21 Oct
2002
21 Oct
'02
8:42 a.m.
pleased to announce the release of 'unmatched'.
'unmatched' is a simple effort to recreate some aspects of
the tone shaping of a real instrument amplifier. unlike
classical convolution techniques, it uses an IIR filter
to emulate an original impulse response, trading impulse
response fidelity for execution speed. it will easily
allow realtime setups combining compression, reverb and
other cpu-bound effects even on an aged system.
http://quitte.de/unmatched.html has the rationale and
implementation (GPL).
tim
21 Oct
21 Oct
10:28 a.m.
On Mon, Oct 21, 2002 at 02:39:05 +0200, Tim Goetze wrote:
pleased to announce the release of
'unmatched'.
'unmatched' is a simple effort to recreate some aspects of
the tone shaping of a real instrument amplifier. unlike
Excellent. Can I ask how you did the FIR -> IIR mapping? Is it neccesary
or significantly better to use doubles for the coefficents?
Have you experimented with adding a delay line and LP filter to simulate
reflections off the back wall of the cabinet?
- Steve
1:15 p.m.
Steve Harris wrote:
On Mon, Oct 21, 2002 at 02:39:05 +0200, Tim Goetze
wrote:
a quick test (%s/double/float/g) shows the cpu usage doubling,
but i'm unsure what may cause this huge performance drop.
doubles should, on average, mean we seldom hit the denormal number
bounds, or at least less frequently than with floats. i also expect
doubles to be beneficial to the filter stability by minimizing
round-off error, though i may be wrong here. lastly, x87 uses 80 bits
of precision internally, so it hardly means a performance penalty,
unless the cpu D-cache has problems keeping the filter history in
reach.
the converger program is at http://quitte.de/Converger.py --
it won't run without the mandala module, but it hopefully
serves well to explain the algorithm (and probably better
than my distorted translation into human language.)
pleased to announce the release of
'unmatched'.
'unmatched' is a simple effort to recreate some aspects of
the tone shaping of a real instrument amplifier. unlike
Excellent. Can I ask how you did the FIR -> IIR mapping? Is it neccesary
or significantly better to use doubles for the coefficents? Have you experimented with adding a delay line and LP
filter to simulate
reflections off the back wall of the cabinet?
actually the next thing i'd have done do to come closer to the
original impulse would be an added IIR operating behind a delay
line, yes. the current plugin response does not capture the later
parts of the original response so well (now that you mention it,
i think it must be those reflections). however personally i'll
probably do more research on nonlinear effects before refining
this method.
tim
6:43 p.m.
On Mon, Oct 21, 2002 at 07:11:41 +0200, Tim Goetze wrote:
Yes, you can always pep things up with a nonlinear effect of somekind
before the cabinet+speaker sim. In FFT convolvers you take several
impulses at different amplitudes and shift between them.
I wouldn't worry too much aboout being very close the the impulse, the're
only for a particular input amplitude and I can't remember when they came
from so may not be fantastic.
I'm wondering if this technique can be used for reverbs too - generate a
purely "white" synthetic reverb tail, and apply an IIR the aproximate
shape of the rooms impulse to it to make it sound more real...
- Steve
Steve Harris wrote:
Lets just put it down to chache issues and ignore it ;)
On Mon, Oct 21, 2002 at 02:39:05 +0200, Tim Goetze
wrote:
a quick test (%s/double/float/g) shows the cpu usage doubling,
but i'm unsure what may cause this huge performance drop. pleased to announce the release of
'unmatched'.
'unmatched' is a simple effort to recreate some aspects of
the tone shaping of a real instrument amplifier. unlike
Excellent. Can I ask how you did the FIR -> IIR mapping? Is it neccesary
or significantly better to use doubles for the coefficents? doubles should, on average, mean we seldom hit the
denormal number
bounds, or at least less frequently than with floats. i also expect
doubles to be beneficial to the filter stability by minimizing
round-off error, though i may be wrong here. lastly, x87 uses 80 bits
Yes, though can manually truncate the floats, and floats will give half
the cache footprint, so play better with other software. I dont really
have a feel for when doubles are neccesary for filters yet.
Have you
experimented with adding a delay line and LP filter to simulate
reflections off the back wall of the cabinet?
actually the next thing i'd have done do to come closer to the
original impulse would be an added IIR operating behind a delay
line, yes. the current plugin response does not capture the later
parts of the original response so well (now that you mention it,
i think it must be those reflections). however personally i'll
probably do more research on nonlinear effects before refining
this method.
22 Oct
22 Oct
7:07 a.m.
Steve Harris wrote:
can't ... ;) here's the output of gcc -O2 -S. this is what is
added for every coefficient in the float version wrt doubles:
...
fstps -4(%ebp)
flds -4(%ebp)
...
afaict, it truncates the algorithm's accumulator to float.
if we change the constant coefficients to come as floats
(appended 'f') this truncation is omitted, and the cpu cost
is the same (maybe a tad more) as for doubles. i conclude
that constant real values really need the 'f' if operating
with floats in gcc, and that there's little reason to use
them in DSP on x87 anyway if memory effects aren't essential.
on average, the branched truncate costs more in this filter
than simply ignoring denormals. for this particular filter,
there doesn't seem to be a good reason to switch to floats
at least on the system i use. if i was pressed for a rule,
i'd say use doubles unless you need to store lots of them.
a quick test
(%s/double/float/g) shows the cpu usage doubling,
but i'm unsure what may cause this huge performance drop.
Lets just put it down to chache issues and ignore it ;) doubles
should, on average, mean we seldom hit the denormal number
bounds, or at least less frequently than with floats. i also expect
doubles to be beneficial to the filter stability by minimizing
round-off error, though i may be wrong here. lastly, x87 uses 80 bits
Yes, though can manually truncate the floats, and floats will give half
the cache footprint, so play better with other software. I dont really
have a feel for when doubles are neccesary for filters yet. Yes, you can always pep things up with a nonlinear
effect of somekind
before the cabinet+speaker sim. In FFT convolvers you take several
impulses at different amplitudes and shift between them.
currently, i'm looking at what a sine wave looks like after
it's been handled by a good distortion stomp box. the way it
shapes the wave form seems easy to grasp, but as usual, i
am hesitant to implement what i don't understand fully ...
I wouldn't worry too much aboout being very close
the the impulse, the're
only for a particular input amplitude and I can't remember when they came
from so may not be fantastic.
the recording seems to be of decent quality, and the iir
response irons away most of the noise anyway. but the most
important thing is i like the sound of it, which i do a lot.
i've tried about every of your impulses and, would you
believe it, liked the fenders the least. i regularly play
a fender super 60, for ten years or so. :) got to take
a response from it someday myself.
I'm wondering if this technique can be used for
reverbs too - generate a
purely "white" synthetic reverb tail, and apply an IIR the aproximate
shape of the rooms impulse to it to make it sound more real...
very interesting thought indeed. do you have a good response
for trying this? (sorry, "steve's flat" doesn't qualify,
"albert hall" is more like it ;)
tim
7:44 a.m.
On Tue, Oct 22, 2002 at 03:09:52AM +0200, Tim Goetze wrote:
I have some, but there not included because there 1) very long 2) legally
dubious. I dont think that deriving an EQ curve form them can be dodgy
though - the hard part is more likly to be getting a purly white reverb -
I suspect that all those allpass's and combs dont even out very well.
PS "Steve's flat" was captured with a half knackered monitor and a three
quarters knackered mic. I should probably do it again, as I have moved :)
- Steve
on average, the branched truncate costs more in this
filter
than simply ignoring denormals. for this particular filter,
there doesn't seem to be a good reason to switch to floats
at least on the system i use. if i was pressed for a rule,
i'd say use doubles unless you need to store lots of them.
No, there is no good reason to switch to floats here. I havn't found that
doubles are generally better than floats though - but I also dont have a
way of prediciting it either. It would save be a lot of time if I did.
currently, i'm looking at what a sine wave looks
like after
it's been handled by a good distortion stomp box. the way it
shapes the wave form seems easy to grasp, but as usual, i
am hesitant to implement what i don't understand fully ...
I've found that the precise shape of the transfer function is important to
the sound, do you have a model for the transfer function or are you just
going to smaple it?
the recording seems to be of decent quality, and the
iir
response irons away most of the noise anyway. but the most
important thing is i like the sound of it, which i do a lot.
i've tried about every of your impulses and, would you
believe it, liked the fenders the least. i regularly play
a fender super 60, for ten years or so. :) got to take
a response from it someday myself.
Sadly heres where my zero knowledge of amps kicks in ;) I wouldn't know a
fender form a hole in the ground.
I'm
wondering if this technique can be used for reverbs too - generate a
purely "white" synthetic reverb tail, and apply an IIR the aproximate
shape of the rooms impulse to it to make it sound more real...
very interesting thought indeed. do you have a good response
for trying this? (sorry, "steve's flat" doesn't qualify,
"albert hall" is more like it ;)
9:04 a.m.
Steve Harris wrote:
i'll rather try and come up with a formula. a simple mapping
will not work, because the transfer function moves the peaks
of the sine to happen earlier (i guess looking at the derivative
of the signal may play a role here). i have some ideas, but i'll
need to read up on what's actually happening inside the
circuitry. and if it turns out to include fabs() and branching,
i'll try and combine it with a compression effect (if it isn't
some sort of already, which i suspect).
I have some, but there not included because there 1) very long 2) legally
dubious. I dont think that deriving an EQ curve form them can be dodgy
though - the hard part is more likly to be getting a purly white reverb -
I suspect that all those allpass's and combs dont even out very well.
i guess (don't know much about reverb design [yet]) that in
order to get a truly 'white' reverb the number of delay lines
approaches infinity. which turns a decent implementation into
a real programming challenge. :) nonetheless, one could start
with what we have -- usually people complain that there's not
enough 'color' in our digital reverbs ('gray'?).
currently,
i'm looking at what a sine wave looks like after
it's been handled by a good distortion stomp box. the way it
shapes the wave form seems easy to grasp, but as usual, i
am hesitant to implement what i don't understand fully ...
I've found that the precise shape of the transfer function is important to
the sound, do you have a model for the transfer function or are you just
going to smaple it? Sadly heres where my zero knowledge of amps kicks in ;)
I wouldn't know a
fender form a hole in the ground.
they all sound like a stratocaster, whatever instrument is
plugged in. ;) no seriously, their sound is very bright; at
best you'll call it very 'electric'.
I'm wondering if this technique can be used for
reverbs too - generate a
purely "white" synthetic reverb tail, and apply an IIR the aproximate
shape of the rooms impulse to it to make it sound more real...
very interesting thought indeed. do you have a good response
for trying this? (sorry, "steve's flat" doesn't qualify,
"albert hall" is more like it ;) PS "Steve's flat" was captured with a
half knackered monitor and a three
quarters knackered mic. I should probably do it again, as I have moved :)
and i should really do the fender here and compare the original
to the convolved/iir-filtered emulation. but it's more important
to get the nonlinear parts right first i think.
tim
1:42 p.m.
On Tue, Oct 22, 2002 at 03:00:03PM +0200, Tim Goetze wrote:
As a first cut, have you tried valve_1209 + valve_rec_1405 + [some LP
filter] plugins? These make a fairly conventional tube model when applied
in that order, and the rectifier includes some line sag effects which are
the source of some of the compression effects (modulo my very basic
knowledge of electronics).
- Steve
i guess (don't know much about reverb design
[yet]) that in
order to get a truly 'white' reverb the number of delay lines
approaches infinity. which turns a decent implementation into
a real programming challenge. :) nonetheless, one could start
with what we have -- usually people complain that there's not
enough 'color' in our digital reverbs ('gray'?).
Also, 1 delay will give you white, so maybe its just the middleground that
causes problems. Its possible that a linear waveguide based reverb will
shift the frequencies less.
You're right though, we should try an off the shelf reverb first.
PS
"Steve's flat" was captured with a half knackered monitor and a three
quarters knackered mic. I should probably do it again, as I have moved :)
and i should really do the fender here and compare the original
to the convolved/iir-filtered emulation. but it's more important
to get the nonlinear parts right first i think.
3:42 p.m.
Steve Harris wrote:
As a first cut, have you tried valve_1209 + valve_rec_1405 + [some LP
filter] plugins? These make a fairly conventional tube model when applied
in that order, and the rectifier includes some line sag effects which are
the source of some of the compression effects (modulo my very basic
knowledge of electronics).
what i don't get about the valve (the rect influence seems
negligible) is that it only shapes the negative portion of a
sine. without a LP, there's faint aliasing. it doesn't sound
so bad though, in fact i like it.
however i'm into harsh distortion now, and have fed the fender
three sines from zero to full gain (three octaves of 'c').
the shaping it does has some important properties that i'm only
beginning to understand, but it really does look fundamentally
different. it seems to self-oscillate at the zero-crossings,
and otherwise compresses the sine much like your valve does,
only the output signal is not a straight line where compressed,
but rather another irregular but smooth oscillation about a DC
value. a (realtime) electric circuit simulator would come in
handy here. i have a schematic of a tube amp circuit but i'm
not apt at drawing conclusions from it.
tim
On Tue, Oct 22, 2002 at 03:00:03PM +0200, Tim Goetze
wrote:
the problem with converging IIR is that either the IIR or the
converging algorithm or both don't like impulses that are too
lively. the converger will simply fall dead when it sees an
exponentially decaying white noise impulse. it makes sense,
my understanding of its nature has it that the IIR really likes
to oscillate smoothly. i'd love to see the response of a good
concert hall, that will surely clear some questions on my part.
i guess (don't know much about reverb design
[yet]) that in
order to get a truly 'white' reverb the number of delay lines
approaches infinity. which turns a decent implementation into
a real programming challenge. :) nonetheless, one could start
with what we have -- usually people complain that there's not
enough 'color' in our digital reverbs ('gray'?).
Also, 1 delay will give you white, so maybe its just the middleground that
causes problems. Its possible that a linear waveguide based reverb will
shift the frequencies less.
You're right though, we should try an off the shelf reverb first. PS "Steve's flat" was captured with a
half knackered monitor and a three
quarters knackered mic. I should probably do it again, as I have moved :)
and i should really do the fender here and compare the original
to the convolved/iir-filtered emulation. but it's more important
to get the nonlinear parts right first i think.
7:03 p.m.
On Tue, Oct 22, 2002 at 09:38:26 +0200, Tim Goetze wrote:
the problem with converging IIR is that either the IIR
or the
converging algorithm or both don't like impulses that are too
lively. the converger will simply fall dead when it sees an
exponentially decaying white noise impulse. it makes sense,
I don't understand why. Is it just beacuse the kernle is a bit too long to
be stable, or is there someting more to it than that?
what i don't get about the valve (the rect
influence seems
negligible) is that it only shapes the negative portion of a
sine. without a LP, there's faint aliasing. it doesn't sound
so bad though, in fact i like it.
Thats because its only odd harmonics (or is it even?). It should be
follwed by a DC offset remover. This is pretty normal for a tube
simulation. The aliasing is because I never got round to oversampling the
valve. It needs some more optimisation to make that a good idea, and LP
filters are so much easier ;)
however i'm into harsh distortion now, and have
fed the fender
three sines from zero to full gain (three octaves of 'c').
the shaping it does has some important properties that i'm only
beginning to understand, but it really does look fundamentally
different. it seems to self-oscillate at the zero-crossings,
and otherwise compresses the sine much like your valve does,
only the output signal is not a straight line where compressed,
but rather another irregular but smooth oscillation about a DC
value. a (realtime) electric circuit simulator would come in
Hmmm... I'm quite out of my depth here, but could this be a class B effect
I'm not correctly modelling? I suspect a class B crossover discontinuity,
effected by the mostion and mass of the speaker cones (pretty high for a
guitar amp I'd guess) could look like that. Any chance of a short sample,
to make sure I'm thinking of the right shape.
Infact, this should just happen, if you use the crossover plugin and
put the output through your speaker cone IIR you should see a similar
shape, as the LP characteristic of the IIR damps the abrupt 0 sticking
point from the crossover.
- Steve
23 Oct
23 Oct
8:41 a.m.
Steve Harris wrote:
yep. :) they should build hw lagrange- or sinc-interpolation
into those fantastic 'multimedia' cpus. would save us a lot
of work.
still i'm not convinced that a sine should only be compressed
if < 0 by valve_1209.so.
i've copied (950 + 50 silent) samples of each into
http://quitte.de/fender-sine.wave -- the quitte.de quota is
measured in k, not M :(. the two oscillations mentioned above
seem to be one actually, their period of about 12 samples is
independent from the sine frequency that is fed to the amp, i
double-checked with a 'f#' sine (that got lost accidentally,
it does not look substantially different however). actually
i should record the output from the circuit not the speaker.
