One other issue about "tight" builds that rely on very slim modules: they're a bitch to work with live unless you have toothpicks for fingers. So, yeah...if the 914 is going to see lots of live tweaking (and it sounds like it might), you're better off with the bigger module. Now, let's see what sort of gibberish I can kick out HERE... (pause for /u to fire up the grid and go berserk)
.......OK. Now, this is a serious AF drone machine:
You've got TWO voices here, actually. Most of the top row is the first, and the second is hiding down there in the lower right.
Top row: I switched your main drone VCO up. Way up. The Mindphaser is actually more related to something like a Buchla 259 than a typical VCO, which means you've got two oscillators in a configuration where one's the "source" and the other's the "modulator", but of course that's like the "serving suggestion" on a frozen dinner. The reality will be that you'll be able to use this in a huge number of configurations, all of which will allow you to not only generate pitch, but also lots of different methods of timbral generation. This sort of oscillator is PERFECT for elaborate waveshapes to feed into the 914 via the Quad VC Mixer. Plus, there's a surprise about that mixer in a bit...
Anyway, the other "odd thing" there is that single fader. WTF is it there for? Ahhhh...it's there because Moog 914s ALWAYS have those two I/O jacks, even if it's Uli making 'em. And why those two jacks exist at either end of the 914 is so that you can insert an attenuator for manual feedback control, allowing you to use the 914 as a RESONANT filterbank, sort of akin to the Serge Resonant EQ. The fader is the attenuator in question, natch...and I put a fader there so that you can have better incremental control AND have a quick visual feedback on your foldback level in the feedback path. Patch it from the 914 OUT to its IN (which would be the fader's OUT being sent there), and you've got it. After that, there's an interesting VCF/VCA combo that can work that way...or as a lowpass gate for yet another "sneak West Coast feature". Plus, you could also make this a bandpass gate or highpass gate, since the VCF in there is a multimode. This also lets you have some global timbral control over the drone voice. And the last thing up top is a Bastl Ciao!, which also contains a dual stereo mixer in addition to your headphone preamp and isolated 1/4" outs. Why a mixer? You'll see...
Bottom row: The Frequency Central Infinitely Maybe gives you a mixer for incoming CVs as well as the usual noise generator for sample and hold and random gates. So you can actually S&H a composite CV signal there, not just merely one source. This is also your main clock, but if you need to override that with the Neutron's, that's cool too. Then the next thing is an omsonic melodic contour generator...a "complex sample-and-hold" that gives you controllable quasi-melodic patterns. But then, I wasn't content there, so I also added a Penrose quantizer after that, which you can ping with, say, the random gates from the I.M.. Feed it an LFO curve and have the gates lock the new value, and you can assign those scalar values to whatever pitches you want. The Penrose has the sole drawback of being a kit module, but if you don't have suitable solder-fu, you can probably find a builder.
Now, the mixer surprise. It's actually the next module, a CVable quadrature LFO, which outputs four identical sines on the same frequency...but the outputs rotate phase by 90 degrees. Sound-wise, this doesn't affect anything, but when this is in LFO mode, you get the outputs working in a "cascading" phase rotation, and when this is hooked up with each phase tap going to each VCA CV, you can then "riff" the Mindphaser's outputs to create yet MORE timbral variation. Then there's the quad LFO...but after that is a little something from DPW Design that gives you a CV mixer, an adder, a comparator, and attenuverters so that you can polarize incoming signals...both inputs are, yep, attenuverters. And as for the comparator...that device lets you send a gate out when you exceed a certain voltage threshold; in this case, the A input is the "threshold" and the B becomes the "signal", so that when A exceeds B, you get a gate output which you can use anywhere, and when B > A, there's no gate.
VCAs are there after this so that you can have level control over two modulation signals or, if needed, you can "rob" one for audio, although the drone "voice" really isn't set up like that. Its single VCA in the Optocore should suffice for CV over final audio levels. Most of the time, though, these will see modulation level control action. Next is your dual ADSR...which actually works pretty well here, since you only have two VCF/VCA outputs from the voicing. Then Voice #2, all Dreadbox. The Hysteria stayed, and I got it a friend in the Eudemonia VCF/mix/VCA module. That gives you a basic single oscillator voice in just 20 hp.
Then there's the last two modules. Both of these are Frequency Central Stasis Leaks, which give you your choice of reverb, tap delay, or chorus. There's also one for each "voice", plus these take your mono audio signals and "stereoizes" them to add some spatial aspects. So, how those work is that each stereo output gets fed to the stereo inputs on the Ciao! and then you have mixing control between Voice 1 (drone) and 2 (basslines, melodics, percussives, etc etc), and this feeds that to the headphone pre AND the 1/4" outs.
Not too shabby, I think. The most spendy thing in there is the Mindphaser, but given what it can do as opposed to a pair of basic VCOs, the price is very justified. And the rest of the module complement doesn't ever have a price tag above $240, and most of them are actually down in the $50-150 range. Hell, I myself would dig playing THIS build!