[LAU] [Zynaddsubfx-user] yoshimi bug fixes

cal cal at graggrag.com
Sat Oct 24 16:32:37 EDT 2009


fons at kokkinizita.net wrote:
> On Sat, Oct 24, 2009 at 12:16:58PM +0100, Folderol wrote:
> 
>> Very interesting information. It's fascinating how complicated something
>> like a 'simple' metering system is!
> 
> Well, it's not really complicated...
> 
> 'Traditional' VU meters had no active electronics
> at all, just a bridge rectifier and a resistor. The  
> dynamic response was determined only by the mechanics
> of the meter. This means that a 'real' mechanical VU
> was not just a V or A meter with a different scale,
> it had to be engineered to have the right response.
> 
> Now such a moving coil meter is just a classic mass +
> spring + damping system, which corresponds to a second
> order low pass filter. So to get the correct response
> from a software VU you take the absolute value of the
> signal and apply the low pass filter. For a VU the
> filter was defined in the time domain: with a step
> input It should reach 99% of the final value in 300ms,
> overshoot to between 101% and 101.5%, and finally 
> come to rest at 100%. The overshoot means sub-critical
> damping, so the filter is 'ringing' (but not much).
> The -3dB frequency for a VU filter is around 2Hz.
> 
> The filter used in jkmeter is similar, but has exactly
> critical damping, it does not overshoot, and the -3dB
> frequency is around 1Hz. The other difference is that
> the input to the filter is not the absolute value of
> the signal but the square of it. This means you can't
> direclty compare the 'speed'. For short signals (less
> than 125 ms) the K meter is faster than a VU would be,
> for longer ones it is a bit slower, and it also falls
> back slower than a VU. If anyone is interested in the
> exact time and frequency domain curves just ask off-
> list.

I started off my working life with 17 years in radio broadcasting, so the
"traditional" vu meter and it's rigidly defined characteristics were very
much a staple of life. I think that experience still colours my thinking.

Just to drift back to the zyn/yoshi metering for a moment - with 0.031
I've gone back to the standard zyn characteristics, and just eliminated
the need to grab the master lock. I think it's better now.

Thank you Fons, your generosity in sharing your knowledge is very much
appreciated.

cheers, Cal




More information about the Linux-audio-user mailing list