On Tue, Oct 22, 2002 at 09:38:26 +0200, Tim Goetze
wrote:
yes, the filter quickly becomes unstable during convergence.
i think it's simply against the nature of the iir, which i take
to prefer smooth oscillation. even if there existed some
'magical' set of coefficients that could emulate the noise
impulse, finding it (i can only argument on the basis of my
feeling for the algorithms involved) will require a different
method than the one employed (foreach coefficient a small
step, then combine what looked reasonable).
the problem with converging IIR is that either
the IIR or the
converging algorithm or both don't like impulses that are too
lively. the converger will simply fall dead when it sees an
exponentially decaying white noise impulse. it makes sense,
I don't understand why. Is it just beacuse the kernle is a bit too long to
be stable, or is there someting more to it than that? what i
don't get about the valve (the rect influence seems
negligible) is that it only shapes the negative portion of a
sine. without a LP, there's faint aliasing. it doesn't sound
so bad though, in fact i like it.
Thats because its only odd harmonics (or is it even?). It should be
follwed by a DC offset remover. This is pretty normal for a tube
simulation. The aliasing is because I never got round to oversampling the
valve. It needs some more optimisation to make that a good idea, and LP
filters are so much easier ;) however
i'm into harsh distortion now, and have fed the fender
three sines from zero to full gain (three octaves of 'c').
the shaping it does has some important properties that i'm only
beginning to understand, but it really does look fundamentally
different. it seems to self-oscillate at the zero-crossings,
and otherwise compresses the sine much like your valve does,
only the output signal is not a straight line where compressed,
but rather another irregular but smooth oscillation about a DC
value. a (realtime) electric circuit simulator would come in
Hmmm... I'm quite out of my depth here, but could this be a class B effect
I'm not correctly modelling? I suspect a class B crossover discontinuity,
effected by the mostion and mass of the speaker cones (pretty high for a
guitar amp I'd guess) could look like that. Any chance of a short sample,
to make sure I'm thinking of the right shape. Infact, this should just happen, if you use the
crossover plugin and
put the output through your speaker cone IIR you should see a similar
shape, as the LP characteristic of the IIR damps the abrupt 0 sticking
point from the crossover.
yup, it definitely looks more like it on the scope. but the
aliasing the crossover introduces is simply unbearable. i think
i'll go for a ride and show some waves to the guy who fixed my
fender amp recently, he knows a lot about the circuitry.
tim
ps: i've updated the 'unmatched.html' page to show graphs of both
the original and the plugin response.
1:39 p.m.
On Wed, Oct 23, 2002 at 02:37:17PM +0200, Tim Goetze wrote:
still i'm not convinced that a sine should only be
compressed
if < 0 by valve_1209.so.
I'm not either. Tubes don't do that.
Tube distortion is famous for emphasizing even harmonics, but
it certainly isn't *all* even harmonics.
A bit of googling turns up some oscilloscope shots of
tube clipping such as these:
http://www.ax84.com/pdr/18watterbuzz.html#ResistiveLoad
Notice that while the clipping is certainly asymmetrical,
it's certainly NOT one-sided.
This is also worth considering, by a guy who's put a lot
of time into building analog (solid-state) tube amp
simulators:
http://home3.netcarrier.com/~lxh2/distampt.html
--PW
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
7:06 p.m.
Paul Winkler wrote:
This is also worth considering, by a guy who's put
a lot
of time into building analog (solid-state) tube amp
simulators:
http://home3.netcarrier.com/~lxh2/distampt.html
also the other shots the same guy has (pod distortion
actually sounds quite good imo):
http://home3.netcarrier.com/~lxh2/pod.html
where a sine morphs into something almost perfectly square
as the gain rises (much like your first link's scope
images from the real tube thing).
i'm not sure how to correctly model this clipping in sw.
steve's valve.xml is, afaict, not guaranteed to be band-
limited if we extend it to support this kind of shaping.
nice googling.
tim
7:48 p.m.
On Thu, Oct 24, 2002 at 01:01:45 +0200, Tim Goetze wrote:
http://home3.netcarrier.com/~lxh2/pod.html
where a sine morphs into something almost perfectly square
as the gain rises (much like your first link's scope
images from the real tube thing).
i'm not sure how to correctly model this clipping in sw.
steve's valve.xml is, afaict, not guaranteed to be band-
limited if we extend it to support this kind of shaping.
No, it will not be bandlimited. I'l have to think about how to do this.
Maybe just by using an oversampled Chebyshev(sp?) and increasing harmonic
level with the amplitude.
I've been playing with the crossover dist, I've implemented some better
crossover behaviour and fixed a serious (and stupid) bug, which is
probably the source of some of what sounded like aliasing.
It only does something meaningful if you run it with smoothing=1.0
The fixed version is in
http://plugin.org.uk/snapshots/swh-plugins-2002-10-24.tar.gz
It also contains a decent mono compressor for anyone else reading this.
- Steve
8:06 p.m.
Steve Harris wrote:
I've been playing with the crossover dist, I've
implemented some better
crossover behaviour and fixed a serious (and stupid) bug, which is
probably the source of some of what sounded like aliasing.
:)
It only does something meaningful if you run it with
smoothing=1.0
The fixed version is in
http://plugin.org.uk/snapshots/swh-plugins-2002-10-24.tar.gz
It also contains a decent mono compressor for anyone else reading this.
that's good to hear.
603k! ah well i needed a fresh set anyway,
you changed the plate since .2.8, right?
tim
24 Oct
24 Oct
3:48 a.m.
On Thu, Oct 24, 2002 at 02:02:51 +0200, Tim Goetze wrote:
603k! ah well i needed a fresh set anyway,
you changed the plate since .2.8, right?
Yes, though it could still do with some work. I'm not sure what exactly,
I've never played with a real plate. Its a physical model, and they can
be tricky to tune, but a few well placed filters might do the trick.
- Steve
7:32 a.m.
Steve Harris wrote:
On Thu, Oct 24, 2002 at 02:02:51 +0200, Tim Goetze
wrote:
i like the .2.8 version as it stands (though iirc you said
it didn't do what you intended it to).
tim
603k! ah well i needed a fresh set anyway,
you changed the plate since .2.8, right?
Yes, though it could still do with some work. I'm not sure what exactly,
I've never played with a real plate. Its a physical model, and they can
be tricky to tune, but a few well placed filters might do the trick.
8:31 a.m.
me wrote:
Steve Harris wrote:
yesterday's release fails to compile because util/waveguide_nl.h
is missing. the version from the 0.2.8 tarball seems to work.
tim
On Thu, Oct 24, 2002 at 02:02:51 +0200, Tim Goetze
wrote:
i like the .2.8 version as it stands (though iirc you said
it didn't do what you intended it to). 603k! ah well i needed a fresh set anyway,
you changed the plate since .2.8, right?
Yes, though it could still do with some work. I'm not sure what exactly,
I've never played with a real plate. Its a physical model, and they can
be tricky to tune, but a few well placed filters might do the trick.
8:58 a.m.
On Thu, Oct 24, 2002 at 02:27:04 +0200, Tim Goetze wrote:
>i like the .2.8 version as it stands (though iirc
you said
>it didn't do what you intended it to).
I dont think it sounds quite right, but maybe im expecting it to be more
like a synthetic plate reverb, than an actual metal plate.
yesterday's release fails to compile because
util/waveguide_nl.h
is missing. the version from the 0.2.8 tarball seems to work.
Thats fine, its the same, but I've added it to the makefile, cheers.
- Steve
23 Oct
23 Oct
8:27 p.m.
On Thu, Oct 24, 2002 at 12:45:27AM +0100, Steve Harris wrote:
No, it will not be bandlimited. I'l have to think about how to do this.
Maybe just by using an oversampled Chebyshev(sp?) and increasing harmonic
level with the amplitude.
it'll probably help that this will typically be run in front of
some kind of lowpass filter. unfiltered overdrive is harsssshhhhhh.
--PW
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
24 Oct
24 Oct
3:44 a.m.
On Wed, Oct 23, 2002 at 05:24:48 -0700, Paul Winkler wrote:
On Thu, Oct 24, 2002 at 12:45:27AM +0100, Steve Harris
wrote:
You say that like its a bad thing ;) But yes.
Looking at the scope traces from that guys site (which I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages. Its not that suprising, it seems to be one of the
things that gives 303's there distinctive bite, and might play well with
further waveshaping stages.
- Steve
No, it will not be bandlimited. I'l have to think about how to do this.
Maybe just by using an oversampled Chebyshev(sp?) and increasing harmonic
level with the amplitude.
it'll probably help that this will typically be run in front of
some kind of lowpass filter. unfiltered overdrive is harsssshhhhhh. 
5:06 a.m.
On Thu, 2002-10-24 at 08:41, Steve Harris wrote:
Its not that suprising, it seems to be one of the
things that gives 303's there distinctive bite, and might play well with
further waveshaping stages.
Talking of 303s, how come Freebirth had to be pulled? I heard
propellerheads threatened them as they had "copied their layout". Which
begs the question, why isn't Roland suing P/heads for copying the layout
of the 303/808/909????
-Lea.
5:32 a.m.
On Thu, Oct 24, 2002 at 10:02:02 +0100, Lea Anthony wrote:
Talking of 303s, how come Freebirth had to be pulled?
I heard
propellerheads threatened them as they had "copied their layout". Which
begs the question, why isn't Roland suing P/heads for copying the layout
of the 303/808/909????
I don't think it was "copying the layout" I think it was copying the
bitmaps that did it.
- Steve
6:03 a.m.
Steve,
First up, my mistake, it was Reborn.
This is not my understanding. The P/H bitmaps were NOT part of the FB
release. They basically stated that if you had an original copy of
Rebirth, then you could copy the bitmap files from your CD to your
freebirth directory. No, it was definately a case of "You copied our
layout".
<quick google search>
Aha! From the site:
"Unfortunately I have been asked by Propellerhead Software to remove
ReBorn from my website and refrain from distributing it. Although ReBorn
was an original work, it copied the look, feel and operation of
Propellerhead's ReBirth program and as such was in breach of copyright."
Apparently they had written permission from Roland for their layout.
That SUCKS big time. They want to use someone elses layout but then get
pissy when someone uses theirs.
Time to write an email.... >:(
-Lea.
On Thu, 2002-10-24 at 10:29, Steve Harris wrote:
I don't think it was "copying the
layout" I think it was copying the
bitmaps that did it.
6:30 a.m.
On Thu, Oct 24, 2002 at 10:59:38 +0100, Lea Anthony wrote:
"Unfortunately I have been asked by Propellerhead
Software to remove
ReBorn from my website and refrain from distributing it. Although ReBorn
was an original work, it copied the look, feel and operation of
Propellerhead's ReBirth program and as such was in breach of copyright."
Oh yeah, I remeber this discussion now. I think that either there was more
going on than meets the eye or the author just didn't want any hastle.
Personally I think I would have stood up to them, but its up to the
individual really.
- Steve
7:03 a.m.
On 24 Oct 2002, Lea Anthony wrote:
That SUCKS big time. They want to use someone elses
layout but then get
pissy when someone uses theirs.
Time to write an email.... >:(
Umm. You really should read the archives on this topic first. The
discussion got rather out of hand and I'm pretty sure all the bases were
covered.
Taybin
--
http://www.piratesvsninjas.com
11:26 a.m.
On Thu, Oct 24, 2002 at 08:41:36AM +0100, Steve Harris wrote:
Looking at the scope traces from that guys site (which
I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages.
Output transformer? Certainly they don't have infinite response...
It's not clear to me where and how those scope shots were taken.
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
12:34 p.m.
On Thu, Oct 24, 2002 at 08:23:32 -0700, Paul Winkler wrote:
On Thu, Oct 24, 2002 at 08:41:36AM +0100, Steve Harris
wrote:
Could be. Aparently there is often HP filters insode amps to strop the low
freqs dominating though.
Some of the scope traces are annotated with where they were taken from but
its all greek to me.
- Steve
Looking at the scope traces from that guys site
(which I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages.
Output transformer? Certainly they don't have infinite response...
It's not clear to me where and how those scope shots were taken.
1:43 p.m.
On Thu, Oct 24, 2002 at 05:31:04PM +0100, Steve Harris wrote:
On Thu, Oct 24, 2002 at 08:23:32 -0700, Paul Winkler
wrote:
I'm looking at a schematic for an 18-watt marshall at the moment...
there's a cap between the input gain tube and ground which *might*
be a HPF, but I don't really grok tube circuits well enough to be sure.
On Thu, Oct 24, 2002 at 08:41:36AM +0100, Steve
Harris wrote:
Could be. Aparently there is often HP filters insode amps to strop the low
freqs dominating though. Looking at the scope traces from that guys site
(which I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages.
Output transformer? Certainly they don't have infinite response...
It's not clear to me where and how those scope shots were taken. Some of the scope traces are annotated with where they
were taken from but
its all greek to me.
On the 18watter page, they only say whether a resistive or speaker
load was used. From that, I now suspect the scope was looking at
the voltage across the load.
BTW, the 18watter page has sound samples here:
http://www.ax84.com/pdr/18watter.html
it's a nice little amp. the mp3s have nasty mp3 artifacts,
and it sounds kinda nasal like he miked it oddly or used an odd cabinet,
but the bridge pickup sample sounds kinda like AC-DC.
Now THAT's a classic rock tone.
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
25 Oct
25 Oct
4:19 p.m.
Paul Winkler wrote:
On Thu, Oct 24, 2002 at 05:31:04PM +0100, Steve Harris
wrote:
an introduction to their basic operation is here:
http://www.eecs.umich.edu/~mmccorq/tubes/introduction/introduction.html
starts with 'what is a tube' and ends up having handled
almost everything i find inside my super 60.
tim
On Thu, Oct 24, 2002 at 08:23:32 -0700, Paul
Winkler wrote:
I'm looking at a schematic for an 18-watt marshall at the moment...
there's a cap between the input gain tube and ground which *might*
be a HPF, but I don't really grok tube circuits well enough to be sure. On Thu, Oct 24, 2002 at 08:41:36AM +0100, Steve
Harris wrote:
Could be. Aparently there is often HP filters insode amps to strop the low
freqs dominating though. Looking at the scope traces from that guys site
(which I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages.
Output transformer? Certainly they don't have infinite response...
It's not clear to me where and how those scope shots were taken. 
24 Oct
24 Oct
12:34 p.m.
Hi,
I haven't followed this thread all the way through, but it looks like
you're all having an interesting time on an interesting subject.
I record mostly through a Pod Pro these days. I've set up a number of my
own amp models. Having used the device for about a year, I'll point out that
(for me anyway) the most interesting changes come from changing the speaker
box part of the equation. A Hi-Watt sounds very different through a closed
single 10" speaker vs. a 4x12 box.
If the scope traces are from a microphone you on the right track. If
they're not, well then you're probably modeling only part of the system.
Mark
-----Original Message-----
From: linux-audio-dev-admin(a)music.columbia.edu
[mailto:linux-audio-dev-admin@music.columbia.edu]On Behalf Of Paul
Winkler
Sent: Thursday, October 24, 2002 8:24 AM
To: linux-audio-dev(a)music.columbia.edu
Cc: Paul Winkler
Subject: Re: [linux-audio-dev] [ann] unmatched - a LADSPA amp tone
On Thu, Oct 24, 2002 at 08:41:36AM +0100, Steve Harris wrote:
Looking at the scope traces from that guys site (which
I will read later
today), there is some fairly serious HP filtering going on sometime after
one of the amp stages.
Output transformer? Certainly they don't have infinite response...
It's not clear to me where and how those scope shots were taken.
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
23 Oct
23 Oct
3:09 p.m.
On Wed, Oct 23, 2002 at 02:37:17 +0200, Tim Goetze wrote:
still i'm not convinced that a sine should only be
compressed
if < 0 by valve_1209.so.
Well, the valve plugin is an aproximation of a single valve, a guitar amp
has several if I understand correctly.
To make distortion I generally use a softish valve followed by some stuff,
followed by an inverter (to make the DC offset smaller), followed by some
stuff, followed by a harder valve. That will give something roughly like
the scope shot Paul posted after this.
Stacking the valve plugins is the key to making them sound interesting.
As I have said before, if someone comes up with a chain of effects that
sounds even vaguely like a guitar amp I will combine it into one plugin.
That should be easier to use and faster.
I dont think the sign of the signal will effect the sound of the output,
just the appearance. Remember the DC offset and the phase shift. The
valve plugin has none, I would expect a real valve to have a fair bit. It
shouldn't noticably effect the sound though.
i've copied (950 + 50 silent) samples of each
into
http://quitte.de/fender-sine.wave -- the quitte.de quota is
Cheers. Just looked at it. It was what I was expecting from your
description.
>Infact, this should just happen, if you use the
crossover plugin and
>put the output through your speaker cone IIR you should see a similar
yup, it definitely looks more like it on the scope.
but the
aliasing the crossover introduces is simply unbearable. i think
i'll go for a ride and show some waves to the guy who fixed my
fender amp recently, he knows a lot about the circuitry.
The crossover plugin was a shot in the dark, based on a vague description
of the effect by someone who understood the electronics. Now I have seen
your recording, and have applied some brain power to it it's obvious what
I did wrong, and how to fix it.
- Steve
7:17 p.m.
Have you guys heard amplitube? Man, those sounds rock!
-Lea.
On Wed, 2002-10-23 at 20:06, Steve Harris wrote:
On Wed, Oct 23, 2002 at 02:37:17 +0200, Tim Goetze
wrote:
still i'm not convinced that a sine should
only be compressed
if < 0 by valve_1209.so.
