[sdiy] OTA performance (was SSM chip reissue)

[Neil Johnson]

Mattias Rickardsson wrote:
> Interesting. I've had some filters sometimes lacking resonance in the
> low end, and the increaswingly unmatched 2164s could be a reason. But
> wouldn't the capacitors need to be matched just as good? AFAIK only
> the RC constant determine the pole frequencies, so if one of the caps
> are 5 % off, then the filter could perform worse if the gain cells are
> perfectly matched - compared to having them 5 % off in the other
> direction.
>
> How well do you specify the tolerance of your SVF capacitors? :-)

A 5% tolerance is about 0.4 dB. At high filter frequencies, where the
gain of the 2164 cells is close to 0 dB, the error is less than 0.1 dB
if you believe the datasheet.  At 40 dB down the errors are creeping
up (0.24dB is about 3%), and down at the low frequency end of things
the gain cells are probably quite a way off (lets say 60dB attenuation
for a 1000:1 range).  And as George Hearn found out in his initial
experiments with the Irwin 2164 linearisation circuit the gain
response at very low gain (high attenuation) is far from ideal:

George wrote:
> There is also a not so nice side effect that the gain response becomes
> non-linear below a certain level. In most applications this is not
> significant as it is at very low gain.

Nice to have some meaty technical discussions here again!

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