Can't get this triangle LFO right

[blackieNYC]

I have this lovely triangle on the scope coming off pin 13, the looped input of a Schmidt inverter (40106), but I can't get it to drive the transistor in this PWM distortion.  The square wave works pretty well if I use that. If I connect pin 13 to a 100k series R and a 10uf, as I read some where, it disappears when I connect it to the base of the transistor, and there's no modulation.  The schematic only gives you the square. http://www.jiggawoo.eclipse.co.uk/guitarhq/Circuitsnippets/snippets.html see the PWM circuit further down the page. Oh please.

[blackieNYC]

I have a triangle LFO that, to clarify, I want to introduce in between a couple inverters in the 40106.  A 500k pot to ground from tha point works fine.  So I want the LFO to imitate the 500k pot, as it swings a transistor up and down.  I do t really understand how that's done.  And sweeping the pot up and down sounds great, but a square wave just sort if turns it on and off.
A triangle, or sine, would be great.  Here's another LFO.  But how to integrate it?  And I think I need some LFO amplitude control
http://circuitworkshop.com/forum/index.php?topic=1136.msg8939#msg8939    See the simple triangle and square generator.  Can this be tied to pin 3 of the 40106 PW modulator?

[R.G.]

You're probably running into some of the fine-print details of using transistors as variable resistors.

Bipolars as variable resistors are are really only good for signals well under 1V, possibly LOTS under 1V if you do want linear. The place that pot is in the circuit looks like the pot side of that timing cap is getting a 0-5V signal from the stage driving it. Most of the "variable resistors" in the world that handle multi-volt signals are either mechanical pots or LDRs. Not that it can't be done, but at first guess, I'd suspect that is one of the first places to look. Above about half a volt across it, a bipolar is either off, saturated on, or looks like a constant current sink. If you're not using a pot or an LDR, every "variable resistor" comes with a long and different set of special conditions that have to be satisfied to make it work right.

If I get what you're trying to do, you want to have a triangle LFO modulate the duty cycle of a PWM. My first shot at that would be with four opamps; two to generate the triangle, and two to do the PWM. The reason for so many is to leave you places to control things like the size of the LFO, it's DC voltage in the middle, the zero-modulation duty cycle, etc.

There is a tendency in English language use and perhaps in all human psychologies to think that if something is named, it is known. LFOs - what could be simpler? Generate a modulation waveform of your choice and connecticate it up to the thing to be modulated. The reality is that the "thing to be modulated" has as many special cases and considerations to be satisfied as any prospective future spouse to make it all go smoothly.  Even when it goes smoothly, it's because you have accidentally satisfied the special conditions without knowing it.

Getting an LFO to modulate something that can be modulated requires knowing at least
-  The DC conditions the modulated-thing needs to be modulated, such as the absolute limits of modulation signal up and down, and the "zero modulation level"; will all modulation be up from some minimum, down from a maximum, or wobbling around some middle ground. I've known all three types.
- The kind of modulation - AC, DC, voltage or current. Or incident light (LDRs) or magnetic field (Hall effect devices) or temperature ( there are variable resistances depending on temperature); it goes on.
- The compliance of the LFO input; if you need to feed in a voltage, how much current do you have to feed in to make that voltage happen, and is this linear or not? Case in point (and I'm giving away a favorite trick of mine here) if you want to modulate something exponentially, you can feed a LINEAR voltage modulation into the base of a transistor. The collector current is then an exponential function of the voltage. Works for diodes too, but it's trickier to get them separated. So to get exponential variation in something like an OTA operation, feed the input current pin (which is the base of a transistor or a transistor based current source in all I've seen) with a LINEAR voltage source that can supply the current needed by the base; this is nonlinear, of course, so you need a low impedance voltage signal, which is limited to no more than X current. Same for current-input modulators; there some voltage range needed to make that current come true. Your LFO driver had better do this right, or it won't work.
- The control curve of the LFO and the modulated thing; these have to be crafted to get your final modulated signal to be overall linear, exponential, square law, reverse log, whatever it is you want. That also means you have to first know what you want.

So I'm guessing you've missed one of these details, possibly the voltage compliance of a bipolar transistor as a variable resistance, or the necessary current drive to make the input of the bipolar do the right thing, or the necessary voltage swing on the base of the transistor to make it be a linearly varying current source, something like that.

Just a guess.

[blackieNYC]

Thanks! I can see this.  I think the amplitude is not enough, so the swing of the "variable xistor" is not enough. I can get a little modulation going when it's a square wave, just because it tends to stick out more. It isn't full swing either.
As far as the audio path, I'm worried if I swap these 40106 inverters for op amps that I will lose this nifty buzzy synthy thing that's happening.  Op amps as the modulator and generator, and the inverters as the distortion?  Or will I indeed get the same kind of sound, but more flexibility if I use op amps as the audio because of the 386 in front will push the waveform into this distortion?  You suppose?

[R.G.]

As far as the audio path, I'm worried if I swap these 40106 inverters for op amps that I will lose this nifty buzzy synthy thing that's happening.  Op amps as the modulator and generator, and the inverters as the distortion?  Or will I indeed get the same kind of sound, but more flexibility if I use op amps as the audio because of the 386 in front will push the waveform into this distortion?  You suppose?

Most of effects work is about side effects - simple circuits that have consequences that are unintended as far as classical design goes, but have an interesting sound. So yeah - opamps for the LFO/modulation because that gives you predictability where you need it, and inverters in the audio path because you like the sound. Opamps and 386 for distortion too/as well? Maybe.

How does that sound? That's the only good question for pedal work.