Well, the valve plugin is an aproximation of a single valve, a guitar amp
has several if I understand correctly.
To make distortion I generally use a softish valve followed by some stuff,
followed by an inverter (to make the DC offset smaller), followed by some
stuff, followed by a harder valve. That will give something roughly like
the scope shot Paul posted after this.
Stacking the valve plugins is the key to making them sound interesting.
As I have said before, if someone comes up with a chain of effects that
sounds even vaguely like a guitar amp I will combine it into one plugin.
That should be easier to use and faster.
I dont think the sign of the signal will effect the sound of the output,
just the appearance. Remember the DC offset and the phase shift. The
valve plugin has none, I would expect a real valve to have a fair bit. It
shouldn't noticably effect the sound though.
i've copied (950 + 50 silent) samples of each
into
http://quitte.de/fender-sine.wave -- the quitte.de quota is
Cheers. Just looked at it. It was what I was expecting from your
description.
>Infact, this should just happen, if you use
the crossover plugin and
>put the output through your speaker cone IIR you should see a similar
yup, it definitely looks more like it on the
scope. but the
aliasing the crossover introduces is simply unbearable. i think
i'll go for a ride and show some waves to the guy who fixed my
fender amp recently, he knows a lot about the circuitry.
The crossover plugin was a shot in the dark, based on a vague description
of the effect by someone who understood the electronics. Now I have seen
your recording, and have applied some brain power to it it's obvious what
I did wrong, and how to fix it.
- Steve
24 Oct
24 Oct
5:09 a.m.
On Thu, 2002-10-24 at 01:12, Tim Goetze wrote:
Lea Anthony wrote:
It's a windows app so I don't know how much extra overhead windows adds
on per submodule. Getting a pure cpu usage for the submodules will be
hard, but an approximation is not much of a problem.
Also, I must ask this, how much of the LAD stuff is designed with 32bit
audio in mind? I ask because that's what I work in, then dither down to
16bit when mastering.
-Lea.
Have you guys heard amplitube? Man, those sounds
rock!
haven't heard it yet, no.
if you happen to use it, it'd be very interesting to hear
how much cpu load the different sub-modules generate.
5:42 a.m.
On Thu, Oct 24, 2002 at 10:05:49 +0100, Lea Anthony wrote:
Also, I must ask this, how much of the LAD stuff is
designed with 32bit
audio in mind? I ask because that's what I work in, then dither down to
16bit when mastering.
My code it is all at least 32bit internally (much is 64bit) and all the
I/O is 32bit. IMHO working with anything else on modern cpus is silly.
Dithering down can be done as the last step in a jack process (if you have
a 16bit card) and you have the choice of rectangular (crap, very fast)
triangular (OK, fast) or shaped w/ error feedback (expensive but good).
I strongly recommend some sort of dithering if you're monitoring or
recording offf a 16bit card. I'l commit some changes later today that make
it more efficient.
- Steve
5:56 a.m.
That's great news! For some reason I was thinking that most LAD stuff
works in 44.1/16 which really isn't the best for professional audio. I
only dither down when mastering to a CD.
Hmm.. ok. Does anyone go up to 96Khz sampling?
Just trying to get a feel what people see as acceptable etc.
-Lea.
PS: I never use 96Khz but I'm sure there are plenty who would want to
use LAD technology for DVD authoring in the future.
On Thu, 2002-10-24 at 10:38, Steve Harris wrote:
My code it is all at least 32bit internally (much is
64bit) and all the
I/O is 32bit. IMHO working with anything else on modern cpus is silly.
Dithering down can be done as the last step in a jack process (if you have
a 16bit card) and you have the choice of rectangular (crap, very fast)
triangular (OK, fast) or shaped w/ error feedback (expensive but good).
I strongly recommend some sort of dithering if you're monitoring or
recording offf a 16bit card. I'l commit some changes later today that make
it more efficient.
- Steve
6:32 a.m.
On Thu, Oct 24, 2002 at 10:52:06 +0100, Lea Anthony wrote:
Hmm.. ok. Does anyone go up to 96Khz sampling?
A few people do, yes. It eats through ADAT and TDIF channels very quicly
though.
PS: I never use 96Khz but I'm sure there are
plenty who would want to
use LAD technology for DVD authoring in the future.
Some poeple think its an improvement for source recoridngs too, even if
you're going to output to CDs. I dont have any 96k AD converters, and hifi
analogue isn't really something I care about.
- Steve
7:31 a.m.
Lea Anthony wrote:
On Thu, 2002-10-24 at 01:12, Tim Goetze wrote:
graphics should not be a problem (bitmap blits don't
create relevant cpu load on recent systems). a rough
figure for the system's total cpu load for every effect
running on its own (if that is possible) would be a good
start.
if you don't run "ms access" or other busy-waiting apps
in the background the system part of the cpu load should
be negligible.
Lea Anthony wrote:
It's a windows app so I don't know how much extra overhead windows adds
on per submodule. Getting a pure cpu usage for the submodules will be
hard, but an approximation is not much of a problem. Have you guys heard amplitube? Man, those sounds
rock!
haven't heard it yet, no.
if you happen to use it, it'd be very interesting to hear
how much cpu load the different sub-modules generate. Also, I must ask this, how much of the LAD stuff is
designed with 32bit
audio in mind? I ask because that's what I work in, then dither down to
16bit when mastering.
most of what i do is 64 internally and 32 for inter-module
exchange.
tim
23 Oct
23 Oct
9:02 p.m.
Have you guys heard amplitube? Man, those sounds rock!
seconded. i've heard amplitube and its amazing. if you have it, you
can toss out every cabinet you've ever used except your favorite, just
for nostalgia's sake. if they put amplitude in a small red, round box
(hint, hint) and sold it, it would be the biggest thing since
... well, you guess.
--p
24 Oct
24 Oct
6:35 a.m.
On Wed, Oct 23, 2002 at 09:01:03 -0400, Paul Davis wrote:
Hehe, thats what people said about bass guitars and the 303 ;) From what
I've heard form guitarists, if its significantly better than the current
generation of pod's etc. it could be a real replacement though.
- Steve
Have you guys
heard amplitube? Man, those sounds rock!
seconded. i've heard amplitube and its amazing. if you have it, you
can toss out every cabinet you've ever used except your favorite, just
23 Oct
23 Oct
7:55 p.m.
Steve Harris wrote:
good. do you think the 44100/12 oscillation is a speaker
or output transformer effect?
now i've seen your code i know where a good deal of
what i have called aliasing is from. :)
did you mean to start the sample loop at index 1 in
crossover_dist_1404.xml? 0 sounds much better. i'm
afraid it still aliases a little, but it does look
like it's basically doing the right thing iiutc.
the problem with both the valve and the xover i think
is that the further you drive them, the edgier the
discontinuity gets. it would have to be computed at
sub-sample resolution, and interpolated in a band-
limited fashion for the real 'hard driving' i think.
my reasoning is that the problem we are facing here
is the same as in the generation of band-limited
square (or sawtooth, if you will) oscillation.
tim
Stacking the valve plugins is the key to making them
sound interesting.
ah, now i dig it. maybe you could make a note about the intended
use in the valve doc?
i've
copied (950 + 50 silent) samples of each into
http://quitte.de/fender-sine.wave -- the quitte.de quota is
Cheers. Just looked at it. It was what I was expecting from your
description. >Infact,
this should just happen, if you use the crossover plugin and
>put the output through your speaker cone IIR you should see a similar
yup, it definitely looks more like it on the
scope. but the
aliasing the crossover introduces is simply unbearable. i think
i'll go for a ride and show some waves to the guy who fixed my
fender amp recently, he knows a lot about the circuitry.
The crossover plugin was a shot in the dark, based on a vague description
of the effect by someone who understood the electronics. Now I have seen
your recording, and have applied some brain power to it it's obvious what
I did wrong, and how to fix it.
24 Oct
24 Oct
4:05 a.m.
On Thu, Oct 24, 2002 at 01:50:29 +0200, Tim Goetze wrote:
ah, now i dig it. maybe you could make a note about
the intended
use in the valve doc?
Yeah, I should probably do that. Documentation (well english in general!)
is not my strong point.
good. do you think the 44100/12 oscillation is a
speaker
or output transformer effect?
I'm 90% sure its from the crossover between the two halves of the class B
amp (thats the power amp IIRC).
now i've seen your code i know where a good deal
of
what i have called aliasing is from. :)
Er, yeah. <shuffles feet>
did you mean to start the sample loop at index 1 in
crossover_dist_1404.xml? 0 sounds much better. i'm
I blame fortran. ;)
afraid it still aliases a little, but it does look
like it's basically doing the right thing iiutc.
I didn't think the aliasing would be audible. It uses a linear interp to
go from the bulk of the sub lobe to the true zero corssing, but that should
be OK. You do have smoothing set to 1.0 right? Otherwise it does something
crazy (fix for that later).
the problem with both the valve and the xover i think
is that the further you drive them, the edgier the
discontinuity gets. it would have to be computed at
There definatly shouldn't be a discontinuity. If there is then its a bug.
Also, the crossover effect is not amplitude dependent, maybe it should be,
but the way it was described to me it isn't. Another thing to mention here
is that many people claim the crossover effect isn't a desirable feature of
amps.
I think the valve needs to be applied gently (looking at the scope traces
from yesterday its a fairly mild effect), and then some harder clipping
needs to kick in.
my reasoning is that the problem we are facing here
is the same as in the generation of band-limited
square (or sawtooth, if you will) oscillation.
Yes, but if the aliasing isn't too bad though it can probably just be
solved with a bit of oversampling, 4x or 8x should be enough, though the
CPU use wil go up 4x or 8x. The basic valve should be optimised.
I can reduce the cost if I move the curve from e^-x to 2^-x with a fudge
factor, but does anyone know how this will effect the sound? Its not too
much work, maybe I should just try it.
- Steve
7:31 a.m.
Steve Harris wrote:
drat. i wish i was better at understanding electronics.
the amp guy pointed in this direction too.
the aliasing is not that strong, you're right. maybe i'm
over-sensitive.
what i gather from the scope shots and the analog distortion
guy's pages is that we want a range of clipping from none to
square. if we're going implement 'square' clipping -- which
we'll need for the real rough tone -- we'll have to come up
with a bandlimited solution. the current valve clipping is
real soft compared to those scope shots, isn't it? if we
stack the valve to get to hard clipping, the net result will
carry the sum of all aliasing i'm afraid.
my feeling is that oversampling alone will not do, although
you're probably right in that it will get rid of most of
the aliasing.
i've been working a little on a square oscillator recently
that integrates 32-sample windowed sincs centered at the
wave discontinuities. the results are very promising, although
not perfect, because the straight parts of the wave are
simply that, and integrating the sinc actually is a crude
approximation. cpu usage is fairly low though.
i'm thinking about how to apply the technique to produce
square clipping, or better a [0 .. 1] range of clipping,
but the integration approach is awkward.
tim
On Thu, Oct 24, 2002 at 01:50:29 +0200, Tim Goetze
wrote:
yeah, i guess we'd all be writing poems instead of code if
we did better in this field. ;)
ah, now i dig it. maybe you could make a note
about the intended
use in the valve doc?
Yeah, I should probably do that. Documentation (well english in general!)
is not my strong point. good. do you
think the 44100/12 oscillation is a speaker
or output transformer effect?
I'm 90% sure its from the crossover between the two halves of the class B
amp (thats the power amp IIRC). afraid it
still aliases a little, but it does look
like it's basically doing the right thing iiutc.
I didn't think the aliasing would be audible. It uses a linear interp to
go from the bulk of the sub lobe to the true zero corssing, but that should
be OK. You do have smoothing set to 1.0 right? Otherwise it does something
crazy (fix for that later). the problem
with both the valve and the xover i think
is that the further you drive them, the edgier the
discontinuity gets. it would have to be computed at
There definatly shouldn't be a discontinuity. If there is then its a bug.
Also, the crossover effect is not amplitude dependent, maybe it should be,
but the way it was described to me it isn't. Another thing to mention here
is that many people claim the crossover effect isn't a desirable feature of
amps.
I think the valve needs to be applied gently (looking at the scope traces
from yesterday its a fairly mild effect), and then some harder clipping
needs to kick in. my reasoning
is that the problem we are facing here
is the same as in the generation of band-limited
square (or sawtooth, if you will) oscillation.
Yes, but if the aliasing isn't too bad though it can probably just be
solved with a bit of oversampling, 4x or 8x should be enough, though the
CPU use wil go up 4x or 8x. The basic valve should be optimised.
9:05 a.m.
On Thu, Oct 24, 2002 at 01:19:20 +0200, Tim Goetze wrote:
Did he give the impression that he though it was an important part of the
sound? I'm not clear on this.
I'm 90%
sure its from the crossover between the two halves of the class B
amp (thats the power amp IIRC).
drat. i wish i was better at understanding electronics.
the amp guy pointed in this direction too. the aliasing is not that strong, you're right.
maybe i'm
over-sensitive.
No, no. Aliasing really ruins analogue effects IMHO. On Friday and/or over
the weekend I'm dig out my bass and try it.
what i gather from the scope shots and the analog
distortion
guy's pages is that we want a range of clipping from none to
square. if we're going implement 'square' clipping -- which
Yes, that sounds about right.
we'll need for the real rough tone -- we'll
have to come up
with a bandlimited solution. the current valve clipping is
real soft compared to those scope shots, isn't it? if we
stack the valve to get to hard clipping, the net result will
carry the sum of all aliasing i'm afraid.
Yes, it will compund it. I can't think how to apply oscillator
antialiasing to wavewhaping though. Maybe it can be done.
my feeling is that oversampling alone will not do,
although
you're probably right in that it will get rid of most of
the aliasing.
Its how most people tackle to problem. I think I can get the valve cheap
enough that it will be practical.
i'm thinking about how to apply the technique to
produce
square clipping, or better a [0 .. 1] range of clipping,
but the integration approach is awkward.
I doubt that the electronics ever produces hard square clips.
- Steve
9:53 a.m.
Steve Harris wrote:
make sure you try the iir 'convolver' with the bass. i
like the combination a lot.
for square clipping, i'm thinking about interpolating where
the input will hit the limit, giving us a sub-sample point
in time, on which one'd center the sinc. the quality of the
clipping point interpolation would be crucial. if this
works for square clipping, moving the clipping level and
blending with the straight signal might work to produce
a smooth range of clipping. still needs more thinking though.
that'd be great.
On Thu, Oct 24, 2002 at 01:19:20 +0200, Tim Goetze
wrote:
Did he give the impression that he though it was an important part of the
sound? I'm not clear on this.
unfortunately he didn't have the time to give more than
a few hints, which mostly covered electronic, not sonic
aspects. what i got out of it basically explains the
scope shots paul has thrown in.
I'm
90% sure its from the crossover between the two halves of the class B
amp (thats the power amp IIRC).
drat. i wish i was better at understanding electronics.
the amp guy pointed in this direction too. the aliasing
is not that strong, you're right. maybe i'm
over-sensitive.
No, no. Aliasing really ruins analogue effects IMHO. On Friday and/or over
the weekend I'm dig out my bass and try it. we'll need
for the real rough tone -- we'll have to come up
with a bandlimited solution. the current valve clipping is
real soft compared to those scope shots, isn't it? if we
stack the valve to get to hard clipping, the net result will
carry the sum of all aliasing i'm afraid.
Yes, it will compund it. I can't think how to apply oscillator
antialiasing to wavewhaping though. Maybe it can be done. my feeling is
that oversampling alone will not do, although
you're probably right in that it will get rid of most of
the aliasing.
Its how most people tackle to problem. I think I can get the valve cheap
enough that it will be practical. I doubt that the electronics ever produces hard square
clips.
don't the scope traces suggest they do? i'll try and record
the circuit output from the fender, it has line-out and fx
send.
tim
10:33 a.m.
On Thu, Oct 24, 2002 at 03:49:12 +0200, Tim Goetze wrote:
They dont look very square to me, there is also some kind of dip (looking
at the pod traces) in the top of the clip waveform, which is even more
pronounced in the brit classic and highgain waveforms.
I suspect this dip is some kind of reflection from the limit, but I can't
tell without trying one.
- Steve
I doubt that
the electronics ever produces hard square clips.
don't the scope traces suggest they do? i'll try and record
the circuit output from the fender, it has line-out and fx
send.
11:37 a.m.
Steve Harris wrote:
On Thu, Oct 24, 2002 at 03:49:12 +0200, Tim Goetze
wrote:
They dont look very square to me, there is also some kind of dip (looking
at the pod traces) in the top of the clip waveform, which is even more
pronounced in the brit classic and highgain waveforms.
by 'square' i meant an abrupt change of the slope, not
necessarily into static dx = 0. sorry for being unprecise.
