[sdiy] Another SSM2164 question

[Neil Johnson]

Hi Tom,

Tom wrote:
> Neil wrote:
> > Tom wrote:
> > > The quick and dirty way to compensate the resonance is simply to  
> > > feed (some of) the Freq CV to the resonance VCA. As Neil said, the  
> > > VCA can be considered to control  "damping", so as the frequency  
> > > climbs, the damping increases.
> >
> > I tried this.  Its a nasty hack and I didn't like the result.
> 
> It *is* a nasty hack, but what was the result?

While it almost worked, it was not perfect - there was still some 
clearly-perceptable variation in the self-oscillation frequency.

The additional trick was a couple of diodes to drop a bit of level, then 
a pot to fine-tune the compensation (I originally marked it as "Q 
COMP"). 

Q CV -----|>|--|>|--/\/\/\/\-----GND
                        T
                        +------> to Fc buffer/mixer

I used 1N4148 for the diodes, and I think a 10k pot for the
trimmer.  The diodes force the compensation to kick in at high Q levels.

Note that at the same time the overall behaviour of the filter at high Q 
was serious waveshape warping and distortion, giving an almost 
ring-modulator-like sound - one input being the source, the other being 
the filter 'oscillator'. Some may actually like that effect. 

> > One thing I noticed on my filter (may not apply to all filter  
> > designs) is that it looks like the limiting factor is the slew rate  
> > of the op-amps.<snip>
> > One thing to try would op-amps with faster slew rates (I'm using  
> > TL07x family).
> 
> Ok, easy enough to definitely be worth a shot!
> Ian Fritz thought the effect was due to op-amp phase shifts and  
> suggested (an electronotes solution) compensating these using small  
> caps (few pF) over the VCA V-to-I resistors.

I'm sure there will be this going as well, but at this stage I think the 
slew rate limit is the main candidate, based on observations of what's 
happening to the waveform on a scope. 

> > In the meantime, I've opted to limit the feedback amplitude to  
> > minimize this effect, while still allowing self-oscillation, using  
> > a scheme very similar to that used in the Oberheim OBx.
> 
> Is this why you have so much more going on around the feedback VCA  
> than Roman's design? There's a straightforward differential mixer  
> first up, followed by the VCA stage, but then there's that network of  
> 33K resistors and diodes as well.

Yes, very similar to what Oberheim did in the SEM and OBx (amongst others). 

> Presumably these provide gain-limited feedback. Is this close?

The opposite - being in parallel to the damping gain cell, it sets a 
lower limit on the damping (or, sets an upper limit on the Q) of the 
feedback path. The diodes ensure that it only has an effect at higher 
amplitudes, and keeps out of way during normal (non-self-resonating) 
filtering operation. 

> I've got a question about that though; I've used diodes in that  
> arrangement as a clipper for fuzzes, so why doesn't that happen here?

Because in a fuzz box you are hard-limiting the output of a buffer, 
flattening off the top/bottom of the waveform and introducing higher 
harmonics (fuzz). 

In this application the diodes are used to fix the lowest damping level 
during self-oscillation - as the amplitude increases so more current 
flows through the diodes, increasing the damping and this stabilizing the  
amplitude. 

> Good luck with your filter. I'll let you know if I find out anything  
> useful.

Thanks, and cheers. Now working on schematic entry and then a PCB. 
Thanks to an A3 laser printer from a local Freecycler (who, ironically 
is/was on the list as well - hi Phil!) I can now print out my schematics 
BIG! 

Neil
(likes his diagrams in A3)

--
http://www.njohnson.co.uk

-----------------------------------------
Email sent from www.virginmedia.com/email
Virus-checked using McAfee(R) Software and scanned for spam