[sdiy] Instability in an OTA-C LP cell

[ASSI]

On Wednesday 14 July 2010, ASSI wrote:
> On Wednesday 14 July 2010, Olivier Gillet wrote:
> > Has anybody any idea about what's going on? Why would the fix
> > work (or appear to work)?

[Olivier sent me some more schematics with this topology and some more 
questions.]

> It's a minor mystery to me that this circuit works at all.
[...]

That mystery resolves when having had enough coffee and recognizing 
that the the opamp provides a feedback loop around the OTA.  I had 
incorrectly assumed it was just buffering the output, but instead it 
actually drives the OTA output current to zero and hence no extra DC 
path is required.  This will also take care of the amplification 
problem I had suspected (the opamp has open-loop gain into the 
feedback path with plenty of headroom to recover the missing 33dB).

The explanation for the lock-up is still the same: the circuit goes 
belly-up if either the output range or the common-mode input range of 
the opamp is exceeded, since that breaks the feedback loop and then 
the OTA output will saturate towards one of the supply rails.  So, why 
does the 47k resistor help?  The 47k together with the parasitic 
capacitance at the pin will provide another lowpass pole that perhaps 
produces just enough roll-off to keep the opamp from producing an 
overshoot that would otherwise lock it up.  You should be able to see 
this on the scope when driving the input with a square wave.  then 
again, there may be other explanations for this, but I'd be wary to 
rely on it functioning always.

In summary, the circuit as is likely works OK for lower input 
amplitude or at higher supply voltage.  The safer solution would be to 
use a rail-to-rail, non phase-reversing opamp.


Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Wavetables for the Terratec KOMPLEXER:
http://Synth.Stromeko.net/Downloads.html#KomplexerWaves