I doubt
that the electronics ever produces hard square clips.
don't the scope traces suggest they do? i'll try and record
the circuit output from the fender, it has line-out and fx
send. I suspect this dip is some kind of reflection from the
limit, but I can't
tell without trying one.
you mean you intend to change valve to have adjustable
slope where it compresses? sounds good, if that is what
you mean.
i did some testing with a simple
in -> valve -> invert -> valve -> convolver -> out
setup and it's beginning to sound like fairly good
distortion when the input signal is strong enough.
aliasing is very, very faint, with both valves set
for maximum saturation (and "character" = 1.). could
use some more saturation/clipping though.
two things: 'in' should probably be HP-filtered, because
the low frequencies become too dominant. and the valve
clipping is too strong at low input levels. it seems to
adapt the clipping level to the input level. i think that
it should rather have a certain threshold beyond which
saturation becomes more noticeable. 'weak' input sines
are flattened almost completely, at very low amplitudes,
rendering a sound resembling that of a torn speaker.
tim
12:39 p.m.
On Thu, Oct 24, 2002 at 05:32:22 +0200, Tim Goetze wrote:
Nope, that would be hard ;) I was thinking of having a second, hard
clipping alg. and bringing that in for high ampltudes.
I suspect this
dip is some kind of reflection from the limit, but I can't
tell without trying one.
you mean you intend to change valve to have adjustable
slope where it compresses? sounds good, if that is what
you mean. i did some testing with a simple
in -> valve -> invert -> valve -> convolver -> out
setup and it's beginning to sound like fairly good
distortion when the input signal is strong enough.
aliasing is very, very faint, with both valves set
for maximum saturation (and "character" = 1.). could
use some more saturation/clipping though.
Yes. Its not really meant to be distortion at that level, more of a
saturation thing. I usually put something hard in the middle (near the
inverter, a cliper or sinus wavewrapper, something like that. Probably
wont make a good guitar sound though.
two things: 'in' should probably be
HP-filtered, because
the low frequencies become too dominant. and the valve
Yes. Aparently they do this in the hardware too.
clipping is too strong at low input levels. it seems
to
adapt the clipping level to the input level. i think that
It doesn't. there are no time domain effects, and it doesn't adapt to
amplitude - just a static trasfer function. It could be an interrelation
with the two valves, but I dont think so...
it should rather have a certain threshold beyond
which
saturation becomes more noticeable. 'weak' input sines
are flattened almost completely, at very low amplitudes,
rendering a sound resembling that of a torn speaker.
Try turning down some of the parameters. You wont get good, hard
distortion out of the valves, more of a natural warmth.
- Steve
25 Oct
25 Oct
4:20 p.m.
Steve Harris wrote:
i put in a music-dsp shaper (archive credits patrice tarrabia
and bram) after the inverter, and with some eq (hp basically)
before entering the first valve, it sounds surprisingly good.
especially with the bridge pickup, the neck pickup is still a
bit muddy.
after some more tinkering, i guess you're right. for some
reason, guitar notes die away very quickly with this setup.
On Thu, Oct 24, 2002 at 05:32:22 +0200, Tim Goetze
wrote:
Nope, that would be hard ;) I was thinking of having a second, hard
clipping alg. and bringing that in for high ampltudes.
oh yes, please keep on bringing them on.
I suspect
this dip is some kind of reflection from the limit, but I can't
tell without trying one.
you mean you intend to change valve to have adjustable
slope where it compresses? sounds good, if that is what
you mean. in ->
valve -> invert -> valve -> convolver -> out
setup and it's beginning to sound like fairly good
distortion when the input signal is strong enough.
aliasing is very, very faint, with both valves set
for maximum saturation (and "character" = 1.). could
use some more saturation/clipping though.
Yes. Its not really meant to be distortion at that level, more of a
saturation thing. I usually put something hard in the middle (near the
inverter, a cliper or sinus wavewrapper, something like that. Probably
wont make a good guitar sound though. clipping is
too strong at low input levels. it seems to
adapt the clipping level to the input level. i think that
It doesn't. there are no time domain effects, and it doesn't adapt to
amplitude - just a static trasfer function. It could be an interrelation
with the two valves, but I dont think so... Try turning down some of the parameters. You wont get
good, hard
distortion out of the valves, more of a natural warmth.
yep, you've managed that quite well. in fact the whole thing
is slowly getting close to what i expect from a decent amp.
there are three issues for me with the current setup:
* it doesn't noticeably prolong sustain.
* the attack phase is 'flat', compared to the ringing
of the real thing.
* the sound gets muddy and faint when you turn down the
volume at the instrument, instead of keeping loudness and
reducing distortion.
i suppose that these could all be improved by applying the
right sort of compression. i've done fairly quick tests with
applying gain before the first valve, and with the new
compressor from your set. the gain fails miserably, and i
can't seem to find the right setting and position for the
compressor.
tim
5:14 p.m.
On Fri, Oct 25, 2002 at 10:15:20 +0200, Tim Goetze wrote:
OK, I have some pending (easy) improvements to the meters, then I'l get
onto this.
That would be the lack of compression
Nope, that
would be hard ;) I was thinking of having a second, hard
clipping alg. and bringing that in for high ampltudes.
oh yes, please keep on bringing them on. i put in a music-dsp shaper (archive credits patrice
tarrabia
and bram) after the inverter, and with some eq (hp basically)
before entering the first valve, it sounds surprisingly good.
especially with the bridge pickup, the neck pickup is still a
bit muddy.
OK, time for a quick lesson - whats the difference between the pickups? My
bass (cheapish active 4 string) has two sets, a single wide one and a pair
slightly offset (they say duncan on them FWIW). There is a knob that, I
think crossfades between them.
It doesn't.
there are no time domain effects, and it doesn't adapt to
amplitude - just a static trasfer function. It could be an interrelation
with the two valves, but I dont think so...
after some more tinkering, i guess you're right. for some
reason, guitar notes die away very quickly with this setup. there are three issues for me with the current setup:
* it doesn't noticeably prolong sustain.
* the attack phase is 'flat', compared to the ringing
of the real thing.
OK... could this be a property of the cliping? Or is it always there?
* the sound gets muddy and faint when you turn down
the
volume at the instrument, instead of keeping loudness and
reducing distortion.
This is another compression effect I suspect.
i suppose that these could all be improved by applying
the
right sort of compression. i've done fairly quick tests with
applying gain before the first valve, and with the new
compressor from your set. the gain fails miserably, and i
I think that the compression effect in a guitar amp is much faster than
what you get from a general purpose compressor. I think it is more like a
slow acting saturation. My Valve rectifier plugin (valve_rect) was an
attempt to capture that, but I expect it needs adjusting, or rewriting.
- Steve
6:44 p.m.
On Fri, Oct 25, 2002 at 10:11:02PM +0100, Steve Harris wrote:
OK, time for a quick lesson - whats the difference
between the pickups? My
bass (cheapish active 4 string) has two sets, a single wide one and a pair
slightly offset (they say duncan on them FWIW). There is a knob that, I
think crossfades between them.
You get more lows and more volume as you go farther from the bridge.
There will be other tonal differences due to the fact that you have
two different kinds of pickups. (the "pair slightly offset"
sounds like a typical Precision aka P-bass pickup).
So any "muddy" effect will likely sound worse with a middle
or neck pickup than with a bridge pickup.
--PW
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
6:52 p.m.
Steve Harris wrote:
the neck, usually placed exactly where the second octave is on
an over-sized fretboard, has far less dominant harmonics than
the bridge. your bass probably is a fender 'precision' type.
usually the split pickup is in the 'neck' position, though it
is much closer to the bridge than on a guitar, because the
sound becomes too dull if you pick up too little harmonics i
guess.
always. i think it might be because a real valve amp sort of
'spreads the excess power' by turning the wave into almost
square at this point, where the guitar signal carries a lot of
energy. more of a limiter effect, right?
On Fri, Oct 25, 2002 at 10:15:20 +0200, Tim Goetze
wrote:
OK, I have some pending (easy) improvements to the meters, then I'l get
onto this.
:)
Nope, that
would be hard ;) I was thinking of having a second, hard
clipping alg. and bringing that in for high ampltudes.
oh yes, please keep on bringing them on. i put in a
music-dsp shaper (archive credits patrice tarrabia
and bram) after the inverter, and with some eq (hp basically)
before entering the first valve, it sounds surprisingly good.
especially with the bridge pickup, the neck pickup is still a
bit muddy.
OK, time for a quick lesson - whats the difference between the pickups? My
bass (cheapish active 4 string) has two sets, a single wide one and a pair
slightly offset (they say duncan on them FWIW). There is a knob that, I
think crossfades between them. * it
doesn't noticeably prolong sustain.
* the attack phase is 'flat', compared to the ringing
of the real thing.
OK... could this be a property of the cliping? Or is it always there? * the sound
gets muddy and faint when you turn down the
volume at the instrument, instead of keeping loudness and
reducing distortion.
This is another compression effect I suspect. I think that the compression effect in a guitar amp is
much faster than
what you get from a general purpose compressor. I think it is more like a
slow acting saturation. My Valve rectifier plugin (valve_rect) was an
my thoughts, too.
attempt to capture that, but I expect it needs
adjusting, or rewriting.
i just tried it before the first valve and it does help a
good deal to have a better attack [in fact i got carried
away strumming a funk pattern, a first time for playing the
guitar through the box], but i'm slowly running out of cpu.
i guess a real HP has to be added, too (the pre-eq is done in
the mixer now) because the lower band still has too much shaping
power over the sound (sort of an auto-wah effect, emphasizing
the 'flat attack' problem).
tim
8:25 p.m.
On Sat, Oct 26, 2002 at 12:47:53 +0200, Tim Goetze wrote:
always. i think it might be because a real valve amp sort of
'spreads the excess power' by turning the wave into almost
square at this point, where the guitar signal carries a lot of
energy. more of a limiter effect, right?
Maybe, or maybe the valve reacts to transients in some way - that could
partly be a side effect of the HP filter, emphasised by some valve effect
we haven't modelled yet.
the neck, usually placed exactly where the second
octave is on
an over-sized fretboard, has far less dominant harmonics than
Cheers, I really should learn how to play the thing sometime.
* the attack phase is 'flat', compared to the
ringing
of the real thing.
OK... could this be a property of the cliping? Or is it always there? i just tried it before the first valve and it does
help a
good deal to have a better attack [in fact i got carried
away strumming a funk pattern, a first time for playing the
guitar through the box], but i'm slowly running out of cpu.
OK, the CPU problems can be improved. Looking at the rectifier I haven't
really optimised it at all. Do you have a feel for whether it should be
faster or slower?
- Steve
26 Oct
26 Oct
11:58 a.m.
Steve Harris wrote:
i think that what we are missing is this transition from hard
to soft clipping as the input becomes weaker soon after the
string has been struck.
can you drop a hint about how you came up with the valve
algorithm?
probably it should be quicker to react. it's not easy to sort
out though, i still haven't got around to get the pre-eq right.
four (or more, if more list members care) ears are certainly
better at sorting this out than my biased two. i've put three
files for testing setups on http://quitte.de:
/power-chord.flac -- a low power chord (root, fifth and eighth)
you know this one from the guys that want to be really bad.
the more distortion the better. should really give a wall
of sound, without any auto-wah at all.
/arpeggio.flac -- a fast-paced arpeggio on an unwound string,
this one should sound familiar from ego-maniac solo work.
should sound clear and sharp, with lots of distortion.
/riff.flac -- excerpt from a jazz melody. i'm particularly
interested in getting this to sound right. unlike the first
two, this was done on the neck pickup, so it's a good
candidate for testing for muddyness. should, but needn't go
with little distortion, but ringing attacks.
there's no eq or other fx on these recordings.
tim
On Sat, Oct 26, 2002 at 12:47:53 +0200, Tim Goetze
wrote:
i've found playing along with a prominent and slow one- to
three-shot delay a fun crash course in both rhythm and
harmony, without having to know a thing about musical theory.
try not to move the left hand too much in the beginning, rather
use all fingers and all strings.
the neck, usually placed exactly where the second
octave is on
an over-sized fretboard, has far less dominant harmonics than
Cheers, I really should learn how to play the thing sometime. always. i
think it might be because a real valve amp sort of
'spreads the excess power' by turning the wave into almost
square at this point, where the guitar signal carries a lot of
energy. more of a limiter effect, right?
Maybe, or maybe the valve reacts to transients in some way - that could
partly be a side effect of the HP filter, emphasised by some valve effect
we haven't modelled yet. i just tried
it before the first valve and it does help a
good deal to have a better attack [in fact i got carried
away strumming a funk pattern, a first time for playing the
guitar through the box], but i'm slowly running out of cpu.
OK, the CPU problems can be improved. Looking at the rectifier I haven't
really optimised it at all. Do you have a feel for whether it should be
faster or slower?
1:21 p.m.
On Sat, Oct 26, 2002 at 05:53:53 +0200, Tim Goetze wrote:
OK, I think we can get that by modifying the rectifer.
Maybe, or maybe
the valve reacts to transients in some way - that could
partly be a side effect of the HP filter, emphasised by some valve effect
we haven't modelled yet.
i think that what we are missing is this transition from hard
to soft clipping as the input becomes weaker soon after the
string has been struck. can you drop a hint about how you came up with the
valve
algorithm?
I didn't, it came from a paper, one of the COST6 ones I think.
- Steve
2:51 p.m.
On Sat, Oct 26, 2002 at 08:08:29 +0200, Tim Goetze wrote:
I think I've found it, Ragnar Bendiksen's thesis:
http://www.notam02.no/~rbendiks/Diplom/Kurveforming.html
but it doesn't look that familiar, and I can't read norwegian, so maybe
there is a translation out there. Definately the same code though.
I just more than doubled its speed BTW (using Andrew Simpers pow(2,x)
aproximation and some hackery). I'm still not convinved that it should be
used for all the stages, I suspect a chebychev would be better for the
output, it would also alow us to model different amps with the same code.
We'd need to calculate the added harmonics from an amp though, maybe by
putting a sin through and FFTing it.
- Steve
I didn't,
it came from a paper, one of the COST6 ones I think.
can't seem to find a trace of cost6 with google. do you
remember where you dug it up?
6:27 p.m.
Steve Harris wrote:
but it doesn't look that familiar, and I can't
read norwegian, so maybe
there is a translation out there. Definately the same code though.
i can decipher some (danish and swedish, though i could not
tell one from the other :), but it doesn't seem to contain
anything you don't know already. thanks.
I just more than doubled its speed BTW (using Andrew
Simpers pow(2,x)
aproximation and some hackery). I'm still not convinved that it should be
that is great to hear! i can almost see the box glowing from
the heat when the valves are turned on. the iir 'convolver',
btw, suffers greatly when competing for the cache with other
plugins now (8-9% @ 64/44.1). think i'll try the floats again.
though the k6 really is cache-starved compared to modern cpus.
used for all the stages, I suspect a chebychev would be
better for the
output, it would also alow us to model different amps with the same code.
We'd need to calculate the added harmonics from an amp though, maybe by
putting a sin through and FFTing it.
i've uploaded yet another recording, line-sine-16.flac, which
contains the line out (left) and the sine that went in. seems
our total transfer function until that point needs more work.
tim
27 Oct
27 Oct
6:25 a.m.
On Sun, Oct 27, 2002 at 12:23:28 +0200, Tim Goetze wrote:
I'm sure it does. I've been playing with myb ass and an ecasound chain:
ecasound -i /dev/dsp -o /dev/dsp -el:valve,1,1 -el:dcRemove -el:amp,4
-el:valveRect,0.5,0.5 -el:sinusWavewrapper,2 -el:valve,1,1.5 -el:dcRemove
-el:amp,-10
It needs some filters and more work, but it sounds pretty good to me.
I'm having a problem than when the input is 0, the output is infinite. I'm
not sure if its a result of the optimisations I did, or it it would do
that anyway.
I haven't listent to your recording yet, but I do that soon.
- Steve
that is great to hear! i can almost see the box
glowing from
the heat when the valves are turned on. the iir 'convolver',
btw, suffers greatly when competing for the cache with other
plugins now (8-9% @ 64/44.1). think i'll try the floats again.
though the k6 really is cache-starved compared to modern cpus.
Well ,the more CPU you have spar, the more tempting it is to un extra
stages and the more cache starved it gets.
used for all
the stages, I suspect a chebychev would be better for the
output, it would also alow us to model different amps with the same code.
We'd need to calculate the added harmonics from an amp though, maybe by
putting a sin through and FFTing it.
i've uploaded yet another recording, line-sine-16.flac, which
contains the line out (left) and the sine that went in. seems
our total transfer function until that point needs more work.
8:45 a.m.
Steve Harris wrote:
On Sun, Oct 27, 2002 at 12:23:28 +0200, Tim Goetze
wrote:
live and learn. i'm looking into a few other places right
now to see where more cycles can be spared. the alsa out
node looks like a good candidate (i'm driving only one of
the ten output channels but zeroing the remaining nine at
every pass). should turn on -O2 instead of -g for the host,
too. :)
that is great to hear! i can almost see the box
glowing from
the heat when the valves are turned on. the iir 'convolver',
btw, suffers greatly when competing for the cache with other
plugins now (8-9% @ 64/44.1). think i'll try the floats again.
though the k6 really is cache-starved compared to modern cpus.
