Added a video demo of the AS3372 analog signal processor, and the VCV Rack interface for it. The 3372 is Alfa Rpar’s clone of the Curtis CEM3372, notable for being the filter in some classic polysynths: Oberheim Xpander, Sequential Prophet 600 & T8, Chroma Polaris and others. Alfa Rpar’s design added to the CEM3372 with a voltage controlled panning block. So in addition to the 4-pole analog filter you’ve got 6 (six?!) VCAs onboard. Fun!
Thought this might be interesting to share: analog 3340 VCO being manipulated with a series of sync pulses. The phase between the positive and negative sync pulses are modulated very slightly by an LFO. I won’t go into details as it dives pretty deep on how the 3340 handles sync. Needless to say it produces a pretty interesting effect. On the scope it looks somewhat like PWM on the saw wave.
Having too much fun making little bleepy bloopy things with this setup of just a few hardware oscillators and a filter. A bit of a wierd drone thing is this; Rack-wise it uses an LFO and sequencer to tweak pulse width on a VCO. Hardware-wise it’s got two analog VCOs and a VCF. The analog oscillators are at about the same pitch, with one of them modulating the filter. The filter output is routed back to PWM the VCO, creating a bit of a feedback loop since that VCO is modulating the filter.
Unrelated to the above and soon to be Rack-related, I’m wrapping up hardware design on a dual-VCO / VCF board. Once that board is designed it’ll need a VCV Rack module for a frontend. The board itself is an AS3394 (“synth-on-a-chip” as used in the Sequential Six Trak and many other synths) and an SSI2130 (modern VCO with thru-zero FM and other cool stuff). Should be quite the synth voice once it’s working.
Time to go back to circuit design 'cause this is about as good as it gets with me at the helm. Would love to see what an actual musician could do with this setup. All analog here: transistor noise source, VCOs, VCF, VCAs. That noise source uses a single transistor VCA which honestly kind of sucks. Pops every time it turns on, so take that as a win to use it to set tempo. LOL-- make your weakness your advantage!
Not so much a demo, but here’s what’s coming up module + hardware-wise. This is a combo is the AS3394 (“synth-on-a-chip”) and SSI2130 (modern synth VCO). The '3394 is a clone of the CEM3394, as used in the Sequential Six Trak and many other synths of the 80s. It’s a single VCO voice, with VCF and VCAs onboard. Being only a single VCO it’s a little thin. Hence the addition of the SSI2130. Introduced a few years back, the 2130 is a VCO with a lot of onboard features. Notably thru-zero FM, but it’s also got a mixer onboard for the pulse/saw/tri waves, a couple sync modes, and is really stable.
This is the most complex board I’ve done and uses two 8-channel DACs, some a number of switches for patching, and an AS3364 for additional VCAs. It will plug into the backplane in the same way as the other boards I’ve done (previous in this thread). One output is the synth voice (two VCOs going through the VCF) and the second output is just the SSI2130 VCO.
There will likely be a round or two or hardware debugging / revisions (fingers crossed not more than two!). While both the '2130 and '3394 are close to datasheet implementations, it’s by no means trivial getting it all wired up. Block diagram:
Some interesting features:
Some other cool stuff as well. And it’s all analog audio path with analog modulation options. Digital control elements from VCV Rack. The good stuff. Next up is designing getting a layout for the above. UI isn’t exactly my forte, and with getting 2x VCOs, VCF, and modulation all presented it’s going to be a decent size layout.
So I lifted @henk.lasschuit 's patch from the One Voice thread. Ported to the Zoxnoxious analog synth with some rash modifications. The patch uses two sequences: so I patched each one to a separate VCO. Then made The Big Deal here to have the audio path VCO sync’d to a master VCO, master VCO not on the audio path. The audio path is from the sync’d VCO, through a VCF. What’s most interesting is the clock enables/disables sync and is also used to modulate the sync method on the Zoxnoxious 3340 VCO via Sync Phase. The 3340 has a positive and negative sync, which isn’t quite hard/soft sync despite that I labelled it as such on the module.
Apologies to Henk since this is quite an abuse of his original. His patch is pretty cool and you ought to checkout the original. It was a great starting point to showoff some of the sync functionality with the Zoxnoxioius '3340 module.
Just really went off the deep end getting noise out of this. Per the YouTube description, it’s “audio simulation of an exorcism of buzzing robotic bees”.
Synth geek patch info: it’s just a VCO and VCF/VCA. The VCO modulates filter cutoff & VCA panning. One channel of the VCF/VCA output modules the VCO pulse width, frequency, and provides a sync signal to the VCO. LFOs are there to vary the modulation depth.
Super great stuff! Congrats @brer_rabbit! Do you plan to make an open source DIY project of this with kits to build?
