Theo Verelst Original Synth page

Original because is the synth I learned to play chords and scales and not too complex songs on, which I built somewhere halfway highschool (15 years or so ago). Still some decently valid thinking in there. In fact it still works, crancking it up with a stabelized 12V supply is all that is needed for most keys to work (two or three have probably gathered up too much dust).

I want to remark here that until some time ago I didn't think much about the machine, and not too many people from university or music scene have seen it at all, let alone in action, except for some people in the time I made it, simply because starting from my purchase of a Korg Poly 800 around the end of highschool, I gathered up a professional array of synth equipment like a DX7, a DW8000, Rev-7, Rx-15, bit-one, tg500, tx802, and some more, so I simply didn't feel much need for it. Only lately I've been somewhat forced by the arising notion that people in the graphics project I was in had such a high mind about their own computer engineering capabilities, and apearently a fairly low of mine to generate some serious correction. And what is better than circuitry that actually works?
Apart from that I rediscovered the joy of electronics design, and the fact that it is one of those areas where even today, and that is what I doubted before, without much of an updated view on part availability, it may pay extremely well to construct one's own devices and circuitry.

Outline

The keyboard had a mixed analog digital disign, with parts associated with on the one hand electronic organs, and on the other hand with analog synthesizers. A top octave (2pow(1/12 spaced) divider ic with high frequency clock generator and followed by 4 stage dividers per half note generate 5 octaves of fully polyphonic square waves. A simple mixing rail gathers the singals of the keys being pressed trough coupling resistors, feeds them into an opamp based mixing amplifier annex vca annex filter, which also receives an envelope triggered by a mixing rail envelope follower , and which is fed through another (adjustable but fixed) LP filter to the audio output. Add an adjustable sine wave lfo, small range frequency modulation input to the top oscilator and a link to the vca voltage input, and voila, a 'hybrid' synth as they called some quite expensive korg synth in the days following. Would have loved to be able to design prophets 24bD/oct VCF's.

Mechanics

The mechanincs, as always in my home projects at the time where based on wood, because I could process it, with a plastic plated hardboard tilted front plate holding a set of pots and switches I'd gotten from everything between old tv's and dump stores, and a bar graph led display, one of the other projects I could at some point afford the chip for. I bought a keyboard for te incredible amount of Fl80,= (about $40) when I could afford it, and was never sorry: they should fit metal framed light but solid and non sticky keys like that on modern synths. The DW8000 comes closest.
Unfortunately, the keyboard had to be fitted with contacts, costing another guilder a piece, which for the 4 octave + 1 = 49 keys was too steep. I though a lot about it, and came up with the following. Each key has a piece of fairly strong and bendable hard-cored flatcable bent to form a long legged U lying under it, with legs of about 5 cm, suspended from to rods pressed together containing all these tricked springs kept at the right distance and in place during construction by thick cellotape. The springy ends have a little piece of silver wire soldered unto it (keeping the remaining silver part hidden from tin additives during the soldering), which rest against a global mixing rail perpendicular to it, running over the length of the keyboard, suspended by simple nails (...). The key ends, after the turning point, in other words they go up when a key is pressed, are fitted with a small piece of rubber which presses the piece of silver wire down (away) from the mixing rail in rest position. When a key is pressed, the flatcable bends outward, the silver ended last part hits the mixing rail with fixed strenght, and a note sounds. The nice part being that no matter how hard a key is pressed, the contacts never get bent, it took me quite some thinking and experiments, but appearently I did my job well, most keys still recently worked even without undusting. It has a funny sound to it, a nice 'clang' when the mixing rail is hit. Since the action is only a few millimeters, and the contact force limited, the spring function is (as yet) not temporal.

Some circuits

The top clock generator is a standard 2 inverting cmos (4032? or so) gate generator, with as extension a diode and resistor in series couples with the 'hot' end of the capacitor, to make it modulable in frequency, as ling as the oscilator isn't forced out of operation, that is.
The top divider is a chip (undoubtedly Radio Shack, since they had component characteristics and application notes on the back of the package) with 12 dividers in the range of about 200 to 400 which frequency divide the top frequency into the semitones of one octave.
The per semitone octave dividers are simply 4020 dual 4 stage asynchronous binary dividers.

The mix amplifier/vca/filter unit
I could add distortion unit, because the vca function is achieved by using a FET in non-saturated mode, more or less like this, from memory: 

                       --------------------------
                       |                   |    |
                       |                  | |   |        -------
 Notes    keys         |           |\     | | -----      |     |
    ------  /          |           | \    | | -----     \ /    |
 >--|    |-/ o| Mix    |    .5Vs|--|+ \    |    |     ------   |
    ------    |        |  -----    |   >--------------|    |------->
    ------  / |    -------|   |----|- /               ------   | Audio
 >--|    |-/ o|    |      ----- |  | /  741                    | out
    ------    |------   ----|   |  |/                        -----
    ------  / |     |   |   |   |                            -----
 >--|    |-/ o|     -----  | | | |                             |
    ------    |       |    | | | |                             |
              .       |    | | | |                            ---
     ...      .       |     |   |
                      |     -----
                      |       |
                      O      ---
                   Control
                   in (appr 0-.5Vs)

Probably I'm forgetting some voltage converting R's and some additional C's, and there may have been another buffer amp as well. I'm sure there are more pots in the fet circuit as well, to adjust non-linear distortion figures, or even make the opamp clamp (nice on fifths), this is a long time ago, and the time left playing I spent on stuff like this. Unfortunately, I never realized I could make steeper filters simply by cascading, and that resonance can be achived by feeding back two or more sections.

envelope detector, adsr generator
(d and r common rate)
low frequencey generator

Seriously thought through and feasible extensions