Well ,the more CPU you have spar, the more tempting it is to un extra
stages and the more cache starved it gets. > >used for all the stages, I suspect a chebychev
would be better for the
> >output, it would also alow us to model different amps with the same code.
forgot to ask ... you mean emulate a 'preamp', filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess?
> >We'd need to calculate the added harmonics
from an amp though, maybe by
> >putting a sin through and FFTing it.
>
> i've uploaded yet another recording, line-sine-16.flac, which
> contains the line out (left) and the sine that went in. seems
> our total transfer function until that point needs more work.
btw, the sine was produced by the virus, not the computer.
thus it is 'sample-accurate', the output at one point in time
directly relates to the sine at the same point.
I'm sure it does. I've been playing with myb
ass and an ecasound chain:
ecasound -i /dev/dsp -o /dev/dsp -el:valve,1,1 -el:dcRemove -el:amp,4
-el:valveRect,0.5,0.5 -el:sinusWavewrapper,2 -el:valve,1,1.5 -el:dcRemove
-el:amp,-10
It needs some filters and more work, but it sounds pretty good to me.
just tried it, indeed it does. what i like especially is that
it sounds quite 'juicy' like good distortion is supposed to.
the attack phase is slightly better than with my previous
setup, though there's still some auto-wah. still with the
mixer pre-eq though, it reads '-15 at 80 Hz' so my guess is
the band it attenuates is too low for this purpose. haven't
measured its roll-off yet.
I'm having a problem than when the input is 0, the
output is infinite. I'm
not sure if its a result of the optimisations I did, or it it would do
that anyway.
should be the optimization, i'm getting clean zero out for
zero in with your last public release.
i still remember the green skies a M$VC optimized POV
binary produced, since that day i'm a bit skeptical about
compiler optimizations.
tim
10:21 a.m.
On Sun, Oct 27, 2002 at 02:41:17 +0100, Tim Goetze wrote:
Yes, I'm not sure what the interrealtion between the cabinet and code is
though - I think a simple cabinet delay line + your IIR filter would be a
good place to start.
I think it was a reaction betwen the dc offset filters. Something was very
close to instability.
> >used
for all the stages, I suspect a chebychev would be better for the
> >output, it would also alow us to model different amps with the same code.
forgot to ask ... you mean emulate a 'preamp', filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess? just tried it, indeed it does. what i like especially
is that
it sounds quite 'juicy' like good distortion is supposed to.
the attack phase is slightly better than with my previous
setup, though there's still some auto-wah. still with the
I've fixed the wah noise in the second one I posted, it was due to a
combination of things (lack of inverters, excessive valve settings, lack
of lowcut).
I'm having
a problem than when the input is 0, the output is infinite. I'm
not sure if its a result of the optimisations I did, or it it would do
that anyway.
should be the optimization, i'm getting clean zero out for
zero in with your last public release. i still remember the green skies a M$VC optimized POV
binary produced, since that day i'm a bit skeptical about
compiler optimizations.
This was more of a maths optimisation than a compiler optimisation.
- Steve
1:15 p.m.
Steve Harris wrote:
the 'under-damped' behaviour, as Tom put it, of the power amp
may need to be taken care of. another interrelation of interest
is probably between the output transformer and the speaker, but
this is where i know next to nothing about.
forgot to ask
... you mean emulate a 'preamp', filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess?
Yes, I'm not sure what the interrealtion between the cabinet and code is
though - I think a simple cabinet delay line + your IIR filter would be a
good place to start. I've fixed the wah noise in the second one I
posted, it was due to a
combination of things (lack of inverters, excessive valve settings, lack
of lowcut).
still working on the HP, i only have a generic IIR in my code
which is quite slow, and like to fix/extend as i use the stuff.
tim
2:20 p.m.
On Sun, Oct 27, 2002 at 02:41:17PM +0100, Tim Goetze wrote:
forgot to ask ... you mean emulate a 'preamp',
filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess?
You forgot a couple - then there's the mic, the mic placement,
and the room... there's no end to the madness! :)
--PW
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
3:09 p.m.
Paul Winkler wrote:
On Sun, Oct 27, 2002 at 02:41:17PM +0100, Tim Goetze
wrote:
and then there's PW who's going to code a physical model of
sound transmission through gaseous media, accurate down to
the molecule while giving rather than taking cpu cycles!
his code is going to include a variety of media, including
simulated smoke from various, partly legal sources ...
;) tim
forgot to ask ... you mean emulate a
'preamp', filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess?
You forgot a couple - then there's the mic, the mic placement,
and the room... there's no end to the madness! :)
3:44 p.m.
On Sun, Oct 27, 2002 at 09:04:42PM +0100, Tim Goetze wrote:
Paul Winkler wrote:
I wish :(
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
On Sun, Oct 27, 2002 at 02:41:17PM +0100, Tim
Goetze wrote:
and then there's PW who's going to code a physical model of
sound transmission through gaseous media, accurate down to
the molecule while giving rather than taking cpu cycles! forgot to ask ... you mean emulate a
'preamp', filter and
apply a 'power amp' [and filter again] and a 'cabinet'
stage i guess?
You forgot a couple - then there's the mic, the mic placement,
and the room... there's no end to the madness! :)
6:06 p.m.
On Sun, Oct 27, 2002 at 09:04:42 +0100, Tim Goetze wrote:
Paul Winkler wrote:
>On Sun, Oct 27, 2002 at 02:41:17PM +0100, Tim Goetze wrote:
>> forgot to ask ... you mean emulate a 'preamp', filter and
>> apply a 'power amp' [and filter again] and a 'cabinet'
>> stage i guess?
>
>You forgot a couple - then there's the mic, the mic placement,
>and the room... there's no end to the madness! :)
I think I know how to model the mic placement, model of mic is going a
little far just now ;)
Whats a typical arangement for speaker cones? I remeber seeing single cone
(obviously) and 2x2.
and then there's PW who's going to code a
physical model of
sound transmission through gaseous media, accurate down to
the molecule while giving rather than taking cpu cycles!
his code is going to include a variety of media, including
simulated smoke from various, partly legal sources ...
Dont joke, I remeber seeing a paper about modelling smoke filled air
recently. They didn't say what kind of smoke, but it most have been
something potent if they wanted to model it ;)
The killer plugin would model the acoutstics of a studio when you've just
got in from the pub, its 2am, you've just created a really great sound and
you're too drunk to find a blank DAT ;)
- Steve
7:36 p.m.
On Sun, Oct 27, 2002 at 11:03:23PM +0000, Steve Harris wrote:
Whats a typical arangement for speaker cones? I
remeber seeing single cone
(obviously) and 2x2.
1, 2, and 4 are the most common. It's common to only
mic 1 of the cones, pretty close up - but it's also common
to do any bizarre combination of mics and placements you can
think of.
Then there's the old Ampeg SVT - the bane of many a bass player's
back. The cabinet had 8 10" speakers, weighed nearly 200 lbs,
and could induce loose bowels at 1000 paces. ;)
The main trick with mic placement is that, when you close mic
a speaker, it sounds very different in the middle than it
does near the edge. Have you played with this much?
The killer plugin would model the acoutstics of a
studio when you've just
got in from the pub, its 2am, you've just created a really great sound and
you're too drunk to find a blank DAT ;)
It would have to call an at script to make it sound awful when you
next come in to the studio ;).
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
10:06 p.m.
Paul Winkler wrote:
On Sun, Oct 27, 2002 at 11:03:23PM +0000, Steve Harris
wrote:
Some people also like to stick a mic right between the two cones of a
2-driver amp.
Miking an open cab from behind is also a popular trick - it yields a
completely different
tone.
-dgm
Whats a typical arangement for speaker cones? I
remeber seeing single cone
(obviously) and 2x2.
1, 2, and 4 are the most common. It's common to only
mic 1 of the cones, pretty close up - but it's also common
to do any bizarre combination of mics and placements you can
think of.
11:03 p.m.
On Sun, Oct 27, 2002 at 10:05:24PM -0500, David Gerard Matthews wrote:
Miking an open cab from behind is also a popular trick
- it yields a
completely different
tone.
Some people also put mics in front and behind the cabinet
and flip the "phase" on one of them.
But let's leave poor Steve alone for a while. ;)
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
28 Oct
28 Oct
3:58 a.m.
On Sun, Oct 27, 2002 at 10:05:24 -0500, David Gerard Matthews wrote:
Some people also like to stick a mic right between the
two cones of a
2-driver amp.
Miking an open cab from behind is also a popular trick - it yields a
completely different
tone.
Now you're just being difficult ;)
- Steve
4:06 a.m.
On Sun, Oct 27, 2002 at 04:33:26 -0800, Paul Winkler wrote:
For maximal realism is should probably add a muted sample of someone
shouting "turn that f**king noise off".
- Steve
1, 2, and 4 are the most common. It's common to
only
mic 1 of the cones, pretty close up - but it's also common
to do any bizarre combination of mics and placements you can
think of.
Yup, but for starters, one dynamic mic at a selectable position along the
front I think.
The main trick with mic placement is that, when you
close mic
a speaker, it sounds very different in the middle than it
does near the edge. Have you played with this much?
Not a huge ammount. I owned an amp for a while (now I own a battered box
that smokes and sparks when plugged in and some interesting looking bits).
There is a guy who sells amps in my street, so I'm consider going and
picking up a cheap one.
My provisional plan is to measure the boundary and cone position effects
with my dynamic and a nearfield monitor. The monitor wont give me
eqivalent results, but there are fewer variables.
The killer
plugin would model the acoutstics of a studio when you've just
got in from the pub, its 2am, you've just created a really great sound and
you're too drunk to find a blank DAT ;)
It would have to call an at script to make it sound awful when you
next come in to the studio ;).
27 Oct
27 Oct
8:03 a.m.
On Sun, Oct 27, 2002 at 12:23:28 +0200, Tim Goetze wrote:
i've uploaded yet another recording,
line-sine-16.flac, which
contains the line out (left) and the sine that went in. seems
our total transfer function until that point needs more work.
Did you upload the right file? The left channel is clipped and the right
doesn't look like a pure sinewave, we need to get an unclipped signal from
the output stage for the FT to get the right results.
- Steve
8:54 a.m.
Steve Harris wrote:
On Sun, Oct 27, 2002 at 12:23:28 +0200, Tim Goetze
wrote:
drat. i double-checked and refetched it from the server, the
left channel doesn't clip at all and the right looks like a
pure sine (with some analog noise i admit, though the virus
output is very clean).
a flac bug? i'm using version 1.0.2 fwiw.
tim
i've uploaded yet another recording,
line-sine-16.flac, which
contains the line out (left) and the sine that went in. seems
our total transfer function until that point needs more work.
Did you upload the right file? The left channel is clipped and the right
doesn't look like a pure sinewave, we need to get an unclipped signal from
the output stage for the FT to get the right results.
9:09 a.m.
me wrote:
Steve Harris wrote:
>Did you upload the right file? The left channel is clipped and the right
>doesn't look like a pure sinewave, we need to get an unclipped signal from
>the output stage for the FT to get the right results.
still found some space for line-sine-2.wave -- another
recording from the same sources, uncompressed this time.
tim
10:03 a.m.
On Sun, Oct 27, 2002 at 03:05:36 +0100, Tim Goetze wrote:
still found some space for line-sine-2.wave -- another
recording from the same sources, uncompressed this time.
That looks better. I dont know what was wrong with the flac file, the
others all came out OK.
I have a reasonable bass sound now:
ecasound -i jack_alsa,in_1 -o jack_alsa,out_1 -el:valve,0.6,0.7 -efh:25
-el:inv -el:valveRect,0.5,0.6 -el:inv -el:sinusWavewrapper,2
-el:valve,0.7,0.8 -efh:30 -el:unmatched
It uses about 6% CPU on my athlon, and I'm sure that would be less if it
were combined into one plugin.
I got my girlfriend to try it (she used to play bass a lot), and she said
it sounded pretty good. Its a bit muddy still, but I'm not really sure
where thats coming from, possibly excesive use of the valve code.
The inv plugin is just an inverter, ecasound proabaly has one built in.
- Steve
12:52 p.m.
Steve Harris wrote:
I have a reasonable bass sound now:
ecasound -i jack_alsa,in_1 -o jack_alsa,out_1 -el:valve,0.6,0.7 -efh:25
-el:inv -el:valveRect,0.5,0.6 -el:inv -el:sinusWavewrapper,2
-el:valve,0.7,0.8 -efh:30 -el:unmatched
i take -efh:25 to mean 25 Hz hipass ... looking at the filter
code in there right now. i still need your new valve and rect
implementations to evaluate the right thing.
It uses about 6% CPU on my athlon, and I'm sure
that would be less if it
were combined into one plugin.
is that with a 64 frames block size? that's about 1/3 of what
the k6 should take.
I got my girlfriend to try it (she used to play bass a
lot), and she said
it sounded pretty good. Its a bit muddy still, but I'm not really sure
where thats coming from, possibly excesive use of the valve code.
still some higher-order harmonics missing, i think.
The inv plugin is just an inverter, ecasound proabaly
has one built in.
i use Gain(-1). got to release this darn host sometime, the
processor itself runs stable enough.
tim
5:56 p.m.
On Sun, Oct 27, 2002 at 06:19:24 +0100, Tim Goetze wrote:
Probably, maybe that hard clipper needs writing :(
- Steve
i take -efh:25 to mean 25 Hz hipass ... looking at the
filter
code in there right now. i still need your new valve and rect
implementations to evaluate the right thing.
They dont sound different AFAIK, they are mathematically the same.
is that with a 64 frames block size? that's about
1/3 of what
the k6 should take.
No, 128.
I got my
girlfriend to try it (she used to play bass a lot), and she said
it sounded pretty good. Its a bit muddy still, but I'm not really sure
where thats coming from, possibly excesive use of the valve code.
still some higher-order harmonics missing, i think.
8:45 p.m.
Steve Harris wrote:
it's an ideal case of exp2 approximation then. :)
i'm looking at that line-sine max overdrive wave (please find
it at quitte.de/wave.ps.gz), and i think it just might work to
see its behaviour like this:
the longer the wave is on one side of zero, the more energy is
built up that tries to push it back. as long as the system is
fed enough energy to keep it this side of zero, it is basically
decaying exponentially against a (more-or-less) constant value.
once the energy input becomes too weak to keep the state on
this side of zero, it starts accelerating towards zero and beyond.
the longer it stays there, the more energy is built up pushing
it back, etc.
i am not sure how to do it so it doesn't alias, which i fear a
naive implementation would do.
your thoughts?
tim
code in there
right now. i still need your new valve and rect
implementations to evaluate the right thing.
They dont sound different AFAIK, they are mathematically the same. still some
higher-order harmonics missing, i think.
Probably, maybe that hard clipper needs writing :(
9:57 p.m.
On Mon, Oct 28, 2002 at 02:12:02AM +0100, Tim Goetze wrote:
the longer the wave is on one side of zero, the more
energy is
built up that tries to push it back.
Is it just me, or does that sound an awful lot like a hipass
filter?
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
28 Oct
28 Oct
8:01 a.m.
Paul Winkler wrote:
On Mon, Oct 28, 2002 at 02:12:02AM +0100, Tim Goetze
wrote:
this one is supposed to be nonlinear, and quite so. a hipass
is a linear system unless the implementation is screwed up.
it's an interesting pov though, in fact i've been thinking
about algorithms that work like IIR filters (that are often
used for HP/BP/LP) with the addition of nonlinear terms for
recreating the tube distortion. the maths of such a thing are
not easy to grasp though.
tim
the longer the wave is on one side of zero, the
more energy is
built up that tries to push it back.
Is it just me, or does that sound an awful lot like a hipass
filter?
9:16 a.m.
On Mon, Oct 28, 2002 at 01:30:49 +0100, Tim Goetze wrote:
it's an interesting pov though, in fact i've
been thinking
about algorithms that work like IIR filters (that are often
used for HP/BP/LP) with the addition of nonlinear terms for
recreating the tube distortion. the maths of such a thing are
not easy to grasp though.
In my (limited) experince its really hard to make them stable. But you can
make interesting non-linear filters by using different parameters for the
different halves of the wave cycle. This is what makes the gong sound
gongy (waveguide_nl.h). It could also apply to cone modelling, as the
characteristics of the deflection in the positive and negative directions
will be a bit different.
<rambling>
To make a nonlienar cone filter you could either use different IIR
parameters for signal sign, or you could us different parameters for
different amplitudes (or both).
Different amplitudes would be equivalent to how you model nonliear systems
with FIR impulses - I can't believe I didn't think of it before.
- Steve
11:58 a.m.
Steve Harris wrote:
On Mon, Oct 28, 2002 at 01:30:49 +0100, Tim Goetze
wrote:
yep, that's what i expect, too.
it's an interesting pov though, in fact
i've been thinking
about algorithms that work like IIR filters (that are often
used for HP/BP/LP) with the addition of nonlinear terms for
recreating the tube distortion. the maths of such a thing are
not easy to grasp though.
