The "Fake" PWM method

[mute]

So this has been bugging me for ages..I like to think I know what I'm talking about, but this seems rather simple and I've never really sought to sort it out...so I figured I'd just ask.

The fake "PWM" you can pull off on synths that don't have PWM but do have Osc modulation and Sync, where you set both to square wave, or pulse.. then modulate one osc while having them set to Sync. Mostly known as a favorite trick on 80s roland DCOs like the 3P and 8P. What is technically happening there since they are hard sync'd? It's really phase modulation isn't it? Or is it something else like waveform cancellation is causing actual PWM? Also.. isn't this very similar to how supersaw (not the minibrute enhanced version) works? Perhaps that's how Roland thought it up in the first place?

Or, am I way off on everything?

[Bitexion]

I think it's a form of FM modulation happening. The dx7 can also do that trick with some specific settings, making it sound like PWM.

[Infrasound]

I was of the belief that what can be done on the JX10 (and the 3 and 8P) actually is PWM.

[pflosi]

Technically this trick is actual PWM. As long as the two osc pitches never cross and stay within a certain range.

The higher the pitch of the modulator, the shallower the pulse of the carrier.

[Stab Frenzy]

I always thought that PWM was a trick for making a single oscillator sound like two beating against each other, so making a sound like PWM with two oscs should be easy. The osc sync would make sure you wouldn't have phase cancellation between the two except when you wanted it, ie it would replicate the limitations of trying two make one osc sound like two.

[loungedumore]

All you could ever want to know about pwm http://www.soundonsound.com/sos/Mar03/a ... rets47.asp

Came in handy for the esq1 /sq 80 too.

[calaverasgrande]

I always thought that PWM was a trick for making a single oscillator sound like two beating against each other, so making a sound like PWM with two oscs should be easy. The osc sync would make sure you wouldn't have phase cancellation between the two except when you wanted it, ie it would replicate the limitations of trying two make one osc sound like two.

this.

[Jinsai]

Hard sync makes one wave (the slave) reset its waveform when the other wave (the master) restarts.
The "resetting within a cycle" effect of hard sync effectively changes the shape of the wave, which changes the harmonic structure and thus the timbre or sound of the oscillator.

Pulse waves are unique in that they are just on/off - there's no ramp or slope to them. That means when you reset them, you're effectively changing the duty cycle (the length of the on/off parts of the wave).

What else could do that? Well, once you think about it, you realize that changing the pitch of a pulse wave would have the same effect. A higher pitch produces shorter duty cycles clumped together frequently, lower pitch slightly longer duty cycles less frequently. Of course, because you're changing the duty cycle AND the frequency/periodicity, the pitch varies, too.

Pulse Width Modulation literally modulates the width of the pulses while keeping the actual pitch - the periodicity of the wave - the same. But because of the shifting width, there is some illusion of pitch variation (and thus, "fatness").

Since these things are close together, it's totally possible to fake PWM using hard sync:

Set both oscillators to the same pitch
Set your "slave" oscillator to a square wave
Set your "master" oscillator to a square wave
If you turn "Sync" on now, you probably won't hear any difference. Turn it back off.
Set up an LFO with a triangle or sine wave. Assign it to control the pitch of your slave oscillator, and give it a frequency and depth that is sort of like a reasonably fast police siren. Wee-ooo-wee-ooo, not weeeeeeeeeeooooooo.

Now turn Sync on.

You've got your fake PWM...except it's not really fake. You're effectively modulating the pulse width of oscillator 1, but in a sneaky way.

Slave oscillator has fixed duty cycle (timbre) and periodicity (pitch) on its own. But the LFO you set to it makes the periodicity vary up and down, so you hear the pitch variation.

But when you slave it to oscillator 2, oscillator 2 forces the wave to re-start its duty cycle at a fixed periodicity.

The net effect is the duty cycle changes at the rate your LFO is modulating the slave oscillator, but the pitch is determined by your master oscillator due to the hard sync. It's pretty clever.

This trick only works with pulse waves, because of their on/off nature. With any other sloped or non/on-off waveshape, you end up with timbral changes instead of PWM-ish sounds. (There's a different way to fake that kind of stuff for other waves using square LFOs, but I won't go into that now).

[Bitexion]

I just tried this setup on my modular system, and can confirm that it works

[shaft9000]

yep - pulsewaves rule synthness. their sublimity and utility know no bounds.

[mute]

Jinsai, that is a perfectly detailed explanation. Thanks everyone. Fun to actually talk synthesis for a minute

[calyx93]

My favorite trick for the SQ-80 (and ESQ-1/m) - and thanks to the 3 oscillator architecture, there's still another oscillator remaining to use for sub or detuning.

[LWG]

I always thought that PWM was a trick for making a single oscillator sound like two beating against each other, so making a sound like PWM with two oscs should be easy. The osc sync would make sure you wouldn't have phase cancellation between the two except when you wanted it, ie it would replicate the limitations of trying two make one osc sound like two.

Hello!

Real-time control over the pulse wave's duty cycle has no singular purpose.
The chorus effect to which you are referring is the end result of only one type of controller routing.
The reason the lfo controller routing to pulse width is the most common is that when some synth designers
had the option of using only one or two hard wired controller routings, the lfo-driven chorus effect was usually
the most common.
Its an aesthetic decision, but one equally driven by budgetary design constraints, not a statement about a singular purpose for that one function. In a more open (modular) or complex instrument, the controllers that can be routed to change pulse width are as varied as those that are used on the more familiar vco, vcf, vca destinations.

Many of ARP Instrument's signature sounds used pulse width modulation with the ADSR envelope as the mod source. It was the basis for some of their reed-modeled sounds, as well as one of the main elements in classic sounds such as the "Fuzz Guitar" preset from the Pro Soloist.

Regards,

-L

[LWG]

All you could ever want to know about pwm http://www.soundonsound.com/sos/Mar03/a ... rets47.asp

Came in handy for the esq1 /sq 80 too.

Hello!

With all due respect to the Gordon Reid article, this technique was known to some
ESQ-1/SQ-80 heads, as it was explored back in the early 90's in an article penned by
Kirk Slinkard, in Ensoniq's newsletter, "Transoniq Hacker".
An Ensoniq-related site here, linked to a reprint of Kirk Slinkard's original article as it appeared in T.H.:

http://www.scoreforsale.com/html/pwm_1.html

fwiw, Transoniq Hacker was imho, the best synth-related company newsletter ever published to
date, as most of the articles were applications-based.
The contents of newsletters from the big three around the same period, consisted mostly of ads and
short, artist interviews.

Regards,

-L

[loungedumore]

All you could ever want to know about pwm http://www.soundonsound.com/sos/Mar03/a ... rets47.asp
Came in handy for the esq1 /sq 80 too.

Hello!
With all due respect to the Gordon Reid article, this technique was known to some
ESQ-1/SQ-80 heads, as it was explored back in the early 90's in an article penned by
Kirk Slinkard, in Ensoniq's newsletter, "Transoniq Hacker".
An Ensoniq-related site here, linked to a reprint of Kirk Slinkard's original article as it appeared in T.H.:

http://www.scoreforsale.com/html/pwm_1.html

fwiw, Transoniq Hacker was imho, the best synth-related company newsletter ever published to
date, as most of the articles were applications-based.
The contents of newsletters from the big three around the same period, consisted mostly of ads and
short, artist interviews.
Regards, -L

Indeed . Having owned an esq1 and being a current owner of an SQ80 I have seen this , just thought the s.o.s artical was more .... broad . There was another ( jx10 specific I think ) that was usufull as well out there too .