When using delays with lots of feedback, I noticed there was a tendency for the signal to gravitate upwards, away from the 0 line when using any of the four audio inputs. I assumed there must be a DC bias of some kind.
So to find it, I tried using a sine osc unit and getting audio input 1 to control the V/Oct of that unit. This gave me a shift pitch up by around 39 cents. So I reckon that is around 32mV (1000mV/12 * 0.39). A fixed HP filter at 0Hz after the inpit, solves the problem but i wondered if this could be built in digitally to the 4 audio inputs. Input 3 and 4 had slightly lower offsets.
Is anyone else getting this? As I guess it could be my case.
I just set up the patch you described (sine osc with IN 1/2/3/4 connected to v/oct in) and can confirm a DC offset of 34-35 cents for each input with nothing connected.
I personally like that all of the inputs are DC coupled. Perhaps there could be an auto-calibration routine in the Admin → System Settings that you could run with nothing plugged into any of the jacks, and it could zero out the DC bias on the inputs?
1 Like
I had thought the audio ins weren’t dc coupled but clearly I was wrong. The routine sounds like a good one, if it is possible to implement. On the other hand having to use a high pass filter is not the worst thing.
Oddly, I don’t get a dc offset on the abcd inputs.
The small DC offset on the IN1-4 inputs is due to the additional analog anti-aliasing filtering on those inputs.
Thanks Brian, good to know it isn’t a fault.
How difficult would it be to add in either a digital calibration option or auto filter, maybe add it to admin pages etc, in a future release?
How is this better than just putting the HPF in front of your delay network?
Just from a workflow perspective, I would want to have this set once and hidden. It’s not a major thing of course but when just sampling even it is good to know you have an exact a replication of the audio presented as possible. With that in mind is an offset a more accurate choice than a hp filter? Or is the hp filter at 13.75hz so shallow and low as to be having a zero audible effect?
Unfortunately, auto-calibration is complicated by the fact that the DC offset of the anti-aliasing filter is probably temperature-dependent.
You can set the cutoff of your Fixed HPF as low as as you like.
Honestly, I think you are optimizing something that is not worth optimizing. This DC offset is tiny and inaudible You only came across this problem when you were playing with high feedback delay networks in which case you would still need the HPF. Incorporating DC-blocking HPFs in a delay network is not something to be hidden and forgotten but rather should be part of the patch design.
3 Likes
Ok, fair enough.
Would it be worth adding a switch to enable the built in delays to feature a dc blocking filter in their feedback path then? Maybe you don’t need this if the signal is high passed before entering the delay though.