I love this VCO and that's really something for me to say... if I'd built this oscillator, I'd be very proud. The WR manual's are a bit abstract and that is fine. It allows users to decode it into our own ways like I'll do here. A scope is your friend with this module... or maybe your ear is enough if you're very coordinated in that way. I like writing things down. I have shit for brains so it also helps me to read this again in the future when I've forgotten something.
More to come... I just got this thing! And now have 2x... so maybe I'll have some tips using both together at some point.
Let's have some fun...
This module is a VCO with 'play' in the retriggering of the pulse generation. Because of this, you get interesting 'resets' or glitches and even divisions for interesting effects.
The actual waveforms produced (with exception of the simple square OP) should likely be thought of as a pulse-train instead of a typical saw/tri/sin, etc. This helps make more sense of some of the controls and what they do.
Constant Formant & Constant Wave Modes
Users may be unsure what the switch CONSTANT WAVE and CONSTANT FORMANT actually does. Just flipping the switch with a constant pitch or no FM modulation may seem like it does nothing in certain scenarios. And rightfully so. You have to change the pitch to see the differences... so that's the first bit to realise.
I like to think of it like this:
Do you want the wave pulses to 'run into one another' or not? If you do, put the switch in CONSTANT FORMANT mode. This locks the wave shape's pulses to a fixed width.
On the other side, if you want the waveshape to look the same across any frequency input, put the switch in CONSTANT WAVE mode. This effectively 'squishes' (as you increase frequency) the whole thing in the time dimension... just like a normal function generator or VCO would as you increase pitch.
An Experiment to Demonstrate the Difference
Set to CONSTANT WAVE and set BARREL to 12 o'clock, set FORMANT to just a hair CW of 9 o'clock (9:30 let's say), no inputs, start with PITCH knob fully CCW. Patch output from the FORMANT output.
Look (and hear) the waveform. It will have some shape. Now slowly increase the PITCH knob CW. Look at the wave shape... looks the same doesn't it? Keep going CW with pitch, rescale scope if you need. Same waveform.
Now put PITCH knob back full CCW and flick the switch to CONSTANT FORMANT. Do the same experiment by increasing PITCH knob CW. You should also hear the 'smear' almost like a little sync when the wave form shapes or pulses start to collide and combine. Be very observant of this and go slow increasing the PITCH knob.
Got it! Great!
Air is a clipping stage and VCA. Nothing too crazy here but keep in mind that this oscillator isn't inherently very 'squarey.' But this is just a matter of chopping the tops of those tris/saws/ramps off and you'll get a decent enough sounding square. Use the AIR for this purpose to great effect. Abuse it, in fact. Don't think of it simply in terms of something trivial like "do I want distortion." Think of it more of a waveshaper and you'll likely use it more. But that's just mental
I note soft clipping starts happening around the 2 o'clock mark of the dial.
And since it's a VCA, it will reshape based on external modulation signal shapes. This input alone adds a very wide variety of shapes when coupled with external VCOs especially those that are harmonically related to the fundamental. The gray capped dial is an attenuverter for the AIR CV input. 0 gain is dial centred at 12 o'clock.
The AM of this module via the AIR jack is really nice sounding. Try audio rate modulation of this input for sure!
Formant & Barrel
I've got these two listed together because at some point, you just cant separate them.
I'll explain what they each do first independently, then how they interact. To start, put the VCO in CONSTANT WAVE mode.
Formant (independent explanation)
FORMANT controls the width of the pulses. The more CW you turn FORMANT, the thinner the pulse all the way to an impulse train. This increases the brightness of the spectra. When learning this control independently, be sure to set the Barrel control to full CW. You'll know why later.
Barrel (independent explanation)
BARREL controls the 'tilt' of the waveform pulses from a ramp at full CCW, to triangle at midway, and to a saw at full CW.
All Together Now
The first observation that you'd want to make can be found by doing the following experiment:
Blanking Count 'Slope' Experiment
- Set PITCH to the line between '80' and '320', set FORMANT to 12 o'clock, set mode to CONSTANT WAVE, set BARREL to full CCW. Make sure AIR is set to 2 o'clock or less as to not clip the waveform.
You should have a nice ramp waveform (this is important).
- Now Slowly rotate the FORMANT control CCW while looking at the waveform on a scope until you hear the pitch jump down. Stop there! OK. Something weird has happened.
Let's talk about it.
Note that it did this pitch jump as soon as the rising part of the cycle (the ramp) touched the vertical falling edge of the previous cycle. Just queue this observation... it's not enough to fully understand what's going on... right now we just know that the rising part of the cycle touched the falling part and we got a pitch division... let's continue!
