LFO woes

[Brian Marshall]

Ok,  I've been working on a fairly complex project of my own design.  I am having a big problem with the LFO.  I tried phase shift ocliators, but there are way too many problem with, especially with really low frequencies.

I have had the most success with a 2 opamp triangle generator, and then filtering it either with diodes, and some other parts, or annother integrator.  

the biggest problem i have is getting really low frequencies.  I am not truly aware of the math involved with the triangle generator, but i know that voltage head room play a big part.  I dont seem to be able to get down below  1hz before the square wave starts to become apparent at the triangle output, and it turns my wave in to more of a tangent than a sine wave.

I know you guys are going to say post the scematic, but i really dont want to draw it all up... anyone have any suggestions?

[gez]

This thread explains the basics behind setting up a simple two amp triangle generator.

http://diystompboxes.com/sboxforum/viewtopic.php?t=19520

If I recall you're integrating the square wave passively before it reaches the integrator.  Do this actively instead.  Take another integrator (don't incorporate it into the feedback chain) off the triangle output and use a dual-gain pot so that one half controls rate and the other attenuates the signal going into the second integrator.

PM me if you don't understand and I'll guide you through it.

[puretube]

use better quality capacitor! (low leakage; when using polarized ones: take tantalums)

You`ll get three-minute sweeps from 2.2uF !

[ExpAnonColin]

This design has always worked flawlessly for me, with all non-polarized caps I've used... up to 10uf even, and it's still a triangle wave.

-Colin

[gez]

This design has always worked flawlessly for me, with all non-polarized caps I've used... up to 10uf even, and it's still a triangle wave

Fine, but tolerances of large value caps aren't too great so results (i.e. frequency) can be inconsistent from circuit to circuit, not to mention caps going leaky with time and buggering up everything.

You can get really low sweeps by  varying the proportion of the Schmidt's output voltage applied to the integrator’s input rather than varying the current to control frequency.  If you look at the schematic linked above, this is one way to do it.  Making the 1k stop resistor smaller will get you reeallly slooooooooooooow...

[puretube]

err... which link?

[gez]

err... which link?

the thread I linked to in my first reply.  Scroll down and you'll see a schematic I used as an example on how to set these up.

[ExpAnonColin]

gez-I haven't needed an exact frequency, so I haven't had to worry as such.

-Colin

[Brian Marshall]

This design has always worked flawlessly for me, with all non-polarized caps I've used... up to 10uf even, and it's still a triangle wave.

-Colin

this is almost exactly what i have built

right down to the 68k cap.  I am starting to think the closer i can get to 100k, the lower i can get with out the tangent distortion.  Of course (i think) this would make the amplitude smaller, but that can be amplifed later

i'll try again

[puretube]

follow gez`s path, and use a proper cap!

[Brian Marshall]

This design has always worked flawlessly for me, with all non-polarized caps I've used... up to 10uf even, and it's still a triangle wave

Fine, but tolerances of large value caps aren't too great so results (i.e. frequency) can be inconsistent from circuit to circuit, not to mention caps going leaky with time and buggering up everything.

You can get really low sweeps by  varying the proportion of the Schmidt's output voltage applied to the integrator’s input rather than varying the current to control frequency.  If you look at the schematic linked above, this is one way to do it.  Making the 1k stop resistor smaller will get you reeallly slooooooooooooow...

this is acutally how i had it set up... i took colins drawing back to my bread board and saw it was set up different...

colins setup is not amplitude stable

i still cant get below 1hz without distortion

i'll draw what i have tomorrow, or tonight some time

[ExpAnonColin]

68k cap?!  Jesus, where'd you get that?   :wink:

I actually use a resistor of around 90k.  It's always treated me nicely in terms of being amplitude stable, no distortion.  You want to get as close to 100k as you can without reaching or exceeding, so I use my DMM.  I've gone as low as maybe 1/10hz, and have had no problems.

You know, you might just want to try an ICL8038 or a XR2206, those most likely will be as stable as you'd ever need, and you get the other waveforms without extra components... just a suggestion.

-Colin

[Brian Marshall]

I hear ya colin.  I really want to stick with 9 volts for this though.  If i went up to 18 volts this would work anyways due to the higher headroom.

I'll draw some stuff up soon and post it.

[puretube]

Brian: maybe it`d be easier to take a little trip through the "schematics" pages, and see how other people have done similar circuits.
go looking for phaser- or flanger-, or chorus-LFO schemos...

[Brian Marshall]

phaser and chorus circuits usually dont have that slow of an lfo.

I've been looking for flanger circuits.  they usually use a triangle wave (i think)

[puretube]

OK, I`ll take back the chorus!
But do I love a phase-sweep over a whole song!

[gez]

I hear ya colin.  I really want to stick with 9 volts for this though.  If i went up to 18 volts this would work anyways due to the higher headroom.

Why do you need a waveform with such a large amplitude?  If you’re using FETs as variable resistors then you don’t need/it would be unwise to do this.  If you’re using LDRs then you still need a driver for the LED, which is often a FET or trannie.  Likewise, you don’t need a huge waveform.

I tend to go the other way with these types of circuits and use the lowest supply I can get away with.  If you use a 7621 op-amp it can be run off supplies as low as 2V.  What’s the point?  Current spikes are miniscule so it’s more-or-less guaranteed tick free (though it’s still wise to pay close attention to layout etc).  For a sine wave I tend to use inverters run off a 3V supply.  I use five, yet the total current consumption for the LFO is a mere .5mA.

If you use a pot from Schmidt’s output to centre of a divider/Vref op-amp (inverter if you’re using inverters)/ground with a bi-polar supply, and feed the wiper to the integrator via a resistor then you can get incredibly slow LFO rates.  Just think it through, by reducing the available output voltage of the Schmidt to virtually nothing the current to the integrator is minuscule so it will take ages for it to ramp up and trigger the Schmidt.  If you’re doing this I don’t understand why you’re having a problem.

You don’t have to give everything away, but it would help if you explain what you’re trying to do.

[Brian Marshall]

Why do you need a waveform with such a large amplitude?  If you’re using FETs as variable resistors then you don’t need/it would be unwise to do this.  If you’re using LDRs then you still need a driver for the LED, which is often a FET or trannie.  Likewise, you don’t need a huge waveform.

-------i dont really care about the amplitude in preticular, but the higher the voltage head room the lower the frequency.  the ramp can only go so high.

I'll draw this up soon.[/b]