10 Jan
2017
10 Jan
'17
2:02 p.m.
Hi,
in this German document [1] Sengpil [2] added a note for
balanced out ---> unbalanced in
for electronically balanced out without galvanic isolation to ground. He
mentioned to connect the + signal _and_ ground (not _with_ ground,
just to prevent against a misunderstanding), but the - signal should be
unconnected.
IOW
balanced out + -------------> unbalanced in
- NC
0 ------------->
There's a lot of chaotic half-truth about this topic in forums.
Considering this document, it makes sense why not all
balanced <--> unbalanced connections require the same kind of cable and
especially why I didn't notice an issue for balanced <--> unbalanced
connections before I got the Focusrite. However, the headphone amps of
the Focusrite are still crap and apart from connecting the Focusrite
blanced ---> balanced to itself, I even didn't ensure that the
balanced ---> unbalanced issue mentioned by Fons is the culprit, but it
could be the reason for the bad sound.
It might be useful for others, too.
[1] http://www.sengpielaudio.com/Geraeteverbindungen.pdf
[2] http://www.sengpielaudio.com/Searchengine.htm
Regards,
Ralf
10 Jan
10 Jan
3:34 p.m.
On Tue, January 10, 2017 7:02 am, Ralf Mardorf wrote:
in this German document [1] Sengpil [2] added a note
for
balanced out ---> unbalanced in
for electronically balanced out without galvanic isolation to ground. He
mentioned to connect the + signal _and_ ground (not _with_ ground,
just to prevent against a misunderstanding), but the - signal should be
unconnected.
There are multiple circuits commonly used for symmetrical outputs, some
need the "-" signal left unconnected, some need the "-" signal
connected
to the circuit ground, but the optimal location could be at the receive
end, or could be directly at the output connector.
The Sengpil approach would be required for circuits which just drive both
legs by similar op-amp output circuits with signals of opposing phase.
Some circuits attempt to mimic the behavior of a transformer by
cross-coupled feedback paths, such that you can terminate one leg to
circuit ground and the other leg should increase amplitude to keep the
total amplitude the same.
This paper presented at an Audio Engineering Society conference about 16
or 17 years ago describes some of those circuits and various problems:
http://www.thatcorp.com/datashts/AES5152_Improved_Balanced_Output_Driver.pdf
--
Chris Caudle
4:54 p.m.
Hi Chris,
On Tue, 10 Jan 2017 08:33:04 -0600, Chris Caudle wrote:
There are multiple circuits commonly used for
symmetrical outputs, some
need the "-" signal left unconnected, some need the "-" signal
connected to the circuit ground, but the optimal location could be at
the receive end, or could be directly at the output connector.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
of the cable output, not the device output?!
I don't have time to test it now [1], but soon or later I'll compare
Tip-Ring-Sleeve jack with Ring unconnected to TS jack or cinch jack.
Tip-Ring-Sleeve jack with Ring connected to TS jack or cinch jack and
with - signal and ground short circuit at the TS jack or cinch jack end.
However, likely connecting Ring with Sleeve by plugging in a TS jack
into the Tip-Ring-Sleeve output socket, IOW short circuit at the
output device is an issue.
I'm sceptic that it makes a difference to connect the - signal with
ground at the receiving device.
Regards,
Ralf
[1]
I've got all required jacks, but only 2 meter headphone cable, no other
cable and I don't want to unsolder jacks from balanced cables I'm
using. I've got TRS socket to TS jack adapters, but unfortunately those
short circuit T with R and not R with S. I don't know what those
adapters are good for. Perhaps for usage on both ends of a balanced
cable, to make it an unbalanced cable by using both wires instead of
just one wire or for converting a stereo signal to mono, by assuming
that all gear usually is short-circuit-proof.
6:06 p.m.
On Tue, Jan 10, 2017 at 04:54:24PM +0100, Ralf Mardorf wrote:
I'm sceptic that it makes a difference to connect
the - signal with
ground at the receiving device.
Depends on the type of output. If it is a transformer or a circuit
that controls the voltage difference between the + and - signals
rather than each of them separately, then connecting - to ground
at the input side would compensate for a ground voltage difference,
while connecting it at the output wouldn't.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
7:16 p.m.
On Tue, 2017-01-10 at 17:06 +0000, Fons Adriaensen wrote:
On Tue, Jan 10, 2017 at 04:54:24PM +0100, Ralf Mardorf
wrote:
Hi,
I suspect a misunderstanding.
We are never talking about
unbalanced out --> balanced in
we are only talking about
balanced out --> unbalanced in
The difference would be
balanced out unbalanced in
+ ─────
- ┐
0 ┴────
to
balanced out unbalanced in
+ ─────
- ────┐
0 ────┴
IMO it doesn't make a difference.
