BLIPPER is a non-linear distortion device. It is like a simple spike generator, however it is well-behaved above the cut-off point. (It is a littlelike VCO where the pulse width is tied to the VCO F to maintain the mark time while the space varies: this may be be a little formant-like or reed-like.)
Multiple Synched LFO is a ring of four LFOs which each can soft-synch the next in line. This allows complex heart-beat like rhythms where you have several chambers that beat or don't beat at different rates in response to each other. Complex rhythms result. Because four LFOs is prone to lockup, they also have a little anti-lockup timeout on them.
Diode Chain Non-Linear Waveshaper is a CV or audio waveshaper. Initial idea was for compensating for unlinearity on freq CV of 8038s, but could also be used for waveshaping audio and for shaping breath control signal. (Probably could be simplified to only outer two pots.)
Vactrol Response Simulator is an envelope follower
circuit designed to simulate much of the behaviour of the famous Vactrol optoisolator.
It tends to gate very low level inputs.
Here is some idea of the responses in SPICE for .1, .5, 1, 2.5 and 10 V inputs:
Simple Vactrol Response Simulator is a much simpler version
of the above circuit. Unlike a normal envelope follower which has the same response regardless
of surrounding input, this has a slow attack response for soft signals and a quicker attack
response for louder signals, progressively coming in.
It tends to compress low level inputs.
Here is some idea of the responses in SPICE for .5, 1, 5 V inputs:
FSK Synthesis is not a schematic but a description of a synthesis technique. I have simulated this in SynthEdit and it works as expected, generating tones with the fundamental, two formants, and a scattering of low-level accidental harmonics.
Improvement? I can see scope for a more sophisticated version of FSK, for the scenario where the slave oscillator is fixed pitch. The idea is that as the master oscillator frequency goes down, progressively more FSK sections are introduced, each bringing in a lower formant. So a high note might have two sections (threfore two formants) like the Oscilliscope chart above, but a low note might have five or ten sections, and therefore possible formants. Under the current FSK system, each section takes a fixed period of the wave cycle independent of frequency. But a more frequency-dependent sectioning would still allow enough time for the section to have an audible effect, but the higher formants would become less the lower the note. This matches constant-waveshape "synthetic" sounds more, perhaps.
I guess this could be done by using an LM3914 instead of the comparitor, but VCA-ing the driving signal in reverse of the master oscillator frequency, somehow. And perhaps some system could be used to guarantee the ordering of formant frequencies, so that high notes get high formants while low notes get all of them. The LM 3914 (or 3915?) is pushed to the limits when used as a waveshaping sequencer at high frequencies, it seem (people don't cascade them for audio) however if the rate of transitions between output sections decreases with input frequency, they shouldn't be over taxed.
This is potentially a nice system, because it gives a direct correlation between knobs and harmonics/formants without needing sine wave mixers and so on. What is lost is that there is no way to micro-adjust the mix between formants by adjusting the comparitor. But you could have two adjacent sections at the same pulse width. Just thinking things through, but there would also probably be more accidental harmonics generated for the transision sections, which would be useful: there would also perhaps be some component corresponding to the number of sections active in the voice: if there were 8 sections for a note, you might expect some harmonic activity two 8ves above the main note.
(I see Henry Thomas's XR2066 uses FSK to do a kind of sync: a short sync pulse drives the frequency crazy high, and it returns in some different phase, but at a fixed position for each keyboard pitch. So this gives a sync effect but based on jumping forwards somewhere unknown rather than jumping to zero or high or low. I think this would give more of a soft sync effect than a hard sync effect though, because you wouldn't necessarily jump forward to the same place each time...it seems like a kind of FM really. I guess you could implement a non-intrusive hard synch on an existing no-synch oscillator by having a short FM pulse like Henry Thomas suggests, but switching the pulse off as soon as there is a zero crossing (or high or low) at the saw output. This would mean that the fundamental of the slave was out of phase with the master by some fairly constant small amount, varying with frequency. )
(On the issue of sync, another posting elsewhere spoke of applying little nugde pulses to bring two oscillators into lock. )
DIY Synth Construction: driven by an open source community is a blog entry that includes some material on my ETI 4600 project.
ETI 4600 is an old Flickr set of photos of the ETI 4600 project as at Christmas 2006. It is much more pretty now that I have tracked down the last of the known Cliff K1 and K2 knobs in the universe.
Timeline of Modular Synths outside America is a little timeline for historical interest.
Improving bad breath and wind control is a blog item on wind controllers, and why they are so bad with normal VSTs, and how to fix it.
Here is some recent shots, the first with all the new Cliff knobs installed, just like the EMS VCS III etc used. Actually, I only have completely finished about half the modules but the majority of knobs.
For good info on the ETI 4600, see Brian Martin's page (another Sydney lifer on this thing!)
Here are the schematics:
