Modeling a synths filter?
-
- KVRist
- Topic Starter
- 356 posts since 3 Oct, 2009
Anybody have a general explanation of the easiest, proper way to do this? I want to, one day, try my hand at modeling/simulating my AX-80...however, im about to sell it. So I want to capture some of the vital parts.
- u-he
- 28065 posts since 8 Aug, 2002 from Berlin
The best and easiest way includes verifying your model against the original circuit. I don't think modeling a synthesizer without actual access to it is the "proper" way to go about it
- KVRAF
- 7896 posts since 12 Feb, 2006 from Helsinki, Finland
Speaking of CEM3372 (which is what AX-80 has for a filter), what's the deal with the x100 impedance for the last pole? The rest of the circuit is really elegant (well, at least the method in the patents is), so it would seem reasonable that there was some actual reason for this?
edit: apparently CEM3396 also does this, but there it's only x70 ..
edit: apparently CEM3396 also does this, but there it's only x70 ..
- KVRAF
- 7896 posts since 12 Feb, 2006 from Helsinki, Finland
I actually finally figured out one possible explanation: if you want to use a simple unity gain output buffer, but still raise the signal amplitude by a factor of 100, you can do this by using a 1:100 divider for the last stage's feedback-side base. Since the control current remains the same, this should cut the frequency by a factor of 100 as well, which can be compensated by dividing the capacitor size by the same amount.
There is some biasing required to keep the base voltage below the collector (which now has 100 times the swing), but otherwise it doesn't seem like this would cause any obvious additional problems. I suppose should check with the hardware if there is some asymmetric clipping that would happen with the BC forward bias, but all my tools are packed up so I'm not going to bother, maybe some day.
There is some biasing required to keep the base voltage below the collector (which now has 100 times the swing), but otherwise it doesn't seem like this would cause any obvious additional problems. I suppose should check with the hardware if there is some asymmetric clipping that would happen with the BC forward bias, but all my tools are packed up so I'm not going to bother, maybe some day.
-
- KVRian
- 1097 posts since 28 May, 2010 from Finland
You could also perhaps do an impulse response filter. I.e. Acustica Nebula style. That you sample the gear at different filter settings and then your filter implementation is basically a convolution reverb that mixes between the different samples sampled at different settings. IMO the Acustica Nebula filters sound more analogue than many VSTs.
You can use the Nebula tools to do this and then you can distribute your filter as Nebula library. If it's not something that has demand, then this is probably safer, than putting in the development time.
You can use the Nebula tools to do this and then you can distribute your filter as Nebula library. If it's not something that has demand, then this is probably safer, than putting in the development time.