In my (limited) experince its really hard to make them stable. But you can make interesting non-linear filters by using different
parameters for the
different halves of the wave cycle. This is what makes the gong sound
gongy (waveguide_nl.h). It could also apply to cone modelling, as the
characteristics of the deflection in the positive and negative directions
will be a bit different.
it's interesting to know that valve preamp stages should have
a small bias (to + iirc), and you can in fact observe this in
the amp output i've recorded.
To make a nonlienar cone filter you could either use
different IIR
parameters for signal sign, or you could us different parameters for
different amplitudes (or both).
i'm still trying to model the hard-driven valve in the time
domain. the more i look at it, the more i think the 'building of
potentials' i've tried to describe comes near it. though it would
be nice to base it on hard facts about valve amp behaviour, too.
interestingly, the sine, shaped by the hard-driven amp, looks
a lot like the plots of 'ion conductivity' and 'Aktionspotential'
from human heart analysis.
unfortunately they don't give a formula in the book i have on my
lap. any heart surgeons around? ;)
Different amplitudes would be equivalent to how you
model nonliear systems
with FIR impulses - I can't believe I didn't think of it before.
in fact you described something along these lines a few days
ago. ;)
tim
12:51 p.m.
On Mon, Oct 28, 2002 at 05:53:26 +0100, Tim Goetze wrote:
That is a side effect of class B amps IIRC. I would have though that
preamps would be class A, but maybe not.
Er, did I? Right. Maybe more sleep would be good ;)
- Steve
make
interesting non-linear filters by using different parameters for the
different halves of the wave cycle. This is what makes the gong sound
gongy (waveguide_nl.h). It could also apply to cone modelling, as the
characteristics of the deflection in the positive and negative directions
will be a bit different.
it's interesting to know that valve preamp stages should have
a small bias (to + iirc), and you can in fact observe this in
the amp output i've recorded. i'm still trying to model the hard-driven valve in
the time
domain. the more i look at it, the more i think the 'building of
potentials' i've tried to describe comes near it. though it would
be nice to base it on hard facts about valve amp behaviour, too.
'building of potentials'? Are you refering to your sinc summation idea?
Its a nice idea, especially as you get anitaliasing for free, but I think
it will be hard to turn a hardish clipping shape into a sum of sincs.
interestingly, the sine, shaped by the hard-driven
amp, looks
a lot like the plots of 'ion conductivity' and 'Aktionspotential'
from human heart analysis.
Woooosh. [that was the sound of that going way over my head ;) ]
Different
amplitudes would be equivalent to how you model nonliear systems
with FIR impulses - I can't believe I didn't think of it before.
in fact you described something along these lines a few days
ago. ;)
2 p.m.
Steve Harris wrote:
no, this is the other idea that would try to model a simple
wave shaper (pushing energies and so on). the sinc idea is
reserved for dire emergency situations, it's not really gentle
on the cpu and tough to get right.
if you could see those plots, you'd know exactly what i mean. in
fact they look almost exactly like one half of a sine cycle
after it's been mistreated by a 12ay7 valve at high gain
(quitte.de/wave.ps.gz).
tim
That is a side effect of class B amps IIRC. I would
have though that
preamps would be class A, but maybe not.
they are, afaik. you're right, it's a property of class B.
still my amp's preamp stage shapes the lower half differently
(less) than the top.
i'm still
trying to model the hard-driven valve in the time
domain. the more i look at it, the more i think the 'building of
potentials' i've tried to describe comes near it. though it would
be nice to base it on hard facts about valve amp behaviour, too.
'building of potentials'? Are you refering to your sinc summation idea?
Its a nice idea, especially as you get anitaliasing for free, but I think
it will be hard to turn a hardish clipping shape into a sum of sincs. interestingly,
the sine, shaped by the hard-driven amp, looks
a lot like the plots of 'ion conductivity' and 'Aktionspotential'
from human heart analysis.
Woooosh. [that was the sound of that going way over my head ;) ] 
3:14 p.m.
[Note: I've only been loosely following this thread, but this message caught
my broadcast engineer eye, being that I work with tube gear nearly daily.]
On Monday 28 October 2002 01:55 pm, Tim Goetze wrote:
Steve Harris wrote:
>That is a side effect of class B amps IIRC. I would have though that
>preamps would be class A, but maybe not.
they are, afaik. you're right, it's a property
of class B.
still my amp's preamp stage shapes the lower half differently
(less) than the top.
It would depend upon where in the transconductance curves the tube is biased,
as well as plate voltage. The behavior you desribe sounds like an underbias
condition, where the tube isn't biased approximately in the middle of its
linear range (for class A). Either that or the plate voltage has sagged (old
dried out electrolytic caps?). And if it's a tetrode or pentode amp, the
screen and suppressor bias level and type figure into it dramatically.
>> i'm still trying to model the hard-driven
valve in the time
>> domain. the more i look at it, the more i think the 'building of
>> potentials' i've tried to describe comes near it. though it would
>> be nice to base it on hard facts about valve amp behaviour, too.
>'building of potentials'? Are you refering
to your sinc summation idea?
>Its a nice idea, especially as you get anitaliasing for free, but I think
>it will be hard to turn a hardish clipping shape into a sum of sincs.
no, this is the other idea that would try to model a
simple
wave shaper (pushing energies and so on). the sinc idea is
reserved for dire emergency situations, it's not really gentle
on the cpu and tough to get right.
Ok, from an engineer's perspective:
It seems to me that modeling the transfer function of a tube would involve
running high level triangles at various fundamental frequencies. This data
would then be analyzed to see the necessary time and frequency domain
artifacts of tube operation. You need to see the group delay, and the
frequency response for the fine details -- but the big issue is the simple
'if I put in X volts I get out Y volts' transfer function. This can be
modeled with finite element techniques if you have the exact parameters of
the tube (plate size and shape, interelectrode spacings, grid dimensions,
filament/cathode emission and dimensions, etc).
I have seen this done for large tubes (4CX15,000A's as used in broadcast
transmitters), but not small ones (6146's, 2E26's, as well as the even
smaller ones like the 12AX7 or AU7, or the venerable (and expensive) 5763, or
even the ubiquitous 6L6 and variants like the 807).
The purpose in the transmitter case is to see what frequencies need what
compensation and potential neutralization. Since the 4CX15,000A is used at
frequencies well over 100MHz, knowing this sort of thing (particularly
interelectrode spacing and coupling) is essential is designing the tuned
cavity output and input networks required at FM broadcast frequencies. But
the old '15,000 would make a hoss of an audio amp -- the 15,000 number is the
rated plate dissipation in watts. A pair in class B push-pull can make 50kW
of output at audio frequencies easily.
Certain frequencies must be avoided for these tubes -- if the frequency
response has an area where the output is 180 degrees from the input
potentially destructive positive feedback can occur. If not properly
neutralized and bypassed, these tubes can self-destruct in a pyrotechnic
manner.
A 4CX15,000A takes around 160A at 12 volts AC for filament, and 10KV for the
plate @ 2A or so, with several hundred CFM of air through its fins... A fun
tube -- particularly when it blows..... I can take a picture of one here
(paperweight dud) if desired.
For what it's worth I dug out my RCA Receiving Tube Handbook and its cousin,
the RCA Transmitting Tube Handbook, from the mid sixties. Some curves are
reproduced there, and I can probably get the curves you need from manuals I
have.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11
4:42 p.m.
Lamar Owen wrote:
[Note: I've only been loosely following this
thread, but this message caught
my broadcast engineer eye, being that I work with tube gear nearly daily.]
On Monday 28 October 2002 01:55 pm, Tim Goetze wrote:
it's a 12AY7, afaik, or a 12AX7 (double triodes, right?). the contacts
looked fairly clean the last time i had them out. remembering what
the circuit board and the resistors and capacitors look like, i
rather wonder at everything in there that is biased properly than the
other way around. ;)
Steve Harris wrote:
>That is a side effect of class B amps IIRC. I would have though that
>preamps would be class A, but maybe not.
they are, afaik. you're right, it's a
property of class B.
still my amp's preamp stage shapes the lower half differently
(less) than the top.
It would depend upon where in the transconductance curves the tube is biased,
as well as plate voltage. The behavior you desribe sounds like an underbias
condition, where the tube isn't biased approximately in the middle of its
linear range (for class A). Either that or the plate voltage has sagged (old
dried out electrolytic caps?). And if it's a tetrode or pentode amp, the
screen and suppressor bias level and type figure into it dramatically. Ok, from an engineer's perspective:
...
I have seen this done for large tubes (4CX15,000A's
as used in broadcast
transmitters), but not small ones (6146's, 2E26's, as well as the even
smaller ones like the 12AX7 or AU7, or the venerable (and expensive) 5763, or
even the ubiquitous 6L6 and variants like the 807).
would you say that a 6L6 (output stage in my fender amp) behaves
vastly different from a 12AX7 when overdriven? it seems to me that
most of the actual distortion, at least in this amp here, is
generated in the preamp stage.
Certain frequencies must be avoided for these tubes --
if the frequency
response has an area where the output is 180 degrees from the input
potentially destructive positive feedback can occur. If not properly
neutralized and bypassed, these tubes can self-destruct in a pyrotechnic
manner.
jimi would have loved them. :)
For what it's worth I dug out my RCA Receiving Tube
Handbook and its cousin,
the RCA Transmitting Tube Handbook, from the mid sixties. Some curves are
reproduced there, and I can probably get the curves you need from manuals I
have.
i'm sure frequency responses and Vin/Vout mappings for common
amp tubes (EL34 comes to mind, too) would be very helpful, but
can i ask you to take a look at
http://quitte.de/driven.gif
first?
it shows my badly-driven fender amp's line output (top) vs. the sine
wave it was fed at the same time. i'm wondering how come the output
overshoots every time the input crosses zero. trying to model this
behaviour into a waveshaper, but i'm a bit lost on the physical laws
that rule here.
there's a schematic of a very similar amp at
http://quitte.de/super_5f4_schem.gif
fwiw. the line-out should plug in right after the last tone
control i reckon.
(the waveshaper will probably alias badly, but i'm hoping it
will give us a good start into the heavy sort of distortion.)
tim
5:36 p.m.
On Mon, Oct 28, 2002 at 10:37:18 +0100, Tim Goetze wrote:
it shows my badly-driven fender amp's line output
(top) vs. the sine
wave it was fed at the same time. i'm wondering how come the output
overshoots every time the input crosses zero. trying to model this
I dont think it does, I still think this is just a phase effect.
OTOH it could just be an extreme case of hard clipping and HP filtering.
- Steve
29 Oct
29 Oct
1:54 p.m.
On Monday 28 October 2002 05:33 pm, Steve Harris wrote:
On Mon, Oct 28, 2002 at 10:37:18 +0100, Tim Goetze
wrote:
> it shows my badly-driven fender amp's line output (top) vs. the sine
> wave it was fed at the same time. i'm wondering how come the output
> overshoots every time the input crosses zero. trying to model this
I dont think it does, I still think this is just a
phase effect.
OTOH it could just be an extreme case of hard clipping
and HP filtering.
I would lean this way as well, since the first interstage coupling caps are a
mere .02 microfarads. Even at the circuit impedance there, you will have
some serious overshoot due to the series capacitance.
See the SPICE project for a circuit simulator program that can handle
waveforms in this sort of circuit. SPICE is as old as dirt, but a very handy
program for electrical engineers. The algorithms used in SPICE, once you
have a working netlist, could theoretically be used on audio waveforms, but
it would take a real machine to get realtime results through SPICE. The nice
thing about SPICE is that, if you have the curves for the active devices, and
the schematics for the circuit as a whole (which is the case here), you can
model very accurately the circuit in question.
To find SPICE, search virtually any archive of freeware in the engineering
programs section. SPICE is the Csound of electrical engineering.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11
28 Oct
28 Oct
6:19 p.m.
On Mon, Oct 28, 2002 at 10:37:18PM +0100, Tim Goetze wrote:
i'm sure frequency responses and Vin/Vout mappings
for common
amp tubes (EL34 comes to mind, too) would be very helpful, but
can i ask you to take a look at
http://quitte.de/driven.gif
first?
it shows my badly-driven fender amp's line output (top) vs. the sine
wave it was fed at the same time. i'm wondering how come the output
overshoots every time the input crosses zero. trying to model this
behaviour into a waveshaper, but i'm a bit lost on the physical laws
that rule here.
looking at the distorted signal it look to me like an amp that is clipping,
but it's power supply can't maintain the voltage (eg. relatively high
impendance on the powersupply/series resistance). While +ve or -ve side of
the output pair is used it drains the capacitor from the power supply, hence
the exponential slope from 'spike' till drop.
Frank.
--
+---- --- -- - - - -
| Frank van de Pol -o) A-L-S-A
| FvdPol(a)home.nl /\\ Sounds good!
| http://www.alsa-project.org _\_v
| Linux - Why use Windows if we have doors available?
29 Oct
29 Oct
3:21 a.m.
On Tue, Oct 29, 2002 at 12:18:23 +0100, Frank van de Pol wrote:
looking at the distorted signal it look to me like an
amp that is clipping,
but it's power supply can't maintain the voltage (eg. relatively high
impendance on the powersupply/series resistance). While +ve or -ve side of
the output pair is used it drains the capacitor from the power supply, hence
the exponential slope from 'spike' till drop.
Aha, great. I was thinking about how to rewrite the valve rectifier, and
it occured to me that my revised algorithm would output something that
looked like that in extreme cases.
That means 1) we're nearer to Tim's shape, and 2) my alg. is more or less
correct 3) we may not need to mess about with chebies.
- Steve
8:25 a.m.
Steve Harris wrote:
On Tue, Oct 29, 2002 at 12:18:23 +0100, Frank van de
Pol wrote:
1 + 2 + 3 = great. :)
i've uploaded another image showing how the line output from
the amp changes as the incoming sine increases in volume:
http://quitte.de/3x-driven.gif
admit i hope that frank's explanation holds, and that the
valve_rect can be made to produce similar shaping.
tim
looking at the distorted signal it look to me
like an amp that is clipping,
but it's power supply can't maintain the voltage (eg. relatively high
impendance on the powersupply/series resistance). While +ve or -ve side of
the output pair is used it drains the capacitor from the power supply, hence
the exponential slope from 'spike' till drop.
Aha, great. I was thinking about how to rewrite the valve rectifier, and
it occured to me that my revised algorithm would output something that
looked like that in extreme cases.
That means 1) we're nearer to Tim's shape, and 2) my alg. is more or less
correct 3) we may not need to mess about with chebies. 
10:31 a.m.
On Tue, Oct 29, 2002 at 02:20:32 +0100, Tim Goetze wrote:
i've uploaded another image showing how the line
output from
the amp changes as the incoming sine increases in volume:
http://quitte.de/3x-driven.gif
admit i hope that frank's explanation holds, and that the
valve_rect can be made to produce similar shaping.
Hmmm... if that is sag from the powersupply then it kicks in much lower
than I though. Another thing I guessed wrongly. Should make it easier to
implement though :) - more margin of error.
- Steve
2:12 p.m.
On Tuesday 29 October 2002 10:27 am, Steve Harris wrote:
On Tue, Oct 29, 2002 at 02:20:32 +0100, Tim Goetze
wrote:
> i've uploaded another image showing how the line output from
> the amp changes as the incoming sine increases in volume:
> admit i hope that frank's explanation holds,
and that the
> valve_rect can be made to produce similar shaping.
Hmmm... if that is sag from the powersupply then it
kicks in much lower
than I though. Another thing I guessed wrongly. Should make it easier to
implement though :) - more margin of error.
I should have read this one before replying previously. This most certainly
looks like the result of undersize interstage coupling caps -- a high pass
filter.
With the currents in those tubes being at most a few milliamps, and the power
supply filter caps at 16 microfards, with a choke in the PI arrangement no
less, it is unlikely this is sag, unless the third filter cap (there's a
double-pi power supply filter here -- the first stage of the pi, the one with
the choke, feeds the output tubes. The second stage, since the current is so
much lower, uses a resistor in the top leg of the pi, which is perfectly fine
for this application) is very leaky.
If the power supply was sagging, the 6L6's would drop first, being that they
need more current. You can measure the voltage rails to see (using
appropriate caution, since there is lethal voltage in this circuit) if they
sag.
Now, there are two interstage coupling caps of interest. One is between the
12AY7 first stage and the cascade connected 12AX7, and the other follows the
12AX7 cascade. I can't read the value on the second one. (I know the first
stage is actually a mixed twin stage -- it is simpler to drop one side of the
mixer and treat it as a single stage, from the signal's point of view).
Can a trace be made in the middle? I'm looking for a signal trace situation
here -- the signal after each active component. Of course, you need a scope
that can handle the voltage there, or build a series capacitor audio probe
first.
While much of the 'tube sound' comes from tubes, much more comes from the
types of components used in tube circuits -- the circuit impedance dictates
the technology of the capacitor, for instance. The 'tube sound' can be had
with FET's, as long as they are biased right and you use depletion mode
devices.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11
30 Oct
30 Oct
9:26 a.m.
