[sdiy] MS-20 Filter Clone Queries

[Aaron Lanterman]

On Dec 2, 2008, at 5:09 PM, Kyle Stephens wrote:

> I'm building an MS-20 filter clone, as seen here:
>
> http://electro-music.com/forum/phpbb-files/ms20_13700version_142.gif
>
> In regards to the cuttoff CV expo converter, I haven't any matched  
> PNPs lying around but I do have a few LM394s (matched NPNs) I got  
> cheap from Futurelec. There's some simple rearranging I can do to  
> make the NPNs work, right? I've seen dual matched NPN expo  
> converters elsewhere, but I'm not sure how to apply them to this  
> instance...

PNPs are natural for building a current source, NPNs are natural for  
building a current sink. 3080s and 17000s expect a control current to  
put into them, which is why you see PNP pairs there. A roll-your-own 4- 
BJT OTA expects a control current to be sucked out of it, so in places  
you see that - like the SSM2040 clones and the Buchla VCO triangle  
cores - you will see NPNs in their converters.

I suppose you could use a current mirror to spin around the current  
from a PNP expo pair.

> Also, this schem says it runs on +/-12V, while I'd been planing on a  
> +/-15V standard for my system. In general, is current consumption  
> the main issue for running at 12V? What needs to be changed if I  
> want to run the filter at 15V?


Only thing I think you'd really need to change is R6. U1b is keeping  
the top of R6 at virtual ground, so the -12 and ground through the  
470K resistor create a reference current. If you change to a -15, I'd  
change the 470K to something else that maintains more or less the same  
current. Or... it might not actually matter much - it would just shift  
the range of cutoffs, and you might like where it winds up.

> I like the Doepfer's resonance loop send among other things in that  
> manifestation (http://www.doepfer.de/A106.htm), and want to try out  
> various combinations for the res diode clippers with selector  
> switches, like different, colored LEDs, Si and Ge diodes, FETs wired  
> as diodes (I got the idea from fuzzbox techniques).
>
> Can you drive the filter's input without blowing anything up, and if  
> so, how..?

As far as the audio goes, the most that will happen (pretty much) is  
that you can saturate the OTA inputs. They have a tanh style  
nonlinearity that starts to get its business on in the millivolt  
range, which is why you see the massive 220R/47K divide down ladder in  
front. You might like it. Folks like Antti have spent a lot of time  
thinking about how tanh nonlinearities can be cool.

One thing to watch when building is the control current input pin on  
the OTAs. Accidentally short that to one of the rails, or any voltage  
source, really, and its fried.

My students have killed countless 13700s.

- Aaron