Official 33609 builder's tread. See 1st page for updates.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Started this project years ago and so many things got in the way but finally it's done. To the best of my knowledge... Thank you all and specially Igor (wherever you may be).
 

Attachments

  • IMG_20180520_230524.jpg
    IMG_20180520_230524.jpg
    1.5 MB · Views: 130
Maybe the experts could chime in about the LF distortion? Any updates on that issue? It's very subtle in my unit when pushed, although I tried to match diodes carefully.
Or maybe live with it... It sounds beautifully just a is...
 
Did you ever try to compress 40 Hz with a short release time?
The DIY version distorts as much as the original Neve version.
I took some measurements some time ago:
https://groupdiy.com/index.php?topic=26218.msg822395#msg822395
(Of course it is very important to match the diodes, but the distortion is caused by a 'ripple' on the control voltage.)
 
Ruud,

I've checked your 33609 samle and the distorion is rather horrible. The first part of the wave file - is it true bypass or "hot" bypass? Does the signal go through the circuit?

I've built the 2254 which has a very common sidechain circuit to 33609 and I don't see the THD as in your example. Mine is around 1% at worst. But I can't get as clean a signal as your bypass example. Hence the first question.
 
The first part of the sample is an original Neve 33609 with bypass on the compressor itself.
(Meaning that the signal is passing the electronics but the sidechain/control voltage is disabled.)
 
RuudNL said:
The first part of the sample is an original Neve 33609 with bypass on the compressor itself.
(Meaning that the signal is passing the electronics but the sidechain/control voltage is disabled.)
BTW, did you manage a deal with AMS/Neve, for the implementation of your circuit? I'd be curious to see what magic trick you used.  ;D
 
RuudNL said:
I'm serious.
The comment in answer #957 was just stupid. Ignoring the effects of CV ripple and confusing with CV feedthrough is pure senseless pontification.
There's been many attempts to solve this issue, the most common being the dual time-constant networks, or more complex solutions, such as the one based on cos²+sin²=1, that nobody managed to implement practically, and the addition of a hold circuit.
I suspect your solution was more or less based on the latter.
Can you divulge a bit or is it under non-disclosure?
 
Same problem with distortion at LF.... :(
help please...

Two questions:

4.3. APPLY MAX. OUTPUT FROM SOUNDCARD TO INPUT, AT LEAST +10DBU,
LIM IN, ATT FAST, RECOVERY 50, LIM THRESHOLD TO +4, COMP OFF,
ADJUST RV4 TO +4 AT OUTPUT.

The max I can Archivate is +6 Dbu at the end of the trimmer
I must change the trimmer or some resistor value?


Thanks!
 
ilfungo said:
Same problem with distortion at LF.... :(
help please...

Two questions:

4.3. APPLY MAX. OUTPUT FROM SOUNDCARD TO INPUT, AT LEAST +10DBU,
LIM IN, ATT FAST, RECOVERY 50, LIM THRESHOLD TO +4, COMP OFF,
ADJUST RV4 TO +4 AT OUTPUT.

The max I can Archivate is +6 Dbu at the end of the trimmer
I must change the trimmer or some resistor value?


Thanks!
Can you achieve +4dBu? If yes, you don't really need to change anything.
 
Here is a hint:
The distortion at low frequencies is caused by modulation of the control voltage.
This can be proven by applying a constant voltage instead of the CV and testing with a low frequency.
Distortion will be very low when the diodes in the attenuator are properly matched.
At very low frequencies, the release already sets in during the time that the waveform goes back to the zero crossing point.
This will make a sinewave look more or less like a 'sawtooth'.
There are a couple of ways to reduce this effect.

- You could avoid that the release sets in before a zero crossing.
- You could incorporate a short 'hold' time after a new peak. (Longer than the time of 1 period of the lowest frequency.)
- The word 'hysteresis' comes to mind...

Also a composite release network will reduce distortion. (Fast release on isolated peaks, longer release after a constant gain reduction.) Or even better: frequency dependent release!  ;)
 
RuudNL said:
The distortion at low frequencies is caused by modulation of the control voltage.
Indeed.

- You could avoid that the release sets in before a zero crossing.
- You could incorporate a short 'hold' time after a new peak. (Longer than the time of 1 period of the lowest frequency.)
- The word 'hysteresis' comes to mind...
Believe me, I'm very familiar with this.
None of these options seem easy to incorporate in a 33609 without significant circuit change/add-on.
 
Believe me, the solution is surprisingly simple!
It is only necessary to add a couple of cheap components. (To the compressor and the limiter section.)
But it reduces the LF distortion with a factor 10(!) at fast release settings.

(Unfortunately I can't tell much more, because I have a non disclosure agreement.)
 
RuudNL said:
Believe me, the solution is surprisingly simple!
It is only necessary to add a couple of cheap components. (To the compressor and the limiter section.)
But it reduces the LF distortion with a factor 10(!) at fast release settings.

(Unfortunately I can't tell much more, because I have a non disclosure agreement.)
nOK, I'll give it some thought. So I conclude the deal went good with AMS/Neve...?
 
Back
Top