Anushri: VCA bleed?

Is it normal (or possible with combination of synth settings) that the VCA doesn’t entirely close?

I’d noticed it on mine and just assumed it was a quirk of design (seen it a few times in modular world where VCAs/VCFs don’t totally close). The guy who bought my Anushri has noticed it and I’m now wondering if there was something wrong with my build?

What kind of sound bleeds through the VCA? The filtered oscillators signal, or just the raw DCO signal (in which case it’s not a problem of bleed and it looks like board noise)?

A recording and numbers would help.

Thanks Olivier.

I’m posting this to try to help the guy who bought it (and I’ve sent him a link to the discussion).

I suspect it could be that the VCA Env control is in ‘gate mode’ as he says the last note played can still be heard in the background at a very low level.

I’ve told him to try playing with the Env setting to see if it persists.

i’ve noticed something like this on my anushri with the volume knob on 12 o’clock… it’s not there with the volume full open.

Hi, I’m the guilty party who bought the Anushri from Neil (to whom thanks for the on-going support). I’ve been struggling to make a meaningful recording. The bleed signal seems to be from both the VCO & the DCO, so the pitch of the bleed signal will change when the VCO range is changed and when the sub level is raised then the difference between the sub o/c settings and DCO can be heard as well. Changes to the VCA Env Control do not change the bleed signal level. It is also still there at different volume settings.
Thanks for your help! Apart from this minor issue I think its a great piece of kit and i’m looking forward to spending a lot more time with it.

Could you post a recording (wav file, release time set to 0, a note and then 10s of blank after it)?

@lux: sorry, didn’t know your user name on here :slight_smile:

I’ll keep watching this thread with interest.

This is why so many companies produce VA, real analog gets these little issues :slight_smile:

@6581: as I said in the first post, I’ve seen these issues with analogue modules in my eurorack modular. I guess when you’re attenuating a signal via analogue circuitry, the signal is never (theoretically) silenced, only reduced to a point where you can’t perceive it. Though I’m not speaking with much authority, only experience and common understanding.

OK, here’s an 11 sec recording. I didn’t leave the blank bit alone though, after a couple notes are triggered I first step through the sub oscs and DCO, then detune the VCO. As far as i can tell it sounds disproportionately louder in this snippet.

> I’ve seen these issues with analogue modules in my eurorack modular. I guess when you’re attenuating a signal via analogue circuitry, the signal is never (theoretically) silenced, only reduced to a point where you can’t perceive it

This happens with some exponential VCA designs, but the Anushri uses a linear VCA, and the output is really the input times the CV times a constant. A source of problems could be that the CV produced by the DAC does not exactly reach zero, but this usually causes only minor annoyances. On the SMR4 mkII I have added a resistor to fix that, on the Anushri I trust the MCP4822 to have a good zero.

Anyway, the audio sample doesn’t sound like something I have observed on my builds.

  • The background noise is not VCA bleed but mostly board noise (signals bleeding on adjacent PCB traces due to capacitive coupling) - the “noise floor” of the synth.
  • My measurements on my unit showed that the SNR was below 72dB - which is the threshold below which patching things make little sense for a DIY project. Here, it is at about 25dB. One thing I have noticed is that the envelope of your sound shows 16 clear steps (4 bits of dynamic) which means that 8 bits of DAC precision are lost (this would imply a drop of SNR by 48dB, which is close to what I measure). You must be pushing the gain on your soundcard to record this, don’t you? So I suspect that either the VCA DAC has been damaged, or more probably you intentionally play notes with a very low velocity. Is that the case?

And check that no device has sent a MIDI CC 7 (volume change) to your unit!

Would be interesting to know how that was recorded, the noise floor seems very high. When I said I’d noticed a small amount of bleed it was nothing like that kind of level, more of a barely perceivable tone. And I monitor a little too loudly most of the time.

Signal path for me: Anushri->Mackie Onyx->Focusrite Compounder (as a limiter)->Adam monitors.

When I exported the file from Logic I saved it as an 8bit file to keep the size down, otherwise it was 1.9MB and i couldn’t upload that here . I’ve uploaded the more representative 16 bit version to my soundcloud page as Anushri 2 http://soundcloud.com/lux-redux/anushri-2.
The recording was created by “knob jamming” as per the manual, not by playing the keyboard.

I still can’t tell whether something wrong is happening on your unit; or whether your sound is intentionally weak, causing the noise floor to be magnified.

Can you post a recording of another sound - say a standard saw tooth with sub/DCO off, and with filter in a middle position with a LFO cutoff modulation?

@ neilbaldwin - i just joined today to discuss this. I’ve lurked here for while after buying a Polivoks Shruthi from the MI shop. I can’t solder (tried and failed) and i’m not familiar with electronics.

I monitor quietly and this issue was noticed initially on headphones. My Signal path is Anushri>Mackie CR1604>M-audio BX5. Recording made via Venom Keyboard. Couldn’t swear to CC message, i unfamiliar with managing them closely. certainly i haven’t told any device to do anything

OK, I’ve tried to make the recording you’ve suggested pichenettes. The filter is apparent in the bleed signal here.

This doesn’t look normal, but since you’re not familiar electronics, there’s not much I can do to help you investigate. The check would consist in measuring the DAC VCA CV voltage. I believe your problem is caused by volume or velocity attenuation, causing the VCA to operate on only a fraction of its range - requiring you to amplify the signal A LOT and magnifying any bleed/background noise in the process.

Also a simple check would consist in swapping the two DACs to see if one of them hasn’t a zero level problem.

Olivier, what would the best way to get the healthiest signal possible out of the Anushri? It would be interesting to see if indeed the output volume is somehow being attenuated (as you say, MIDI CC 07 is possible but presumably this is not stored - or is it?)