Doepfer

Hi Ingo, Dieter, list,

After looking at the schematic Dieter kindly sent I believe the 
reason for the 'suddenness' of the resonance/self-oscillation is the 
type of filter itself: it's a standard Sallen-Key low pass topology, 
and I've read in several books that for normal use these are not 
regarded as very practical filters, due to the sensitivity of the 
resonance to the component values used, i.e. it is hard to guarantee 
any given value of resonance because small changes in component 
values give rise to large changes in the resonance. I came across 
this before in the work I've done to date on the Korg35-based 
MS10/MS20 filters (there's a little algebra on it in my report), as 
these are also Sallen-Key types, and as then, I'm finding it hard to 
come up with a good way of sensibly representing this problem, and 
how it might be affected and/or alleviated by different circuit 
topologies (and for example different pot laws). However I have 
breadboarded a very simple fixed-frequency Sallen-Key filter, and 
comparing experimental results with some simulations, I now have a 
greater trust in what the simulations are saying (which can be hard 
to interpret), and the upshot is I believe it should be possible to 
make the resonance come in in a more controlled fashion. This should 
be achievable by replacing the resonance pot P5 with a 'reverse log' 
type, law 'C' (keeping it at 50k); and another thing which would seem 
to be a further step in the right direction would be to connect the 
pot wiper to the end of the pot connected to R17 (=47k).

Unfortunately since I don't have one of these modules I cannot try 
it, and so cannot be sure for certain. However Dieter has said he 
will look back at the notes made during the module development to see 
if anything similar was tried then. The simulations show that one 
possible problem with doing this is that the peak in the resonance 
might occur for a lower knob setting: trying it out for real is going 
to be the only way to find this out, but even if it does, the 
solution should be quite simple - tweaking the value of R17 (although 
since we know the topology is very sensitive to small changes in 
component values, the required value might mean the use of a preset). 
Also, since I have only simulated the *filter* configuration of the A-
101-2, it is possible that the change has an undesirable effect on 
the VCA side of things (!?).

I've placed a small file in the 'Files' section showing some plots of 
the effect of changing the pot on my simple experiment. The schematic 
shows the Sallen-Key set-up: the key thing it shares with the A-101-2 
is the topology around the final op amp providing the gain; other 
component values were chosen for convenience (and note since I have 
equal value capacitors, the max gain for self-oscillation needs to be 
three, as opposed to 2 in the A-101-2, hence the different values 
there). The blue trace shows how the filter gain shoots up in about 
the last 5% of the pot travel for a linear pot, which effectively is 
the problem. The pink trace shows how the gain rises much more 
steadily throughout the pot rotation, using a law 'C' pot with the 
wiper connected to the one end as stated above. I should point out 
that the traces were, of necessity, obtained via slightly different 
means for the different pot types: using switches I rigged it so that 
as well as measuring the filter gain, I could simply switch the 
filter caps and resistors out and measure the gain of the non-
inverting op amp set-up. For the linear pot, having measured both 
gains, I inverted the theoretical relationship between the pot 
position and the op amp gain (simple, since the pot is linear), so 
this enabled me to plot the filter gain versus pot position - this 
gave a satisfyingly very close trace to that output from the 
simulation (in which it is easy to relate pot position to everything 
else), which is what convinced me that I could trust what the 
simulations are telling me. For the 'C' pot (robbed from my A-102 as 
I don't have a spare one lying around!), I have no way of telling 
exactly what the relationship between pot position and its resistance 
is (the log pots I use in my simulations are 'guesstimates' from 
datasheets), so I have nothing that I can invert, so the figures are 
literally taken from the physical settings I used (and hence why the 
graph is nowhere near as smooth).

I'm hoping that Dieter or someone can try this suggestion easily 
enough and let us know whether it is as useful as I'm hoping it is!

Tim

--- In Doepfer_a100@yahoogroups.com, <hardware@...> wrote:
>
> > Hi Ingo,
> > 
> > > Even though the
> > > filter itself is rather smooth sounding the self oscilation 
comes all
> > > of a sudden 
> > 
> > I'd be quite happy to cast an eye over the circuit to see if I 
can come 
> > up with a suggestion as to how to change this, but unfortunately 
I 
> > don't actually have this module and so do not have access to the 
> > schematics, so unless someone can send me a copy ... Dieter? 
Anyone?
> > 
> > Tim
> 
> Tim,
> 
> I'll send you the schematics.
> 
> Dieter
>