Lamar Owen wrote:
...
ok, with all of you pointing in that direction, i took a look
at the waveforms a valve_1209 and a hipass produce for a sine.
at lower distortion levels, they do in fact come really close.
no surprise there, the combination is proven to sound good.
comparing the frequency spectra of our setup to the amp output
reveals that there is quite a difference. it is visible at low
gain settings and, as expected, becomes emphasized at higher
distortion levels.
the amp output frequency spectrum changes quite noticeably the
more gain we apply. i've uploaded plots relating to the same
waves that are shown in '3x-driven' to
http://quitte.de/3x-spectrum.gif
(the first spike is unrelated 50 Hz hum if i'm not badly
mistaken.)
maybe it would be possible to use two or more chebyshevs to
create the harmonics needed to emulate this behaviour, and then
blend between them according to the simulated gain?
Hmmm... if
that is sag from the powersupply then it kicks in much lower
than I though. Another thing I guessed wrongly. Should make it easier to
implement though :) - more margin of error.
I should have read this one before replying previously. This most certainly
looks like the result of undersize interstage coupling caps -- a high pass
filter. 
If the power supply was sagging, the 6L6's would
drop first, being that they
need more current. You can measure the voltage rails to see (using
appropriate caution, since there is lethal voltage in this circuit) if they
sag.
the waves shown in the above image were taken from the amp's
line-out with the main volume at min, so it's very unlikely
power supply sag is an important factor i come to think.
Can a trace be made in the middle? I'm looking for
a signal trace situation
here -- the signal after each active component. Of course, you need a scope
that can handle the voltage there, or build a series capacitor audio probe
first.
i'm afraid that's not an option -- the only scope i call my own
is a pure software solution; therefore i'm limited to what i
can feed an audio interface. would this series capacitor thing
be capable of scaling down the voltages appropriately?
While much of the 'tube sound' comes from
tubes, much more comes from the
types of components used in tube circuits -- the circuit impedance dictates
the technology of the capacitor, for instance. The 'tube sound' can be had
with FET's, as long as they are biased right and you use depletion mode
devices.
in fact there are some pretty good analog stomp boxes out there
that sound really valvy. i bet they make use of FETs then.
tim
10:36 a.m.
On Wed, Oct 30, 2002 at 03:22:35PM +0100, Tim Goetze wrote:
in fact there are some pretty good analog stomp boxes
out there
that sound really valvy. i bet they make use of FETs then.
Nearly every stomp box "tube distortion" i've ever opened,
or looked at a circuit for, uses diodes for clipping.
Ibanez TubeScreamer
Danelectro Daddy-O
MXR Distortion+
Pro Co RAT
Peavey Tube-Sound Distortion
Some notable exceptions:
Craig Anderton's Tube Sound Distortion, sounds pretty good:
(uses two stages of a 4049 CMOS inverter for distortion!)
EH Hot Tubes Distortion, haven't heard it:
(similar, but uses 3 stages of the 4049)
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
11:01 a.m.
On Wed, Oct 30, 2002 at 03:22:35 +0100, Tim Goetze wrote:
comparing the frequency spectra of our setup to the
amp output
reveals that there is quite a difference. it is visible at low
gain settings and, as expected, becomes emphasized at higher
distortion levels.
the amp output frequency spectrum changes quite noticeably the
more gain we apply. i've uploaded plots relating to the same
waves that are shown in '3x-driven' to
Thanks for doing that, I haven't had the time. Can you produce a spectrum
plot for the current s/w chain?
maybe it would be possible to use two or more
chebyshevs to
create the harmonics needed to emulate this behaviour, and then
blend between them according to the simulated gain?
It would be more efficient to just calculate the corect chebyshev in
realtime, the problem is that they have lienar CPU cost with the number of
harmonics, 20 harmonics for example will be pretty expensive. As we are
building on the existing harmonics generated by the valve sim though we
shouldn't need that many.
- Steve
12:15 p.m.
Steve Harris wrote:
Thanks for doing that, I haven't had the time. Can
you produce a spectrum
plot for the current s/w chain?
http://quitte.de/sine-spectrum.gif
http://quitte.de/straight-spectrum.gif
the first is generated by this chain:
Sine (1046.5022612023945), # c'''
Valve (.3, .7),
HiPass (200),
Inverter(),
Rect (.5, .6),
Inverter(),
SineShaper (2.),
Valve (.7, .8),
HiPass (30),
the second is the same chain with the SineShaper left out.
It would be more efficient to just calculate the corect
chebyshev in
realtime, the problem is that they have lienar CPU cost with the number of
harmonics, 20 harmonics for example will be pretty expensive. As we are
building on the existing harmonics generated by the valve sim though we
shouldn't need that many.
there's one problem i see: if we employ a chebyshev, it is going
to create harmonics no matter what amplitude our incoming signal
is iiutc. this is unlike the real thing, which will produce a
different spectrum if one is playing softly, or with the volume
control set low. (i'm fond of this effect because it allows me to
change the sound pretty drastically without having to walk over
to the amp.)
tim
1:16 p.m.
On Wed, Oct 30, 2002 at 06:11:11 +0100, Tim Goetze wrote:
I was planning to vary the generated harmonics with incoming
amplitude, but it doesn't look like this is the right appraoch, there are
allready plenty of strong harmonics, a cheby would just make it muddier.
The strong presence of low harmonics is probably the cause of the
softness.
- Steve
Steve Harris wrote:
The sine spectrum looks pretty close to me, except that some of the low
harmonics are too high, probasbly cutting back some of the parameters to
the valve would fix this.
I would guess that if we can fix the low harmonics the right thing to do
is to modulate the strength of the sin shaper with incoming amplitude.
Thanks for doing that, I haven't had the time.
Can you produce a spectrum
plot for the current s/w chain?
http://quitte.de/sine-spectrum.gif
http://quitte.de/straight-spectrum.gif It would be
more efficient to just calculate the corect chebyshev in
realtime, the problem is that they have lienar CPU cost with the number of
harmonics, 20 harmonics for example will be pretty expensive. As we are
building on the existing harmonics generated by the valve sim though we
shouldn't need that many.
there's one problem i see: if we employ a chebyshev, it is going
to create harmonics no matter what amplitude our incoming signal
is iiutc. this is unlike the real thing, which will produce a
different spectrum if one is playing softly, or with the volume
4:37 p.m.
Steve Harris wrote:
The sine spectrum looks pretty close to me, except that some of the low
harmonics are too high, probasbly cutting back some of the parameters to
the valve would fix this.
i've been mucking around a little but it seems real hard to
make the harmonics fit, especially with such a wide range of
possible mixes being aimed at.
I was planning to vary the generated harmonics with
incoming
amplitude, but it doesn't look like this is the right appraoch, there are
allready plenty of strong harmonics, a cheby would just make it muddier.
i can't say i really like the idea of dynamic parameters,
sounds a bit messy to me. to make it work smoothly, there'd
be some discontinuity every time parameters change, or one'd
have to overlap blocks, and how would one go about choosing
their size to begin with ...
The strong presence of low harmonics is probably the
cause of the
softness.
it seems hard to come up with a wave shaper that favours higher
harmonics, i've tried some in the past few days. most of the
shapers that compress the waveform favour even harmonics, while
shapers based on even powers of the input sample generate only
odd harmonics. they're hard to distribute though. afaict, the
mechanism is refined in the chebyshev (even powers = odd harm.
and vice versa).
tim
5:34 p.m.
On Wed, Oct 30, 2002 at 10:30:48PM +0100, Tim Goetze wrote:
it seems hard to come up with a wave shaper that
favours higher
harmonics, i've tried some in the past few days.
maybe you're not filtering enough?
i was just reading about the Ibanez TubeScreamer circuit,
its clipping stage is quite interesting.
http://www.geofex.com/Article_Folders/TStech/tsxtech.htm
There's a pair of parallel diodes in the negative feedback
loop (rather than across the output), and there's
a cap and resistor in series from the neg. input to ground
forming a highpass filter (that is, there's less negative
feedback from high freqs so the circuit gets higher gain
at high freqs).
The filter is at about 750 Hz.
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
31 Oct
31 Oct
4:14 a.m.
On Wed, Oct 30, 2002 at 02:30:53 -0800, Paul Winkler wrote:
On Wed, Oct 30, 2002 at 10:30:48PM +0100, Tim Goetze
wrote:
Thats definatly possible, we could just make the HP filter deeper and
inject a bit of the source signal back into the output (with an apropriate
ammount of delay).
- Steve
it seems hard to come up with a wave shaper that
favours higher
harmonics, i've tried some in the past few days.
maybe you're not filtering enough?
4:17 a.m.
On Wed, Oct 30, 2002 at 10:30:48 +0100, Tim Goetze wrote:
I meant incoming envelope, I think its inevitable given the process that a
valve amp puts the signal through. Infact this is what the rectifier does
now, its just that the shaper isn't right.
- Steve
i've been mucking around a little but it seems
real hard to
make the harmonics fit, especially with such a wide range of
possible mixes being aimed at.
Yup. Hey, if it was easy, everyone would be doing it ;)
I was planning
to vary the generated harmonics with incoming
amplitude, but it doesn't look like this is the right appraoch, there are
allready plenty of strong harmonics, a cheby would just make it muddier.
i can't say i really like the idea of dynamic parameters,
sounds a bit messy to me. to make it work smoothly, there'd
28 Oct
28 Oct
3:54 a.m.
On Mon, Oct 28, 2002 at 02:12:02 +0100, Tim Goetze wrote:
the longer the wave is on one side of zero, the more
energy is
built up that tries to push it back. as long as the system is
fed enough energy to keep it this side of zero, it is basically
decaying exponentially against a (more-or-less) constant value.
...
i am not sure how to do it so it doesn't alias,
which i fear a
naive implementation would do.
I don't know its hard to seperate phase effects from waveshaping, thats
why it needs FFTing. Onve we have the aproximate harmonic series we can
simulate that with a cheby, and if neccesary put the phase back with an
allpass or two.
Generally phase effects aren't really audible, but they may react with
distortion etc. to make a difference.
- Steve
1:46 p.m.
Steve Harris wrote:
I don't know its hard to seperate phase effects
from waveshaping, thats
why it needs FFTing. Onve we have the aproximate harmonic series we can
simulate that with a cheby, and if neccesary put the phase back with an
allpass or two.
ok, i'm slowly beginning to understand this i think. you mean
you plan to generate the harmonics with a time domain shaper
(like beloved valve_1209) and then 'shape the freq domain' with
the chebyshev?
if this is what you mean, i'm not sure how different it will
sound across the input frequency range. for a guitar, we have
close to three octaves, and if the chebyshev has a static
frequency response, there will be noticeable difference afaict.
though i may have a wrong picture here.
Generally phase effects aren't really audible, but
they may react with
distortion etc. to make a difference.
i don't think phase shifting will make a difference (and if
it does, it's good to know you can fix it ;)
tim
5:29 p.m.
On Mon, Oct 28, 2002 at 07:42:04 +0100, Tim Goetze wrote:
Not really. Both are freq (well, and amplitude) domain only.
The chebyshev is a way of generating specific harmonics, eg. you can
output the fundamental + the 2nd harmonic at -6dB, the 4th at -16, the 6th
at -21, etc. You can't easily control the phase though.
The harmonics track the source as you would expect, regardless of
frequency range (though aliasing is an issue, ofcourse).
I don't
know its hard to seperate phase effects from waveshaping, thats
why it needs FFTing. Onve we have the aproximate harmonic series we can
simulate that with a cheby, and if neccesary put the phase back with an
allpass or two.
ok, i'm slowly beginning to understand this i think. you mean
you plan to generate the harmonics with a time domain shaper
(like beloved valve_1209) and then 'shape the freq domain' with
the chebyshev? i don't think phase shifting will make a
difference (and if
it does, it's good to know you can fix it ;)
OK, then all I need to do is split out the distorted sinwave and load it
into octave. How Hard Can It Be(TM). ;)
Fix it is probably the wrong term, I can sort of bash it into compliance.
- Steve
6:10 p.m.
Steve Harris wrote:
The chebyshev is a way of generating specific
harmonics, eg. you can
output the fundamental + the 2nd harmonic at -6dB, the 4th at -16, the 6th
at -21, etc. You can't easily control the phase though.
gee, i remember. i've just taken a look at harmonic_gen_1220,
this must be what you mean. fine if it sounds good, but it's a
serious argument for getting hold of a few more MHz, i bet. :)
> i don't think phase shifting will make a
difference (and if
> it does, it's good to know you can fix it ;)
will not make a difference unless we apply another shaper, now
i understand. i don't think we'll need another if the chebyshev
does well.
tim
27 Oct
27 Oct
3:09 p.m.
Steve Harris wrote:
ecasound -i jack_alsa,in_1 -o jack_alsa,out_1
-el:valve,0.6,0.7 -efh:25
-el:inv -el:valveRect,0.5,0.6 -el:inv -el:sinusWavewrapper,2
-el:valve,0.7,0.8 -efh:30 -el:unmatched
god save our glorious steve! this is really beginning to sound
like an amp. i've found it sounds better for the guitar if the
first hipass is set to something like 100-600 Hz, my personal
pref is 600 (on my 'real' amp, the 'bass' pot is set to zero).
fiddling with the second hipass creates some interesting sounds,
but 30 Hz seems a good compromise.
any thoughts how to tackle the compression issue?
tim
5:59 p.m.
On Sun, Oct 27, 2002 at 08:58:10 +0100, Tim Goetze wrote:
Cheers :) I have no real idea what it shouldsound like, so I just tweaked
it until it sounded OK. I think the key is the sinus, it gives the double
dump from the british amp scope-shot.
ecasound -i
jack_alsa,in_1 -o jack_alsa,out_1 -el:valve,0.6,0.7 -efh:25
-el:inv -el:valveRect,0.5,0.6 -el:inv -el:sinusWavewrapper,2
-el:valve,0.7,0.8 -efh:30 -el:unmatched
god save our glorious steve! this is really beginning to sound
like an amp. i've found it sounds better for the guitar if the
first hipass is set to something like 100-600 Hz, my personal
pref is 600 (on my 'real' amp, the 'bass' pot is set to zero). fiddling with the second hipass creates some
interesting sounds,
but 30 Hz seems a good compromise.
any thoughts how to tackle the compression issue?
Yeah, fix the rectifier. I had a quick look at it and it looks like I did
something fairly obvious (its basicly a compressor, but with a waveshaper
instead of a linear gain).
- Steve
8:45 p.m.
Steve Harris wrote:
Cheers :) I have no real idea what it shouldsound like,
so I just tweaked
it until it sounded OK. I think the key is the sinus, it gives the double
dump from the british amp scope-shot.
/me plays another one, about britannia ruling the waves ...
make that pref 200 Hz for me. this and the sine shaper at .4, and
tell you what, this sounds very much like, but in fact better to
me than the sound i play 90% of the time on the real thing. only
drawback is it becomes faint far too quick when turning the vol
down at the guitar, the compression woe.
i've made a quick recording of what this sounds like:
http://quitte.de/easy.mp3
Yeah, fix the rectifier. I had a quick look at it and
it looks like I did
something fairly obvious (its basicly a compressor, but with a waveshaper
instead of a linear gain).
the shaper code looks familiar. admit i'm hoping for a performance
and quality gain.
tim
10:01 p.m.
On Mon, Oct 28, 2002 at 02:41:08AM +0100, Tim Goetze wrote:
make that pref 200 Hz for me. this and the sine shaper
at .4, and
tell you what, this sounds very much like, but in fact better to
me than the sound i play 90% of the time on the real thing. only
drawback is it becomes faint far too quick when turning the vol
down at the guitar, the compression woe.
how are you plugging in your guitar?
if it's a guitar with passive electronics and you're going
straight to the A/D converter, i'd expect it to be
overly sensitive to the volume control.
i've made a quick recording of what this sounds
like:
http://quitte.de/easy.mp3
Nice!
is this with ecasound?
how did you record it - straight from ecasound, or piping
back out to a speaker / microphone?
I want to get a chance to try this for myself :(
(too busy)
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
28 Oct
28 Oct
8:25 a.m.
Paul Winkler wrote:
i've been using my audio/midi processor for this one. since
it's only ladspa fx, inverters and hipasses, and since i've
ripped the hipass filter coefficient calculation from
ecasound (thanks kai!), it is, as steve put it, 'mathematically
equivalent' to the ecasound command line steve posted (with the
above modifications).
for the recording, the output of '-el:unmatched' is written
straight to the disk, there's really nothing analog involved
after the signal enters the box. huge :)
On Mon, Oct 28, 2002 at 02:41:08AM +0100, Tim Goetze
wrote:
yes, it's passive, two humbuckers. it's going through the
mixer first, the signal's not strong enough to go AD directly.
playing to the control out instead of through the box, the
vol pot sensitivity is what you expect.
make that pref 200 Hz for me. this and the sine
shaper at .4, and
tell you what, this sounds very much like, but in fact better to
me than the sound i play 90% of the time on the real thing. only
drawback is it becomes faint far too quick when turning the vol
down at the guitar, the compression woe.
how are you plugging in your guitar?
if it's a guitar with passive electronics and you're going
straight to the A/D converter, i'd expect it to be
overly sensitive to the volume control. i've made
a quick recording of what this sounds like:
http://quitte.de/easy.mp3
Nice!
is this with ecasound?
how did you record it - straight from ecasound, or piping
back out to a speaker / microphone? I want to get a chance to try this for myself :(
(too busy)
ditch those web pages! it's only money ... ;)
tim
9:08 a.m.
