[sdiy] Compensation for triangle amplitude getting bigger at higher frequencies?

Roman Sowa modular at go2.pl
Tue Sep 5 15:19:52 CEST 2023


What are the odds... I just had a visit from my friend who brought an 
oscillator looking exactly like this, with big tuning problems. Triangle 
core, OTA, TL074 as Schmitt. So I had a chance to confront my mumblings 
with the real world.
The triangle amplitude started to rise at about 600Hz, and climbed more 
than 1V above rails at 10kHz. Integrator starts to change direction 
about 2-3us after the Schmitt has toggled so there's some lag in OTA 
itself. But the Schmitt trigger is a real bugger, at high setting it 
starts to move only when its input was overdriven by 400mV. That creates 
most of the amplitude rise and frequency going flat big time.
I tried to add that feedback cap on the Schmitt, 33-100p works best, but 
the improvement was just by one semitone. So the only option was to use 
Rossum compensation and now it's somewhat OK in most used frequency range.

Roman

W dniu 2023-09-05 o 11:27, Roman Sowa pisze:
> I suspect you have taken those rise/fall times from datasheet and not 
> actually measured them.
> TL07x may indeed swipe its output so fast, but only in certain 
> conditions, when the input is overdriven and output unloaded. If it 
> works as a comparator, it will think for a while what to do in this 
> unfortunate situation of passing threshold, then slowly deciding that 
> maybe it should move its output a litlle up or down.
> If you add a small capacitor in the feedback of the Shmitt trigger 
> opamp, it might help.
> There might also be some parasitic capacitance at OTA inpt, so even if 
> Schmitt changes immediately, the OTA will still push the current in 
> previous direction for a short while. That's easy to check if you look 
> at Schmitt waveform superimposed on integrator waveform
> 
> Roman
> 
> W dniu 2023-08-29 o 15:38, Scott Bernardi via Synth-diy pisze:
>> I've breadboarded a VCO with an OTA (as switch) - Integrator - Schmitt 
>> trigger triangle / square wave generator at its core. I've noticed 
>> that at higher frequencies, the amplitude of the triangle wave gets 
>> larger. Like the output of the schmitt trigger is delayed in 
>> switching.  I'm using parts of a TL074 in the integrator and schmitt 
>> trigger.
>> Is this because the rise and fall times of the TL074 become a more 
>> appreciable part of the waveform time?  TL074 has 13 V/uSec slew rate, 
>> which traverses the ~22v in about 1.7uSec, or the +/-5v trigger points 
>> in a bit under 1uSec.  Even at a top frequency of 16kHz, that's about 
>> 2% of the total period.
>> Is there a way to compensate for this?
>>
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
>> http://synth-diy.org/mailman/listinfo/synth-diy
>> Selling or trading? Use marketplace at synth-diy.org
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> Selling or trading? Use marketplace at synth-diy.org


More information about the Synth-diy mailing list