Tim Escobedo's PWM

[tcobretti]

Has anybody built it?  I've been thinking about it, but I'm not sure how to incorporate the LFO.  I guess you'd just omit the connection from pin 13 to all the other input pins and wire it to the LFO?

[Tim Escobedo]

Yup.

[tcobretti]

Thanks Tim.  While I'm at it, thank you for all your great designs. 

I'll try to work up a vero layout in the next coupla days.  Granted, I still have to order the 40106, so I won't be building it any time real soon.

[choklitlove]

i built it without the LFO and i still like it a lot.  i like it better than the samples i've heard with it.  if you do decide to use it, i'd put a switch in to on/off it.

i used this layout: http://geocities.com/worthekik/pulsewave.html

[Processaurus]

I'd like to try this with envelope control, my guitar synth has some really funny 80s sounding patches that rely on changing pulse width on the two square wave oscillators.

[tcobretti]

i built it without the LFO and i still like it a lot.  i like it better than the samples i've heard with it.  if you do decide to use it, i'd put a switch in to on/off it.

i used this layout: http://geocities.com/worthekik/pulsewave.html

Thanks for the layout, and a switch is a very good idea.

An envelope control would be cool.  This circuit seems like a very cool noisemaker that's also very simple, so some mods to expand on what it does would be nice to have.

[Tim Escobedo]

An optocoupler like on the Uglyface could be grafted on with minimal headaches to implement envelope control. The .005uF cap and the 500k pot of the PWM might have to be tailored to fit the response of the optocoupler, but it should work OK.

It may be possible to get a pseudo "two oscillator out of synch" sound by running the output of the 386 to two different gates, one with a fixed pulse width, the other modulated by the LFO, and the outputs then summed together. The result should sound (and look) like a more animated pulse passing through each duty cycle. Those few unused Schmitt triggers have quite a bit of potential.

[Jaicen_solo]

You could also try using six square waves with fixed pulsewidths. Mixing them together you could then get some interesting waveshapes/harmonics. I'm sure the resolution would be very low, and the tolerances probably aren't that hot either, but I think it would still sound good/interesting.
I've actually built this, and whilst it is pretty good, it has a few drawbacks. The most obvious is the poor tracking of transients which I don't think can be improved easily. Also, it's prone to that nasty 'gate' effect at the end of the notes, which normally is ok, but it just doesn't sustain quite long enough. The second problem you could probably fix using the Rocktave approach, by grafting in an NE570 compander, but again, that's raising the complexity.
I'd like to hear the PWM combined with Tim's square wave shaper, i'm sure that would be interesting.
If anyone's interested,  try combining the Rambler with the PWM, now that's very cool

[Processaurus]

It may be possible to get a pseudo "two oscillator out of synch" sound by running the output of the 386 to two different gates, one with a fixed pulse width, the other modulated by the LFO, and the outputs then summed together. The result should sound (and look) like a more animated pulse passing through each duty cycle. Those few unused Schmitt triggers have quite a bit of potential.

Awsome.

I've been thinking about making a double PWM with a built in time delay of some kind one one of the PWM's for a psuedo 2 oscillator synth sound, but this may do it with 10x less trouble.  It'll be different than 2 unsynced oscillators, because they would be starting their cycle at the same time, and just ending at different times, but it should sound animated.  I wonder if inverting one and then sending it through a cap to another inverter (with resistor to ground), making another pwm stage, would give a phase delay that might make the two sound less synced.

You could also try using six square waves with fixed pulsewidths. Mixing them together you could then get some interesting waveshapes/harmonics. I'm sure the resolution would be very low, and the tolerances probably aren't that hot either, but I think it would still sound good/interesting.
I've actually built this, and whilst it is pretty good, it has a few drawbacks. The most obvious is the poor tracking of transients which I don't think can be improved easily. Also, it's prone to that nasty 'gate' effect at the end of the notes, which normally is ok, but it just doesn't sustain quite long enough. The second problem you could probably fix using the Rocktave approach, by grafting in an NE570 compander, but again, that's raising the complexity.

The crude blue box gate works alright, if you you have enough gain making the envelope it sees on the collector of the transistor it uses. 

[tcobretti]

I'm working on a PCB layout for the PWM with LFO, and I'm wondering if the either the 500k or the 10k pots in the LFO should be a trimpot.  Anybody have any ideas what these pots specifically do?

[Lateksi]

I built the PWM a couple of days ago. The basic circuit sound great, but with LFO it doesn't.
With LFO the pulse width seems just jump between minimum and maximum, which are somewhat controllable with 10k pot. Is that normal?
I'd like it to sound smoother, like when you manually turn the 'width' pot. Also, the LFOs minimum speed could be slower, but that could be achieved with lowering the resistance in series of the 500k pot?

[Jaicen_solo]

Raising the resistance of the speed knob, or increasing the cap to ground in the LFO will both give you slower speeds.
To smooth things out, move the 100uF cap before the 33K resistor. You can increase the 33K too, higher values will give you a more even ramp up/down.

[TELEFUNKON]

does that PWM have an almost 100% sweep range, or  50% ?

[Processaurus]

I'm working on a PCB layout for the PWM with LFO, and I'm wondering if the either the 500k or the 10k pots in the LFO should be a trimpot.  Anybody have any ideas what these pots specifically do?

Speed & depth.   Try it on a breadboard before going to the trouble of making a pcb design though. 

i think i might have tried the lfo too and not been too keen on how it worked, but that was quite a while ago.  looking at it it seems like it would have  a squarish wave form, since its coming from the output of the schmitt trigger inverter, and then getting filtered by the big 100uF cap.  The 33k resistor before it makes it into a lowpass filter, if the cap were before that it would work strangely, probably not smoother but the opposite.  A bigger resistor would make it smoother, but the drawback of that simple approach is that the amplitude gets messed with as the speed changes, lower as the speed increases.  You'd get a real triangular LFO if you took the signal from the inverter's input.  it would probably need an emitter follower to buffer it so it doesn't get loaded down and cause the lfo to latch up.

Also I wonder if the LFO might transfer smoother if the 3904 were replaced with a FET?  Maybe you could avoid the need for a buffer on the LFO if the depth pot were a 1M pot to ground from the inverter input, and a series 1M resistor from the wiper of the depth pot to the gate of the FET.  Sorta like the EA trem or phase 45.  Don't know if you'd get enough sweep without amplifying the LFO though, I think i'll try it out and see how it goes.

[tcobretti]

Thanks for the help, guys.  I plan to make the LFO switchable so I can use it at will.  I really just want it to be a weird effect that gets used very occasionally on recordings, so the LFO will be useful.  However, I need it to be capable of being very slow.

Processaurus, please let us know how it turns out!  I still have to order the parts, so mine won't be built for a while.

[Tuemmueh]

Are you still on the project, tcobretti? Would be great!

[tcobretti]

It has been moved to a back burner.  After listening to Tim's samples, I think the much loved Ugly Face might be a better choice, and instead of the LFO I'm gonna set it up for an expression pedal.