Need Help: Escobedo PWM Build

[blackieNYC]

yes, the 10k pot is LFO depth/output level control, opening and closing the transistor. the 33k is probably an adjustment to the LFO depth so that the variable resistance of the transistor approximates the range of the pulse width pot.  or perhaps keeps the transistor from clipping.

[anotherjim]

The first schmitt trigger output will depend on the last level it reacted to coming from the amp. As the guitar signal decays, it will fall somewhere between the upper and lower input trigger threshold levels about 4.5v. For the sake of argument, lets say below 3v it sees 0 and above 6v it sees 1. If during note decay it had exceeded 6v but then decays towards 4.5v and never again goes below 3v, then the trigger remembers last seeing a 1 and outputs 0 and continues to until a new note swings the input below 3v.

A real music signal is rarely as regular as a sine wave, and appears to wibble and wobble about with little symmetry, but the schmitt trigger makes some sense of it.
 
The second schmitt trigger will tend to normally see a 0 input due to the pull down resistance of the pulse width pot. It will only see 1 when the first trigger outputs 1 and then only for the duration of the time it takes the input capacitor to charge which depends on the value of the pot and/or the LFO transistor - and that of course sets the pulse-width.

Back to the LFO, the 100uF cap and the 33k resistor are to slow the square wave output from the schmitt oscillator into something resembling a shark tooth wave - almost a triangle but with curved slopes. Then the transistor can be seen as a buffer to that, but its also converting the variable LFO waveform into a variable resistance.

[turdadactyl]

Ok, so can we take this a little further up the abstraction ladder?

Let's say I want to take the PWM portion of this circuit (right side, 40106 portion) and instead of using the 386 gain circuitry, use a different distortion section.  I guess I'd have to properly bias the output of the distortion section to square with the thresholds of the Schmitt Trigger, right?

Here's an example I'm trying to hack around with to figure this out.  This is a simple distortion circuit from Premier Guitar.


And for reference, here are the 40106 specs.


When I just hooked the output of the distortion up to pin 1 of the PWM circuit it passed the signal but didn't give me the PWM effect.  That made me think my input voltage to the 40106 wasn't properly triggering it.  My meter said it was around 6V.  Anybody up for educating me a little more?  Thanks again.

[turdadactyl]

Any takers?

[anotherjim]

Try it without C2.
You might get very poor sustain with just that one transistor stage. Remove R3 (emitter to ground) for more gain. Increase R2 to get the collector closer to 4.5v (which the LM386 will have been internally biased to).

[turdadactyl]

Thanks Jim. I'm trying to make sure I understand this fundamentally so I can change it, build off it, apply it to other situations. Why are you trying to get to 4.5v? Is that only because the 386 in the original circuit was biased to that or based on the threshold specs of the 40106?

[anotherjim]

Yeh, if you bias the input somewhere near the middle of the thresholds, you have the best chance of the effect lasting through the guitars notes sustain. The LM386 amplifier happens to bias it's output at half supply volts - if the design used an op-amp instead, it would be biased to do the same. Half supply voltage is usually near enough half-way between trigger thresholds.
If the input was biased close to a threshold, the output will stop early as the decaying signal cannot reach the other threshold.

To keep it hitting the thresholds for as long as possible, you do also need plenty of gain. IIRC, the LM386 is x100, you will struggle to get a single transistor stage anywhere close to that. Another stage, just as drawn, in front of it would do it.

[turdadactyl]

Well in that case I'm going to have a near impossible task of doing this with one transistor. The datasheet says the short from pin 1 to 8 makes the gain 200x.