phase shift oscillators.

[Brian Marshall]

I was trying to use a 4 opamp lfo on my project i am working on, but cant seem to get rid of the 2nd harmonic.  it seems to be leaning to one side, no matter what i do.

a phase shift oscilator is looking a lot simpler, and a lot better.  i havent tried it yet, but i never hear about people hear using them.... am i missing something?

[Paul Perry (Frostwave)]

The problem with phase shift oscillators is that usually you have to vary at least two elements (hopefully resistors!). Sure there are limited range versions where you only vary one element, but these usually have some drawback.
Let me know if I'm wrong, though!

[Brian Marshall]

i think if you vary one of the resistors (to ground) you will get amplitued changes as well as frequency.  i seem to have the same problem with my 4 opamp oscilator, although not that noticeable.

the other thing i was thinking about is using 3 well matched transistors instead of resistors, and controling them with a voltage.  I am thinking this might not work though, because as the voltage drops and rises via oscilation, the neede voltage at the colector would have to change to keep the resistance constant.  Maybe this wouldnt be a problem.  i dont understand bipolars that well.  that's sort of why i havent tried it yet.

maybe opamps could work..... not sure about the opamps impedance to ground?  transconductance amplifiers?? maybe, i dunno.

[Paul Perry (Frostwave)]

I've always found a triangle generator, followed by an overdriven 3080 to convert to a sine, was "near enough for rock&roll".
But, yes, any phase shift circuit could be converted to a phase shift oscillator, sure enough. Just be caregful where you pick the signal off, because one point will be least distorted. Let us know how you go!

[Tim Escobedo]

What range of frequencies do you want? What oscillator circuit were you using? What waveform do you want?

[Brian Marshall]

What range of frequencies do you want? What oscillator circuit were you using? What waveform do you want?

i want a sine wave.... or close.    LFO type speed... .25hz to 2 hz or so,  maybe slower.

the circuit i was using was a 2 opamp triangle generator with one as an integrator, and the other as a comparator, but no matter what i do, it isn not symetrical... it rises faster than it falls.  maybe this is due to loading, but i also tried it with a 2nd integrator, and a 4th opamp to set offset, and de-amplify the signal.  (i only need about a 2 volt swing

Brian

[Tim Escobedo]

Try a circuit like this:
http://www.ozitronics.com/docs/k23.pdf
Diodes are used to soften the peaks of the tri.

BTW, how do you know you've got a 2nd harmonic? And is it bad enough that it really matters in whtever it's controlling?

[Brian Marshall]

well the first two opamps are basically set up exactly like that, and the triangle is not symetrical, so my "sine" wouldnt be either... right?

if i look at the square wave from the comparator, it is not symetrical either, it's tops and bottoms lean back towards 0v beforefinally dropping.

oh, and yeah, it probably isnt a 2nd harmonic, just that non-symetry makes me think 2nd harmonic.  I'll have another look at it when i get my scope, right now i am using a coputer, so not too reliable.  I use my other computer, which does not have internet access, so it is a pain in the ass to get a pic up.

[Tim Escobedo]

Using a soundcard in a computer as the front end of a oscilloscope may not faithfully reproduce the such low frequency signals. I'd guess the inputs are AC coupled. And freqs. of .25hz to 2 hz are damn near DC.

Unless you absolutely need a pure sine, a little bit of distortion of the wave isn't necessarily a bad thing. Especially with a LFO, it may or may not be very noticeable at all.

[gez]

the circuit i was using was a 2 opamp triangle generator with one as an integrator, and the other as a comparator, but no matter what i do, it isn not symetrical... it rises faster than it falls.  maybe this is due to loading, but i also tried it with a 2nd integrator,

This is easy enough to rectify.  It's probably the op-amps you're using.  What are they?  Most can't swing rail-to-rail, in fact many swing further in one direction than the other, hence the asymmetry.

Try CMOS output amps.  If you still have a problem then all you have to do is include a trimmer pot and back-to-back diodes in the resistance chain which sets the frequency (usually from the Schmidt's output to integrator's input).  The trimmer can compensate for any slight errors and you can set a perfect 50:50 mark/space ratio to give you the symmetrical triangle you're after.

You can replace the legs of a phase-shift oscillator with matched FETs (4007 for example) and 'stop' resistors to vary frequency.  I think there's something over at GEO about this.  You'll get a change in amplitude across the frequency range though.

If sine is ALL you're after, the 'easiest' circuit would be wein (edit: meant Wien - Freudian slip!) bridge.  Stable amplitude, low distortion and reasonable range (and probably a lot lower parts count than the thing you're trying to cook up at the moment!).

[Boofhead]

I don't know what circuit you are buillding but a phase-shift oscillator isn't the type that produce tri-waves, phase-shift produces sine (some distorted but still basically sine).

If it's like the one Tim Escobedo posted the link to , which is an OK design,  then it's a Schmitt-integrator oscillator.

On those you get the symmetry correct first (even if it changes the frequency)  *then* you  adjust the timing cap or resistor to get the frequency.

[Brian Marshall]

Using a soundcard in a computer as the front end of a oscilloscope may not faithfully reproduce the such low frequency signals. I'd guess the inputs are AC coupled. And freqs. of .25hz to 2 hz are damn near DC.

Unless you absolutely need a pure sine, a little bit of distortion of the wave isn't necessarily a bad thing. Especially with a LFO, it may or may not be very noticeable at all.

sorry, let me clarify.... while working on it, i was runing it as an audio generator... i know that a sound card wont show much under 50hz