On Mon, Oct 28, 2002 at 01:56:34 +0100, Tim Goetze wrote:
Thats something that worried me, I seem to remeber that the sound of a
guitar changes if its not impedance matched with the AD converter. I'm
pretty sure, mine doesn't match passive guitars. My bass is active though,
and I dont have any DI boxes, so no choice anyway.
Sleep is for the weak (or is that week) ;)
- Steve
how are you
plugging in your guitar?
if it's a guitar with passive electronics and you're going
straight to the A/D converter, i'd expect it to be
overly sensitive to the volume control.
yes, it's passive, two humbuckers. it's going through the
mixer first, the signal's not strong enough to go AD directly.
playing to the control out instead of through the box, the
vol pot sensitivity is what you expect. I want to get a
chance to try this for myself :(
(too busy)
ditch those web pages! it's only money ... ;)
9:42 a.m.
Steve Harris wrote:
Sleep is for the weak (or is that week) ;)
yeah, coffee is cheap (better than counting sheep) ;)
tim
On Mon, Oct 28, 2002 at 01:56:34 +0100, Tim Goetze
wrote:
Thats something that worried me, I seem to remeber that the sound of a
guitar changes if its not impedance matched with the AD converter. I'm
pretty sure, mine doesn't match passive guitars. My bass is active though,
and I dont have any DI boxes, so no choice anyway.
likewise here. anyone knows a suitable AD box that can
team up with coaxial digital in?
using the mixer as preamp probably isn't the best solution,
but it works ok so far and gives some flexibility. i'd like
to make use of the digital in though, two-channel AD is a
bit restricting (m-man audiophile 24/96).
how are
you plugging in your guitar?
if it's a guitar with passive electronics and you're going
straight to the A/D converter, i'd expect it to be
overly sensitive to the volume control.
yes, it's passive, two humbuckers. it's going through the
mixer first, the signal's not strong enough to go AD directly.
playing to the control out instead of through the box, the
vol pot sensitivity is what you expect. I want to get a chance to try this for myself :(
(too busy)
ditch those web pages! it's only money ... ;)
10:04 a.m.
On Mon, Oct 28, 2002 at 03:37:36PM +0100, Tim Goetze wrote:
No, but The Right Thing (tm) is a direct box connected to
a mic preamp. The ones under about US $50 generally are crap.
--PW
--
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
Thats something
that worried me, I seem to remeber that the sound of a
guitar changes if its not impedance matched with the AD converter. I'm
pretty sure, mine doesn't match passive guitars. My bass is active though,
and I dont have any DI boxes, so no choice anyway.
likewise here. anyone knows a suitable AD box that can
team up with coaxial digital in?
11:43 a.m.
Steve Harris wrote:
On Mon, Oct 28, 2002 at 01:56:34 +0100, Tim Goetze
wrote:
Thats something that worried me, I seem to remeber that the sound of a
guitar changes if its not impedance matched with the AD converter. I'm
pretty sure, mine doesn't match passive guitars. My bass is active though,
and I dont have any DI boxes, so no choice anyway.
sorry, i've been unprecise again. i meant, 'what one expects
from a decent instrument preamp' rather than 'what paul
describes'.
in fact the mixer preamp volume sensitivity is ok (it has
been built with mics and guitars in mind i guess). of course
the mixer preamp lacks the compression of an instrument amp.
tim (trying to exercise think-before-write for a change)
how are
you plugging in your guitar?
if it's a guitar with passive electronics and you're going
straight to the A/D converter, i'd expect it to be
overly sensitive to the volume control.
yes, it's passive, two humbuckers. it's going through the
mixer first, the signal's not strong enough to go AD directly.
playing to the control out instead of through the box, the
vol pot sensitivity is what you expect.
5:24 p.m.
Steve Harris wrote:
Sleep is for the weak (or is that week) ;)
- Steve
Thats something that worried me, I seem to remeber that
the sound of a
guitar changes if its not impedance matched with the AD converter. I'm
pretty sure, mine doesn't match passive guitars. My bass is active though,
and I dont have any DI boxes, so no choice anyway.
Yeah, you really should be going through a DI box if you're plugging a
guitar into a console. The mixer
will be able to give you the necessary gain boost, but I don't that's
sonically or electronically equivalent to
the pre-amp stage of an amp or direct box.
-dgm
I want to
get a chance to try this for myself :(
(too busy)
ditch those web pages! it's only money ... ;)
3:47 a.m.
On Mon, Oct 28, 2002 at 02:41:08 +0100, Tim Goetze wrote:
That sounds pretty good. The attack sounds a bit muted to me though, is
that deliberate or should it have more bite?
Yup, I adapted what I had around ;) I think something symmetric and harder
might be better though, I'l see what I can do. Maybe it should just be a
peak following compressor... my original though was that the power supply
sag would cause a distortion, but maybe it wont.
I think the speed will go up dramatically when its bolted into one plugin.
I want to get the ordering right first though. OTOH there is still mic
effects to be added in :(
- Steve
Yeah, fix the
rectifier. I had a quick look at it and it looks like I did
something fairly obvious (its basicly a compressor, but with a waveshaper
instead of a linear gain).
the shaper code looks familiar. admit i'm hoping for a performance
and quality gain.
8:25 a.m.
Steve Harris wrote:
Yup, I adapted what I had around ;) I think something symmetric and harder
might be better though, I'l see what I can do. Maybe it should just be a
peak following compressor... my original though was that the power supply
sag would cause a distortion, but maybe it wont.
whatever it does, it helps. ;) probably because it adds more
distortion in the attack phase. so it may not do exactly what
is intended on the smaller scale, but what is intended to be
modeled overall.
On Mon, Oct 28, 2002 at 02:41:08 +0100, Tim Goetze
wrote:
That sounds pretty good. The attack sounds a bit muted to me though, is
that deliberate or should it have more bite?
it could do with a little more bite, yes. it's the neck pickup
with the tone pot at the guitar closed slightly, that's the way
i usually play the super 60. this way, it stress-tests our setup's
capability of generating higher-order harmonics during attack.
Yeah, fix the rectifier. I had a quick look at it and
it looks like I did
something fairly obvious (its basicly a compressor, but with a waveshaper
instead of a linear gain).
the shaper code looks familiar. admit i'm hoping for a performance
and quality gain. I think the speed will go up dramatically when its
bolted into one plugin.
I want to get the ordering right first though. OTOH there is still mic
effects to be added in :(
speaking for my selfish self, i'm just glad i don't have to
care about mic placement at all anymore.
i'll be happy to create more iir 'convolvers' or extend
'unmatched' for more impulses though, as long as they can be
made to converge.
tim
25 Oct
25 Oct
5:37 p.m.
Steve
Hi. The input impedance of a real amp, along with the output impedance of
the pickups in the guitar, are actually quite sensitive to the loading of
the 1/4" cable going between the two. The volume control is a resistor,
essentially, so you've got an RLC circuit looking into the cable which
becomes a low pass when the resistance goes up.
I hear all these same things when I use the inputs on the Pro Tools
directly. The Pod Pro has a better input amplifier that's much more matched
to the output levels and impedances coming from the guitar.
You might try playing around with the length and type of guitar cable.
Try a 5' and a 20'. I can hear differences with my Pod Pro.
Mark
-----Original Message-----
From: linux-audio-dev-admin(a)music.columbia.edu
[mailto:linux-audio-dev-admin@music.columbia.edu]On Behalf Of Tim Goetze
Sent: Friday, October 25, 2002 1:15 PM
To: linux-audio-dev(a)music.columbia.edu
Subject: Re: [linux-audio-dev] [ann] unmatched - a LADSPA amp tone
Steve Harris wrote:
* the sound gets muddy and faint when you turn down the
volume at the instrument, instead of keeping loudness and
reducing distortion.
24 Oct
24 Oct
4:15 a.m.
On Thu, Oct 24, 2002 at 01:50:29 +0200, Tim Goetze wrote:
Steve Harris wrote:
Er, one more. There is also the valve rectifier plugin, which should
produce some of the compression effects in amps. The time scale probably
needs changing though - I was guessing.
I think it should go between the "preamp" and "poweramp" valves. It
includes the effect caused by power supply line sag. Which is aparently
important.
- Steve
Stacking the valve plugins is the key to making
them sound interesting.
ah, now i dig it. maybe you could make a note about the intended
use in the valve doc?
22 Oct
22 Oct
12:58 p.m.
On Tue, Oct 22, 2002 at 12:41:38PM +0100, Steve Harris wrote:
I recently acquired my first vintage amp -
a '63 Gibson GA-30 RVT in very good shape.
It's pretty versatile and doesn't sound quite like anything else
I've played. I'd be happy to donate some impulses, with various
mics / positions and amp settings, if somebody can tell me a good way
to do it. How do you generate the impulse? (I've tried simply creating
a sample with 1 frame at maximum followed by zeroes, but this seems
not to make it through the D/A conversion - it comes out as dead
silence.)
I also have some very odd impulses lying around somewhere if
I can find them - made by popping balloons in and around
some 20' by 5' steel tanks that were in the basement of
a converted industrial building I used to live in.
--
Paul Winkler
"Welcome to Muppet Labs, where the future is made - today!"
the recording
seems to be of decent quality, and the iir
response irons away most of the noise anyway. but the most
important thing is i like the sound of it, which i do a lot.
i've tried about every of your impulses and, would you
believe it, liked the fenders the least. i regularly play
a fender super 60, for ten years or so. :) got to take
a response from it someday myself.
Sadly heres where my zero knowledge of amps kicks in ;) I wouldn't know a
fender form a hole in the ground.
1:34 p.m.
On Tue, Oct 22, 2002 at 09:55:07AM -0700, Paul Winkler wrote:
I recently acquired my first vintage amp -
a '63 Gibson GA-30 RVT in very good shape.
It's pretty versatile and doesn't sound quite like anything else
I've played. I'd be happy to donate some impulses, with various
mics / positions and amp settings, if somebody can tell me a good way
to do it. How do you generate the impulse? (I've tried simply creating
a sample with 1 frame at maximum followed by zeroes, but this seems
not to make it through the D/A conversion - it comes out as dead
silence.)
OK, well an alternative, for EQ readings is to push whitenoise though.
This doesn't work too well for convolving though, if you want to make it
compatible with a convolver then I suggest you make a sine sweep (10
seconds long, 50Hz to 20kHz should be plenty), and record that. You can
then deconvolve the recording with the sweep and you are left with an
impulse.
I've not seen the DA converter problem you mention before, that's
interesting.
I also have some very odd impulses lying around
somewhere if
I can find them - made by popping balloons in and around
some 20' by 5' steel tanks that were in the basement of
a converted industrial building I used to live in.
Mmmmmm. Yeah, that would be cool if you can find them. We have some old,
linked salt water storage tanks at work that desperatly need impulses
taking. The problem is that its really damp down there, so I didn't want
to take monitors down, and I dont have any acceptable mics that dont need
phantom power. I never thought of a baloon, I was going to try a kids cap
gun.
I have a bigish library of impulses, but I can't remeber where most of
them came from, so I'm reluctant to release them.
- Steve
1:43 p.m.
On Tue, Oct 22, 2002 at 06:31:16PM +0100, Steve Harris wrote:
Everybody should have at least one good dynamic mic!
I've got a Sennheiser 421. An EV RE-20 would be even nicer.
I've not seen the DA converter problem you mention
before, that's
interesting.
What do you usually do?
I also have
some very odd impulses lying around somewhere if
I can find them - made by popping balloons in and around
some 20' by 5' steel tanks that were in the basement of
a converted industrial building I used to live in.
Mmmmmm. Yeah, that would be cool if you can find them. We have some old,
linked salt water storage tanks at work that desperatly need impulses
taking. The problem is that its really damp down there, so I didn't want
to take monitors down, and I dont have any acceptable mics that dont need
phantom power. I never thought of a baloon, I was going to try a kids
cap
gun.
Didn't think of that.
Balloon pops probably have some coloration, but they're much stronger and
probably less colored than handclaps... I don't have a spark-gap
generator--I don't even know what it is really, but I've heard
those are the "proper" tool for acoustic impulses.
The problem with balloons is you have to keep blowing up more
as each one is only good for one go :)
--
Paul Winkler
"Welcome to Muppet Labs, where the future is made - today!"
6:47 p.m.
On Tue, Oct 22, 2002 at 10:40:23 -0700, Paul Winkler wrote:
People often use starter pistols, but I think that would be seriously
unpleasent in a small, concrete tank, and they are hard to get hold of.
You have to deconvolve them, which requires getting an uncouloured
recording of the pistol, which is probably also hard.
Everybody should have at least one good dynamic mic!
I've got a Sennheiser 421. An EV RE-20 would be even nicer.
I have an EV dynamic mic, but it is a vocal mic of some kind, and it has
seen better days, the grill thing round the ouside is dinted and rusty!
It was never that sensitive anyway. My preffered mic is a big ugly black
octava condenser.
I never
thought of a baloon, I was going to try a kids cap
gun.
Didn't think of that. Balloon pops probably have some coloration, but
they're much stronger and
probably less colored than handclaps... I don't have a spark-gap
generator--I don't even know what it is really, but I've heard
those are the "proper" tool for acoustic impulses.
Accoring to google, a spark-gap generator is a bigish dangerous looking
van-de-graff type thing. Hmmm... impulse gathering death sports ;)
The circuits look pretty simple, but I'm not safe with soldering irons.
- Steve
7 p.m.
On Tue, Oct 22, 2002 at 11:44:42PM +0100, Steve Harris wrote:
On Tue, Oct 22, 2002 at 10:40:23 -0700, Paul Winkler
wrote:
I said "good". :)
There is a world of difference between an RE-20 and most dynamic mics.
A few other cool ones are AKG-D220/222/224 (hard to find)
and Sennheiser MD-441 (a bit fragile and very expensive).
Unlike your average stage vocal mic, any of these can make a useable
recording of most source sounds: vocals, drums, amps, acoustic
instruments, animals, trucks....
--PW, refugee from rec.audio.pro
Paul Winkler
http://www.slinkp.com
"Welcome to Muppet Labs, where the future is made - today!"
Everybody should have at least one good dynamic
mic!
I've got a Sennheiser 421. An EV RE-20 would be even nicer.
I have an EV dynamic mic, but it is a vocal mic of some kind, and it has
seen better days, the grill thing round the ouside is dinted and rusty!
2:06 p.m.
Paul Winkler wrote:
I recently acquired my first vintage amp -
a '63 Gibson GA-30 RVT in very good shape.
It's pretty versatile and doesn't sound quite like anything else
I've played. I'd be happy to donate some impulses, with various
mics / positions and amp settings, if somebody can tell me a good way
to do it. How do you generate the impulse? (I've tried simply creating
a sample with 1 frame at maximum followed by zeroes, but this seems
not to make it through the D/A conversion - it comes out as dead
silence.)
maybe widening the pulse will help. steve? i've yet to take one
myself, but if you come up with one you'll get an optimized plugin
for it for free if i can get it to converge (so you can tell me
whether you recognize the sound at all ;).
I also have some very odd impulses lying around
somewhere if
I can find them - made by popping balloons in and around
some 20' by 5' steel tanks that were in the basement of
a converted industrial building I used to live in.
i'd love to have one of those. if nothing else, they'll provide
excellent study material for our intrepid voyagers into the
sonic unknown.
tim
5:09 p.m.
On Tue, Oct 22, 2002 at 08:01:35PM +0200, Tim Goetze wrote:
I just realized I probably won't be able to get at that old DAT
tape for at least another 4 months - it's 700 miles away. :(
Anyway, they sound quite interesting but not a "nice" reverb.
kinda harsh and metallic. the decay was VERY VERY long.
--
Paul Winkler
"Welcome to Muppet Labs, where the future is made - today!"
I also have
some very odd impulses lying around somewhere if
I can find them - made by popping balloons in and around
some 20' by 5' steel tanks that were in the basement of
a converted industrial building I used to live in.
i'd love to have one of those. if nothing else, they'll provide
excellent study material for our intrepid voyagers into the
sonic unknown.
6:49 p.m.
On Tue, Oct 22, 2002 at 08:01:35 +0200, Tim Goetze wrote:
That would require deconvolving anyway, so you may as well got for the
chirp (swept sin), many epopel say there better for things like amps
anway, and you should suffer less from the DA converter.
- Steve
to do it. How
do you generate the impulse? (I've tried simply creating
a sample with 1 frame at maximum followed by zeroes, but this seems
not to make it through the D/A conversion - it comes out as dead
silence.)
maybe widening the pulse will help. steve? i've yet to take one
7853
days inactive
7863
days old
linux-audio-dev@lists.linuxaudio.org
131 comments
10 participants
participants (10)
-
David Gerard Matthews -
Frank van de Pol -
Lamar Owen -
Lea Anthony -
Mark Knecht -
Paul Davis -
Paul Winkler -
Steve Harris -
Taybin Rutkin -
Tim Goetze