Another dual LFO question

[blackieNYC]

I want to do another dual lfo (two frequencies, blended) and I haven't done this with op amps before, only with the 40106 schmidt triggers.  Here's the tri-vibe, which I hastily took down off the breadboard and started to assemble, unfortunately.  Changed my mind and want to try the dual LFO.  Working on perf. Tell me if this is the way to do this:
Duplicate U3a completely - the Vref, 100k, 220k, 12k, and rate control, and connect that at pin 6 of U3b, along with the first LFO, where the inverting input would make a decent mixer.
And, also duplicate the 68K, thus having two 68Ks connected at U3b pin 7, one from each LFO. 
That 68k resistor I'm not so sure about. (I have a hard time understanding FB loops that span two stages in general)
Would this do a nice job of mixing the two LFOs?

[armdnrdy]

I'm not exactly sure what you're trying to accomplish with mixing two LFOs but...

The A/DA Final phase uses a LFO for the sweep and another to add modulation on top of the sweep.

It's a pretty straight forward design that you might get some ideas from.

[blackieNYC]

Thanks - Saw the A/DA, not quite what I'm looking for. I will need them mixed. Aiming for the pseudorandom LFO a bit like geofex but not with the 40106. Two square waves at different frequencies - two sweeps overlapping, adding, subtracting together - waveform looks like a staggered city skyline. Then smoothed out a bit. It can sound like S&H with a square, and a random swirly thing with sine waves.

[dbp512]

If you summed 2 different LFO's you'd end up with a more complicated waveform, but I don't think it would be too random sounding. I haven't read the geofex article yet, so it if says otherwise I'll stick my foot in my mouth. If you're looking for that, then perfect, like you said it would appear more like a city skyline than an LFO wave. But if you're looking for random, some sort of sample & hold would be better. Or you could use a TAPLFO, which has a random waveform.

[blackieNYC]

Understood, D. The pseudorandom LFO is just that, and mathematically it should all cycle round again, and it sounds like it sometimes, because it isn't random of course.  It was very easy to do with the  Escobedo PWM, and you see it on the scope. If the two LFOs are tweaked by ear to some non-harmonic interval, which ain't hard, the square wave waveform is rather complex and interesting. Throw in a third oscillator and the waveform is pretty random. I always find myself selling this geofex idea. Random-steps wah sound, psycho stereo panner, etc.
Anyway, I'm doing the perf and I have an extra op amp, so I'm going to just do it. If they don't mix well, I'll just kill it. The trivibe is certainly good enough as is.

[dbp512]

Oh, I'm sure it would be cool. I just didn't realize it would sound random that quickly. Does that makes sense? I thought we'd be able to detect the pattern but I guess not. Perhaps using different waveforms, like a sine and triangle, could produce some interesting results as well. I haven't done much experimenting with modulation effects, so all I can do is spitball random ideas.

[samhay]

If you decide to add another LFO, you just need to duplicate everything around U3a and U3b. The second depth pot then connects to the (+) input of U2b, which will act as your LFO mixer.
The only downside to this is that your LFO indicator (red LED) will give you an indication of the mixed LFO, which may make it hard to dial in specific rates, whatever. If this is a problem, you can hang an LED and resistor to ground off the output of U3b, which should blink in time with each LFO. You might have to do some tweaking to get this to go out on the bottom of the swing though.

As Vref is already shared with the audio side of the circuit, I wouldn't bother duplicating it solely for the second LFO driver.

[ElectricDruid]

Changed my mind and want to try the dual LFO.  Working on perf. Tell me if this is the way to do this:
Duplicate U3a completely - the Vref, 100k, 220k, 12k, and rate control, and connect that at pin 6 of U3b, along with the first LFO, where the inverting input would make a decent mixer.
And, also duplicate the 68K, thus having two 68Ks connected at U3b pin 7, one from each LFO. 
That 68k resistor I'm not so sure about. (I have a hard time understanding FB loops that span two stages in general)
Would this do a nice job of mixing the two LFOs?

I've got a couple of comments;

I wouldn't duplicate the Vref. There shouldn't be any need. If you've got Vref, you've got it.

The LFO is a schmitt trigger/integrator design. The first op-amp produces a square wave output (might be useful) and the second stage produces a triangle. The 68K feedback takes the triangle back to the schmitt trigger so that it flips into the other state when the integrator cap(s) has charged sufficiently close to the input voltage. The basic action is "Schmitt outputs positive square, integrator charges up. When cap is sufficiently positive, Schmitt triggers and outputs negative square. Integrator starts charging the cap towards -ve voltage. When cap is sufficiently negative, schmitt triggers and outputs positive square. Repeat".

The part I don't understand is that next op-amp which I presume is biasing the thing. The triangle goes through some clipping diodes to get turned into something sine-ish, I get that. But the op-amp I don't get - positive input tied to +9V? what's that about? Consequently, I'm not much help with the mixing part...

Tom

[duck_arse]

if you duplicated the U3 oscillator, up to and including the diodes to Vref, dis-connected Vref from the depth pot, and connect the new osc's diodes to that free end, you could then mix between the two oscs (but with no proper depth control). I thimk.

(didn't RG or PRR once tell us NOT to hang a cap off the output of an opamp, even if it is Vref?)

[blackieNYC]

I should not have said "duplicate v ref". I meant duplicating everything from v ref through.
But it sounds like you don't think I can mix two "U3a"s into one U3b. I'll need to duplicate 3a and 3b.
If I made a panner, that could be very cool, but I would have to add some gain to the following stage, and a voltage divider depth control. This would give me a depth control, and make up for the loss due to the mixer.  I'm hoping to just add a switch to add or not add a second LFO, of the same depth as the first. I don't need full control of everything.
I'll have to rebuild the LFO on the breadboard and experiment.

[ElectricDruid]

I should not have said "duplicate v ref". I meant duplicating everything from v ref through.

Ok, I didn't get that. That's fine.

But it sounds like you don't think I can mix two "U3a"s into one U3b. I'll need to duplicate 3a and 3b.

No, I don't think you can. 3a and 3b are a unit - a whole LFO. You can mix two LFO outputs easily enough though. No problem with that, in principle.

In practice, I think U2b is basically a non-inverting amp stage, which doesn't make an ideal mixer.

If I made a panner, that could be very cool, but I would have to add some gain to the following stage, and a voltage divider depth control. This would give me a depth control, and make up for the loss due to the mixer.  I'm hoping to just add a switch to add or not add a second LFO, of the same depth as the first. I don't need full control of everything.
I'll have to rebuild the LFO on the breadboard and experiment.

Yes, doing a passive mix and then using a bit of make-up gain in the U2b op-amp seems reasonable. As you say, breadboard it and tweak it.