Triangle to sine using OTA question - Paul?

[gez]

Recently I’ve been trying to overdrive a OTA with a triangle wave to get a sine wave and have had mixed results.  Although all the schematics I’ve seen seem to use a 15V dual supply, dumbo here decided to run the OTA at 5V (more convenient to do so).  

The best I’ve been able to achieve is a rounding off of the triangle’s tips but nothing truly sinusoidal, there’s no gradual curving of the sides they’re as straight as a die.  Reducing the input bias current seems to help get a smoother curve, but it’s still not a true sine.

Is this normal, i.e. the best I’m going to get using this method, or is it that linearity of OTAs is poorer at higher supply voltages and therefore the results better?

[Paul Perry (Frostwave)]

I havn't tried roundig a triangle with a 3080 at low voltages, but maybe you need even MORE input!! As you increase input signal, eventually it should give a 'rounded square wave'. Also, for musical purposes, usually you don't need a perfect sine. I've been very suprised at some of the waveforms that sounded 'reasonably pure'
The other thing to try, is some 'post rounding' with back2back diodes.

[Nasse]

http://hammer.ampage.org/files/hypertriangleclock.gif

There was this interesting looking fet sine shaper. Anyone tried this with audio signal?

[gez]

I havn't tried roundig a triangle with a 3080 at low voltages, but maybe you need even MORE input!! As you increase input signal, eventually it should give a 'rounded square wave'. Also, for musical purposes, usually you don't need a perfect sine. I've been very suprised at some of the waveforms that sounded 'reasonably pure'
The other thing to try, is some 'post rounding' with back2back diodes.

Thanks for the feedback Paul.  I'm not sure that I need more input as it squares off pretty harshly with large input signals.  Seems like the 3080 is a little too sensitive (gain too high?) at lower voltages as I'm not getting a 'rounded' square, more of a triangle with, er 'roundy' points.

I agree that you don't need a perfect sine, I'm just experimenting at the moment and seeing what various approaches do.  I've had pretty good results using a 4049 for a triangle LFO then overdriving a 4th stage.  Again, this just rounds off the points but it's less mechanical than a pure triangle.

So far, the purest 'sine' I've got is from diode shaping networks.  I messed around early in the year with these and had mixed results, but I've finally got the current consumption down and a consistent wave (using a regulator).  

I'll mess around a bit more with the OTA, but I'll probably stick to the diodes.  Thanks again - at least I know what I'm looking out for now!

[gez]

http://hammer.ampage.org/files/hypertriangleclock.gif

There was this interesting looking fet sine shaper. Anyone tried this with audio signal?

I haven't tried this circuit but I've done the rectified sine thing with a diode shaping network.  I rigged it up to a simple T-filter for an auto-wah and it sounded superb!  

Wasn't too cool with a tremolo mind.  It worked fine but the tremolo was very subtle - sounded like the depth control wasn't cranked up enough even though it was.  

I'm just in the process of re-jigging the circuit, so I'll have to try it out with some other effects.