You're probably wondering why I called this the 'Blanking Count Slope Experiment'? Soon! Back to more experimentation:
- Set the FORMAT control back to full CW and BARREL to full CW.
- Now rotate the FORMANT control CCW and observe.
The waveform pulse's decay gets longer as you go CCW. At some point, it will take so long to decay, that it turns into a DC level and you wont hear it. For kicks rotate it full CW as well. It will turn into shorter and shorter pulses. We see that FORMANT acts as it was described above in the 'independent explanation' and there are no pitch divisions like before! Why not!?
Let's experiment MORE!
- Set the FORMANT control back to 12 o'clock.
Remember, the barrel controls the 'tilt' of the waveform. This is another way to say that it changes the ratio of the slopes of each sides of the pulse. At 12 o'clock, it is more or less a triangle which means the slope of the rising portion equals the slope of the falling portion of the cycle. Turning BARREL more CCW, it gets more rampy and more CW from 12 o'clock, it gets more sawey.
- Let's start with BARREL set to 1 o'clock. This gives us a waveform whose pulse has a just a slightly faster rise time than fall time.
- Rotate FORMANT CCW very slowly.
No pitch division. Hmm.
- Set FORMANT back to 12 o'clock. Let's now nudge BARREL to just shy of 12 o'clock... maybe 11 o'clock. This changes the waveform now such that the rise time is slower than the fall time.
- Now rotate FORMANT CCW very slowly. And there's our pitch division again.
NOTE: In this video, I think I said 'reset'... don't think of it as a reset a la oscillator sync!
Now we have enough data to make a reasonable explanation. I'll explain it in simple terms (no circuit or too much technical lingo!)
If a previous cycle's pulse collides with the next cycle's pulse
the previous pulse (at this collision point) is falling FASTER than the rise time of the next cycle's pulse, then a division event is instantiated.
And for the observations with no pitch division, we may say:
If a collision happens but the previous cycle's pulse (at this collision point) is falling SLOWER than the rise time of the next cycle's pulse, then there is no pitch division.
So basically, you can only divide pitch when the pulses looks more ramp than saw. This is why you don't get pitch divisions when the BARREL control is past 12 o'clock ish.
But what about these pitch divisions... we need to investigate these a bit further, I think.
Elaboration on the Pitch Division
The pitch divisions can be viewed as 'counts' where the FORMANT pulse train will 'wait' n cycles of the master clock (square output) before firing the next pulse where 'n' is mostly an integer*. The number of counts to wait, 'n,' is incremented +1 when the pulse length exceeds the length of 'n' cycles of the clock.
If the pulse is in length '2.78 cycles' of the master clock, then 'n' is 3 and you'd see a pulse every 3 cycles of the master clock.
If the length of the pulse exceeds 1 cycle of the master clock, then n = 2. You'd see a pulse every 2 cycles of the clock.
When doing experiments with all of the pitch dividing that is occurring, do yourself a quick check from the FORMANT output to the SQUARE output. The SQUARE output is the main core frequency. The FORMANT is after any processing/retriggerring. Put them on a scope at the same time and you'll see about this whole blanking counts thing.
*There is some 'wiggle' between these seemingly 'binary' zones of division and you can spot if you are very careful with knob movement. Here you get lots of interesting Thomas' English Muffin things that happen. I'm guessing this is some 'loose' comparator and when the pulse duration is too long (and of correct polarity), the pulse train cannot retrigger and so it seems like it is waiting master clock pulses. Curiosity! I think the fact that this transition zone ends up being 'semi-snappy' is why the analogue domain is so wonderful.
BTW, the panel notes: utone. I'm guessing that's 'mirco' I don't know. I don't think this term is required and adds unnecessary vocabulary. It's also on a weird spot of the control making it seems like you dime the control to get that 'mode.' I'm assuming this is supposed to describe the pitch division phenomena we are observing. Either way, I'll not be using that term here.
If anyone, including the maker of this VCO, wants to correct me, just contact me (PM33AUD) over at muffs!
Details/Blurbs from Muffs
Think of it as a primary oscillator which feeds the SQUARE out. Then drives a formant generator where the panel controls create interaction between the sections. Sync is soft. Output stage is waveshaped VCA. FM w/ VCA is linear input. Exponential FM via the PITCH jack.
A further note on the linear FM input- it's AC coupled for inter MANGROVE connection. With high FORMANT settings there can be a large DC offset in the FORMANT output so AC made sense for best pitch tracking. For those after DC shifts or linear LFO input there's a tiny solder jumper on the back to go DC. Easy to mod and easy to revert.
The FM INDEX vca is normalled 100% level. Designed to attach an envelope on CV, it drops to min volume when 0v is present, and has exponential response for snappy harmonic braps.