supernatural filters 2pole clip

[gentleclockdivider]

I posted the same question over at gearslutz , no luck yet ..

Just asking if someone could verify this on his/her integra , jdxa , fa or jup 80 ..as these all have the supernatural synth engine .

Initialize a synth patch ( sns NOT pcm .
osc saw A ( mute others
Choose lp 3 , set filters slope to 12 db ( 2pole )
cutof at max
resonance at max ..127
I get a verry unpleasant distorted digital clipping sound when playing lower registers .except for notes E and G .
In the soundfile I will play note c , gradually increasing the cut off untill the clipping occous .


Recorded over usb , setting integra to 44.1 , 48 or 96 khz .the distortion remains.
Could someone test if the same behaviour occurs on his/her machine
Soundfile
https://app.box.com/s/m366sl3oin0o6ukrzbgb4dsz46ckci5d

[EvilDragon]

It's not unlikely that it does happen. Ye olde JD-990 also had an issue with filter clipping at resonances above 40 and you play a chord. Later iterations (XV-5080, say) had oscillator gain parameter which combated this issue successfully. So see if SNS has the same parameter and drop the oscillator gain before filter.

[gentleclockdivider]

It's not unlikely that it does happen. Ye olde JD-990 also had an issue with filter clipping at resonances above 40 and you play a chord. Later iterations (XV-5080, say) had oscillator gain parameter which combated this issue successfully. So see if SNS has the same parameter and drop the oscillator gain before filter.

Yes , but I don't think the sns engine has anythig to do with the jd990 pcm engine .
Here it sounds downright ugly when playing one note , clipping aliasing like
I am aware of the clipping issues in the jd990 and this was resolved with a partial gain parameter in the jv/xv series .
In the the supernatural synth engine there is NO pre-filter partial gain parameter ( when not using the pcm waves )

[JCJR]

I didn't try with my FA, but setting resonance to max is like setting REALLY high gain at the filter resonant frequency.

If a filter can self-oscillate, maybe one could call it "infinite gain" at the resonant frequency. A "well behaved" digital analog emulation might implement "infinite gain followed by soft-clipping" to better emulate saturated analog OTA chips, which can get "mellow sine-wavey" when heavy-overdriven in that situation. OTOH for instance some of the Moog transistor ladder filters didn't get "nice and mellow" driven hard-- They could get rough as a cob like a cheap overdriven transistor fuzzface distortion box. Some people liked it thataway. Not me so much, but no accounting for taste.

So now if you tune the filter wide open, "by definition" with a digital filter you are probably applying VERY HIGH GAIN at or near the nyquist frequency. Applying incredibly high gain to whatever dither noise is in the filter input signal and whatever dither noise is intentionally injected inside the filter algorithm to keep it well-behaved numerically. It wouldn't be so surprising that this might alias rather badly.

Just my off-hand ignorant guessing-- Maybe a smart person would expect something else to happen under those conditions.

[gentleclockdivider]

It's already been confirmed by another jdxa user .
https://www.gearslutz.com/board/electro ... p-bug.html
This issue ( 12db filters) occurs in all Rol.synth that have th sns engine
-jdxa
-integra
-jup 80
-jdxi
-fa