Quad VCA CV issue

I seem to have an issue with my Quad VCA specifically when breaking the normalization between CV inputs.

Steps to reproduce:

  • OSC into IN1. (eg saw from Dixie).
  • Monitor OUT1 (or OUT4 doesn’t matter so long as you can hear the output).
  • Boost1 OFF (Boost doesn’t matter but just set off).
  • Level1 min (ie full CCW).
  • Curve1 Response exponential (ie full CCW. The issue is far less noticeable when using linear curve).
  • CV1 envelope (eg Triggered or cycling Env from Quadra etc).
  • Adjust CV1 Attenuator until OSC is just audible. (ie until you can just hear Saw).

Now patch a cable into CV2 with the other side unconnected.

Expected: Nothing should happen.
Actual: Volume of Saw just got quite a bit louder.

I have tested this on three different power supplies. Issue remains.
I have tested quite a few env’s and LFOs etc. Issue remains.
Notice that if one uses a buffered multiple before sending to Quad VCA CV and
repeating the above steps the issue no longer occurs.

Can anyone repeat this issue on their Quad VCA?

Thanks,
Steve

1 Like

hello

I recreated it with Tides and Mangroves.

The output is slightly louder when I patch a dummy cable into CV 2. It’s really not that much louder with my Quad VCA.

I have a Pittsburgh EP-360

@aroom Thanks for taking the time to try it out.

I just tried Tides too and the effect is still there. Maybe not as much as Quadra and other envelopes I have. I think it’s because Tides outputs 8V envelopes.

For me using Quadra as the envelope generator (with Quad VCA response set to exponential (full CCW) and the CV1 set to around 2-3 o’clock) the output of a Dixie Saw is approx -30dB (peak) but with a dummy cable patched into CV2 it jumps to -10dB! This is pretty far from subtle.

Using a buffered multiple inbetween Quadra and Quad VCA with the same settings described above the output remains approx -10dB (peak) regardless whether the dummy cable is inserted into CV2.

Steve

I had noticed the same behavior, this is because of the sharing of the enter Trig between 1, 2, 3 and 4 cv-inputs.
If you want to avoid this problem you can take the habit to enter your differents signals in the order 4,3,2,1 and then you cannot have sound accidents.

This would only avoid the problem if you patch linearly and never go back and change anything.

Patching a cable into CV2 should not effect the CV level of channel 1 (and so forth).

It would be nice to hear from someone at Intellijel regarding this.

Steve

I’m agree, it should not, but i think it’s a conceptual default of the module, i should be very surprise if Intellijel could do something for us. and you are right this is annoying.

This is what is happening:

Due to the normalling, a CV input at CV1 could possibly go to all four CV inputs (assuming nothing patched into the other three CV inputs). Effectively meaning the four input CV attenuators are in parallel.

Since each attenuator is 10k the equivalent parallel resistance is 2.5k. If the output impedance of your source is something like 1k (which it is on the Quadra) then the voltage divider formed with the 2.5k is larger than you might have expected.

This is made worst by the fact that small differences to the cv input in exponential mode yield large differences in gain based on how the exponential control works ( only 100mV result in a change of 3db).

What we should have done is use 100k attenuators instead of 10k and then the issue is negligible. On our next run of Quad VCA we will be doing this. We can probably mod your Quad VCA if it is giving you grief by replacing those four pots.

Other workarounds
-Buffer the envelope with a Buff Mult (which has a super low output impedence)
-Use the Quad VCA channels in backwards order: 4,3.2,1 so that when you only have one CV source you are only normalled to a single channel and not four.
-ease off on the Exponential mode! Honestly in most application you want linear or a touch of exponential but not full exponential.

1 Like

@Danjel Thanks for replying and explaining the issue.

Given there are no schematics to look at, can you please confirm that only the four (4) CV attenuation pots need to be changed to Linear 100k (B100k) to fix this issue? No other changes are required? Has someone at Intellijel done this fix and confirmed that it works?

Also, when a fix is implemented to newer Quad VCAs could it be made clear in some way? Eg labeling “Rev002.x” on the PCB etc I for one do not wish to purchase another Quad VCA with the current issue. It would be nice to have simple way to ask a retailer/seller “is it rev 2 or above” etc before purchasing and knowing without a doubt that it doesn’t have the issue.

Finally, in regards to the comment:

Hahahah!!! Sorry but this is absurd! You may as well ask that I don’t use more than 75% filter resonance or limit pulse width modulation to <10Hz etc etc It’s just not going to happen! (Besides, I found this VCA issue with response set at approx 12 - 1 o’clock - far closer to linear than exponential).

Thanks,
Steve

Yes we have confirmed this works. Just B100K for the CV attenuators.

We have gone a step further and changed the pcb layout so that there is an extra buffer directly before each attenuator. Essentially we have built a Buff Mult into the module so there will be zero voltage drops.

New pcb revision is rev2.03 to identify this.