[sdiy] Untangling the Oberheim OB-8 filter

[Tom Wiltshire]

Hi Andrew,

> Oh, ok, I get it now. What it appears they are doing (which Tom did
> state in the first email but I just didn't twig that anyone would
> bother making it this complicated) is switching the entire circuit
> between being a 4 pole cascade and a 2 pole SVF - crazy complicated
> stuff!

Isn't it just! It's the only thing I know that's anything like it. Even the Elka Synthex only does a simple switch between HP and LP on a couple of stages. Craig Anderton's MIF does the same thing, but on all four stages, and throws a few extra bells and whistles in (selection of different caps, Allpass, series/parallel switching) but it's still basically a straightforward cascade filter.
But this Oberheim filter changes the entire topology...

> I would have just grabbed either the 2nd or 4th output of the
> cascade and called it quits right there.

Yep, that's the obvious way to go. I think what was going on was that Oberheim had become so associated with the SVF "sound" that they didn't feel they could leave it out. A 2-pole filter would have been close, but all their earlier synths had SVFs. They didn't abandon it until the pole-mixing filter in the Xpander/Matrix-12, and that was such a impressive beast, I doubt anyone cared much it wasn't an SVF.

> Ok, so with that in mind, they lift the ground on the first cap to
> bypass it completely, and then skip the last stages output, so now
> you've got only two caps doing something. You then need to block the
> feedback paths of those two caps to make them into integrators, and
> then add external "in line back to back" diodes to increase the
> damping at high signal levels to control the resonance.
> 
> If you check between the OB-Xa SVF
> http://electricdruid.net/wp-content/uploads/2015/06/OBXa2PoleCEM3320Filter.gif
> and the and the OB-Xa Cascade
> http://electricdruid.net/wp-content/uploads/2015/06/OBXa4PoleCEM3320Filter.gif
> 
> you can see that both circuits are there and there is a whole bunch of
> switching to have both of them together.
> 
> It took me a while, but now hopefully we're on the same page which bit
> of the full circuit is causing trouble?

Well, the 4053 on the left connected to pin 2 of the CEM. In 4-pole mode, that connects from it's common pin 15 to connection "Y". From there, there's a bias network of a diode and 240K to -15V (I think that's what that's for, right?) and then a 100K that connects to pin 6. That's all ok, since it follows the 4-pole schematic you linked above. But what I'd like to know is "Can I ignore the resonance limiting network on the far side of the 1uF cap in this case?" It's still there, and it's connected in some way.

I think it's probably better if I try and get my diagrams tidied up a bit and then I'll post them and you can tell me if it looks reasonable.

Thanks,
Tom