Pro 3 Square Wave PWM Question

[creativespiral]

Finally got my 3!  Amazing board... big grin on my face all day exploring sound design.

I have a question about PWM on Square/Pulse waves.   Not sure if this is a bug, a design limitation, or if I'm just doing something wrong - 
 
After setting up an oscillator to Shape 128 (Full Pulse/Square Wave), and then assigning an LFO to Shape Mod destination for PWM with a Bidirectional Triangle, I'm noticing that the square wave only modulates in one direction (positive), and the negative cycle of the triangle is ignored/clipped, causing the modulation cycle to pause at square shape for half the time.   This makes it so you can't get a full range PWM sweep.

I'm assuming this is a matter of the mod input for Shape Mod not allowing negative values, since 0 shape mod value is the 50% square pulse rather than a mid point 128 value (like on other products.. ie Rev2).  So there's probably two different approaches to tackle this.  Might make sense to set 128 value as 50% square?   Or can the square be set to accept negative modulation to its shape mod parameter?

Or maybe I'm just missing something and there's another way to get a full sweep of PWM on square wave?   

[guyaguy]

Finally got my 3!  Amazing board... big grin on my face all day exploring sound design.

I have a question about PWM on Square/Pulse waves.   Not sure if this is a bug, a design limitation, or if I'm just doing something wrong - 
 
After setting up an oscillator to Shape 128 (Full Pulse/Square Wave), and then assigning an LFO to Shape Mod destination for PWM with a Bidirectional Triangle, I'm noticing that the square wave only modulates in one direction (positive), and the negative cycle of the triangle is ignored/clipped, causing the modulation cycle to pause at square shape for half the time.   This makes it so you can't get a full range PWM sweep.

I'm assuming this is a matter of the mod input for Shape Mod not allowing negative values, since 0 shape mod value is the 50% square pulse rather than a mid point 128 value (like on other products.. ie Rev2).  So there's probably two different approaches to tackle this.  Might make sense to set 128 value as 50% square?   Or can the square be set to accept negative modulation to its shape mod parameter?

Or maybe I'm just missing something and there's another way to get a full sweep of PWM on square wave?

We're Sequential synth twins!

It seems the Shape Mod value needs to be at around 110 or so for a full sweep. It's kind of like the Rev 2 where setting the Shape Mod to center will give you a square and you can sweep from there. I'm not sure why it's not at the center point on the Pro 3 though--presumably to make the most of the shape mod of other wave forms.

[creativespiral]

Yeah, I experimented with dialing up Osc Shape Mod knob to allow the clipping to be avoided, but still, you can't get a full PWM sweep... only 50% cycle to 100%,  nothing between 0% and 50% it seems, which rules out a lot of PWM sound design - doing deep PWM sweeps with one or more oscillator stacked, and the possibility of going through zero in both directions.

My guess is that there's probably a way to just alter the firmware a bit to either allow negative modulation values to the shape mod, or adjusting it so the center-point value of the pulse (128) is the 50% duty cycle square.   I dunno, maybe there's some other design constraint, but would be really nice to be able to have full PWM range on the square/pulse. 

[Quatschmacher]

I wonder if this is similar to something I read about panning not being able to be fully exploited (might’ve been on the Pro 2). In that instance one had to use the mod matrix to gain a full sweep, I think applying a DC offset or something.

[creativespiral]

Just checked for any workaround in mod matrix for full PWM sweep - no DC offset fix possible, since it seems the input of the modulation for shape mod won't accept negative numbers, and the 0 value is the 50% duty cycle on the pulse.   

Also, FYI - you mention a Panning issue.   I just did some testing and figured out what's going on with that:

A value of +127 pans hard right, and -127 pans hard left, if you route in the Mod Matrix:  DC directly to PANNING destination, however, the LFO seems to run at values of 1/4 of that of the Mod Matrix (this is similar to other sequential products).    So the problem here is that the LFO > Panning option at maximum value 128 is only generating about a 32 value relative to the mod matrix... it only pans slightly left/right via the LFO routing... almost imperceptible.   This should be a fairly easy fix... need to add a x4 modifier  (or perhaps even x8 would be good) at the output from LFO to Panning Destination.   x4 would make it so it perfectly pans hard right/left at full value.... x8 would allow you to use LFO with purposeful clipping on both ends of the swing (which is actually very useful for panning sound design)... I would use the x8 multiplier.     

(Currently, if you are desperate for a workaround for LFO Panning hard left / hard right, you can get a full swing by routing the LFO > Panning destination with max value, and then using up three additional MOD slots with LFO > Panning at full value)