I’m trying to figure out the direction from here. The project is open source. Kicad files, Raspberry Pi app code, VCV Rack modules. All at GitHub - brer-rabbit/zoxnoxious: Analog synth / software interface . Some demo-y stuff on the YouTube channel. I’d like to get something into other people’s hands. I’m pretty sure anyone on this board has greater musical talent than myself, I’d like to see what someone can do with it. That said, there seems a very long road to getting there. My roadmap is something like:
That’s the hardware side of things. Software-wise I’d like to put a Rack module together that’d be a utility module for the Raspberry Pi. So you could actually do stuff like connect to a wifi point, get system updates, utility like stuff since the thing runs headless. Basically start to get the environment around it user friendly. And documentation. That’s sorely lacking. I wrote datasheet on the AS3340 module. I need something similar on the others and other docs.
But yeah- overall goal would be to get it in other people’s hands. It’s been well over 18 months of development getting this far. I’d like the project to reach some usable end state eventually!
Here’s a demo of the AS3394’s Waveselect modulation. Background: the 3394, a clone of the same chip used in many 1980s analog synthesizers, uses voltage on a single pin to select between tri, saw, and tri + saw waveforms. How about modulating this pin at audio rate? Here’s a demo of exactly that.
Interesting? Just noise? Curious to hear what you think.
This is one of the more unique (and possibly questionable!) functions of this board. It’s got some interesting functions that I plan to do videos of down the road. Definitely the most complicated board I’ve put together, this thing is a beast.
The “v0.2” version of this board is out to JLCPCB for prototype right now. Should be back within two weeks. I’m expecting this version to have the SSI2130 fully working. What’s that mean? A modern analog VCO with thru-zero FM, sync options, output mixer with VCAs, blah blah blah. That’s integrated with the AS3394 demo’d previously, the VCO/VCF/VCA chip. Combined it’ll be a pretty cool synth voice. And combined it’s way more complicated of a board than I’ve done previously. Geek numbers: 192 nets, 20 opamps, 16 DAC lines, 8 switches… this thing is a beast. Not a lot of free space on that 80x80 mm board.
Some demos coming up once it’s available. Til then, here’s some AS3394 synth fun:
One shortcoming I recently addressed: the Zoxnoxious synth was only able to address 27 channels of control voltages. I doubled this to 54 channels a few weeks back. Net result is I’m able to fully populate the backplane with six voice cards. The next problem is how to demo. I’ve zero musical talent (as exemplified by this thread!). Thank you thank you to the people putting out algorithmic composer modules! That said, if anyone has a patch that I could port to the Zoxnoxious synth, I’d be more than happy to record a demo.
A couple cool things in the works:
Whoa, this is wild. An analog modular synth patch-able trough the vcvrack interface for total recall. Just when you think the world has run out of ideas 
Slightly qualified: the world has run out of good ideas. I’m here to see the bad ideas have their day in the sun 
Finished build of the Zoxnoxious Z5524 board, and it works! Some tweaks would be nice but nothing I’m in a hurry to address.
Full demo coming soon but here’s a sneak peek at some of the FM stuff. Just a scope shot here for now, but this is using Fundamental’s Fade with an LFO to crossfade between modulator waveforms and/or crossfade between carrier waveforms. Nothing fancy. Except it’s a friggin SSI2130 VCO and AS3394 synth-on-a-chip being driven by VCV Rack! 
Tuning: something everyone who uses VCV Rack take for granted. I’m getting to that- finishing up a static autotune method for the Zoxnoxious synth. On a tune request (yes, an actual MIDI tune request, 0xF6 message!) the synth will tune all installed boards in parallel. Static tuning, creating lookup correction tables. I’ve got the 3340 VCO tuned so far. With that done it should be short order to get the other cards tuned. Beyond the VCOs, I should even be able to tune the VCFs to 1 v/octave following the same model.
I consider the tuning is “good enough”, at least it’s way better than it was previously. This is what you get with a 12 bit DAC; though I’ve got a couple things that will improve it a bit down the road. It’s analog, the drift that remains is just character!
Demo shows two VCOs with the same control voltage applied to both. Note the “F” frequency in the scope shot on each one.
Hoping this gets beyond the vaporware phase 'cause there’s some really cool features here!
back in the day I calibrated the pulse width of the VCOs, too/
One more great project name !
That is very well justified & I’ve been considering it. Basically, prevent the user from letting the pulse go to DC? I’d like to get to that.
I’d think pulse width may be something that may only need to be calibrated at startup. My assumption being that it isn’t so temperature sensitive as exponential converters.
Makes sense. I think we did it to save a trimmer and the calibration time. We were trimming for 50%, btw.
