Cherry Audio Forums

Is the same math issue present in the SynthVoice filter?

I assume they share a lot of common code and it is not accident the first 2 instruments in VM are the first two VST.

To my ears CA2600 is not even in the same league as Korg Odyssey VST in terms of analogue modelling. CA2600 sounds years behind it in terms of analogue modelling quality of VSTs.

yes great question, it seems that the 2 pole filter is also not behaving as expected.

i was asking this myself, i can't answer myself..

Hi,
just tried build 86 and this is looking way better. Resonance has to be over 95% though to be comparable with the other 2600 synths but now they behave quite similar.

Cherry Dan wrote: ↑Thu Nov 05, 2020 12:35 amI can assure you that we are as interested in getting these details right as you are in analyzing what we've built, and I commend you for your dedication to this issue.

This has been mentioned before somewhere (I think here in the Cherry Audio forum):
You guys really need to improve your QC/QA with some good/dedicated beta-testers... which apparently, gentleclockdivider would be a good candidate and probably dergundel too

Cherry Dan wrote: ↑Thu Nov 05, 2020 12:35 am I can assure you that we are as interested in getting these details right as you are in analyzing what we've built,

This doesn't make sense , you're telling me no one from the developers team noticed the filter bug ( 2 in a row ) ?
Could you tell me which modelling technique you guys use , circuit modelled using splice ...then simplified to make it run real time ?or
Perhaps a black box approach , making it sound similar to the o.g. using non circuit modelled techniques ?
I honoustly don;t believe there is any actual modelling going on

Hi,

Be sure to allow it to update to the latest build.

Greg

Could you tell me what kind of modelling technique you guys use ?
Thank you
I updated the filter , the weird looking spike is gone , buyt I don't get why there there are partials 400hz below and abovethe cutoff point when self oscillatng .

Hi,

I'm not a developer and I don't know the details of the algorithmic secret sauce, but as you can see the updated filter is much improved.

Thanks for pointing out this issue!

Greg

I did some testing with filter self oscillation and as shown all three 2600 emulations behave quite similarly, the filter of all three synths pass a lot of signal below the cutoff frequency when resonance is turned to 100%.
Also I found that the filter is able to self oscillate without input, what you would expect from this kind of 24dB filter, very good.
As regards the filter, to me everything seems quite right. I used to own a real ARP 2600 some 10 years years ago and at that time I got the TimewARP2600 as a first emulation which I tested against the real machine. They actually were really very close.
But I think you will find a lot of differences anyway, even if you would model the circuits precisely - I don't know how WayOutWares did it.
Anyway, I find it amazing to have the panel almost as big as the real one and to be able to patch AND store and recall what was patched which couldn't be done on the real machine in those days.
Only thing maybe for the moment, the resonance slider could be more effective along it's path.
Hope this brings some new insight.