Regards,
Ralf
I'm sceptic that it makes a difference to
connect the - signal with
ground at the receiving device.
Depends on the type of output. If it is a transformer or a circuit
that controls the voltage difference between the + and - signals
rather than each of them separately, then connecting - to ground
at the input side would compensate for a ground voltage difference,
while connecting it at the output wouldn't.
9:16 p.m.
On Tue, January 10, 2017 12:15 pm, Ralf Mardorf wrote:
We are never talking about
unbalanced out --> balanced in
we are only talking about
balanced out --> unbalanced in
Correct, NOT unbalanced out to balanced in, only balanced out to
unbalanced in.
(whether connection between - and GND is made close to ouput, or close to
next device input)
IMO it doesn't make a difference.
For noise reduction it can make a difference when cables are long, the
connection difference changes whether the voltage difference between the
two devices due to power line leakage currents is higher at the unbalanced
input, or higher at the balanced output. For some types of balanced
outputs which can float effectively on the common mode voltage the power
line noise is lower with that connection.
The other way that connection point makes a difference is that connecting
at the input of the next device changes the capacitance which loads the
grounded output. That could affect stability in some cases.
By the way, what was the output level from the device with bad sound when
connected to the unbalanced input? Section 1.2 of the paper I referenced
earlier points out that when grounding one side of that style of output
circuitry, the ungrounded leg will be driven +6dB to keep the nominal
output level the same, which is twice the output voltage. If that drives
the output into clipping, large currents can flow through the grounded
side. If the output was near the listed device maximum it could have been
clipping just slightly on peaks. I would not expect that, usually an
unbalanced input is designed for lower limits, so I would expect the
unbalanced input to clip before the balanced output even in single ended
connection, but hard to know without checking.
--
Chris Caudle
10:20 p.m.
On Tue, Jan 10, 2017 at 07:15:51PM +0100, Ralf Mardorf wrote:
The difference would be
balanced out unbalanced in
+ ─────
- ┐
0 ┴────
to
balanced out unbalanced in
+ ─────
- ────┐
0 ────┴
IMO it doesn't make a difference.
Consider that the screen is not a perfect (zero
resistance) conductor and that there can be a
voltage difference 'E' between the two ends
(typically mains hum).
Then compare
balanced out unbalanced in
A ─────
B ┐
0 ┴──── E
signal at input = A - B - E
to
balanced out unbalanced in
A ─────
B ────┐
0 ────┴ E
signal at input = A - B
assuming the output amplifier controls A - B and
not each of them separately w.r.t. it local ground.
For example a transformer output will do that, but
it can also be done without a transformer.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
11 Jan
11 Jan
12:27 a.m.
Chris, Fons,
ok, this could be an issue. A while back I mentioned that I never build
usable audio circuits on a stripboard, while the same circuits were
usable on etched circuit boards. The route of short conducting
paths and especially grounding paths already could make a difference.
Cable length here usually is between 1.5 m and 3.0 m. Maybe it makes a
difference were connecting "-" to ground. I'm still sceptic, that it
does cause the issues for the Focusrite. It at least seems not to be
impossible. The problem is, that nothing in the user manual gives a
hint, to avoid plugging in TS jacks into the balanced outputs.
Regards,
Ralf
PS:
When I worked for Brauner, I soldered some of the Klaus Heyne Edition,
my claim is that all Klaus Heyne changes are fraud. Reminds me of
freeze-dried, neutral point earthing, peas and power strips
http://www.vovox.com/en/home-audio-hifi/products/power-supply/
"Yes, power cords too, have an influence on sound quality. This is hard
to believe and even harder to explain. But it nevertheless cannot be
denied, once you have tried it. Often it is precisely this cable that
transforms a good audio chain into a peak unit . A truly rewarding way
to perfect high-end systems." -
http://www.vovox.com/en/home-audio-hifi/products/power-supply/textura-power/
https://www.thomann.de/gb/vovox_initio_power_100.htm
https://www.thomann.de/gb/glockenklang_high_end_powercable.htm?ref=prod_rel…
10 Jan
10 Jan
6:01 p.m.
On Tue, Jan 10, 2017 at 08:33:04AM -0600, Chris Caudle wrote:
There are multiple circuits commonly used for
symmetrical outputs, some
need the "-" signal left unconnected, some need the "-" signal
connected
to the circuit ground,
and some don't care. The Sengspiel document simplifies things a bit
too much by just considering one unspecified 'E' type.
Ciao,
--
FA
A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)
2674
days inactive
2674
days old
linux-audio-user@lists.linuxaudio.org
8 comments
3 participants
participants (3)
-
Chris Caudle -
Fons Adriaensen -
Ralf Mardorf