Dave Vosh's interview with Serge

When I started the serge system back in '73, the guiding principle for the work was to make available to the musician as many of the possibilities inherent in electronic hardware, even though they might not seem to be of traditional musical interest. An example of this is our "DUAL VARIABLE SLOPE GENERATOR". Not only does this module allow the generation of simple attack-release envelopes, but it can also function as a vert accurate voltage controllable portamento device (with separate up and down rates.), as envelope followers, or as wide range trapezoidal audio and sub-audio oscillators. Not only that, but the module is very good for generating sub-harmonic tones. Now, as you know, the sub-harmonic series does not fall into the tempered musical scale and therefore, perhaps, it wasn't sought out by traditionally minded synthesizer designers...

Anyway, such a module is extremely multi-purpose as you can see. Back in '73 it was quite a change from the established practice: modular synthesizers were made up of many separate purpose modules tailored each to do their own function. In my design work, however, I saw that technology would allow multi-functionality quite easily: so why not? And why not have functions with no apparent traditional musical purpose?

The benefits and drawbacks of this approach have become very clear over the years. First the drawback: essentially, the multipurpose modules are harder to learn to play gracefully. Anyone who takes a look at our faceplates can see why: there are more than 5 or 6 times as many inputs, outputs and controls on our things than on typical machines. These are of course needed to achieve multi-programmability. Thus, it is up to the musician to sort these possibilities out for himself,.and discover their use. More work. But, eventually, much more rewarding for the user since possibilities don't dry up as easily as they do on typical single purpose modular systems...

The benefits are twofold: first, cost: in traditional synthesizer designs a musician has to buy many more modules which do only their thing in order to get the same wide range of functions as on the serge. But really, the major benefit is the fact that our approach offers a far greater palette of effects: in my experience, having heard the work of tens of musicians using the serge, the music is truly electronic and not, in some case, even remotely similar to traditional musical instruments (as was the goal of most commercially available synthesizer machines: to duplicate the traditional concepts of musical sound) This has been' the greatest satisfaction to me personally, since I am the most interested in new musics.

To recapitulate: the basic approach of the serge is to let electronics itself provide possibly useful musical functions; not the traditional way, which is to take, as Moog did with the ADSR envelope generator, a musical analytical concept: in Moog's case, the attack and decay characteristic of traditional instruments; and to make the electronics bend to achieve that rather limited concept. It's all esthetics! Free electronics from pre-conceptions! If the electronics allow a timing or wave-forming possibility with no apparent musical purpose, then the serge will incorporate this possibility as an available feature. It's up to creative musicians to find a musical use; and they do just that...

Back in '73 modular synthesizers cost in the tens of thousands non-inflated dollars, and therefore could only be afforded by commercial studios or schools. The problem was accessibility of the equipment for people: most musicians I knew didn't want to foresee a life in school just to have access to the tools needed for making their own music. This is why the serge began (and will probably continue) being available in kit form. Our desire was to make the machines available to people, so that individuals could have them in personal studios. Why not make music like graphic artists do their work: in private studios? Nowadays, we've kept the philosophy. Typically, an exceptionally well rounded serge can be purchased in kit form for about the price of a good motorcycle. The dream is thousand of personal studios across the country doing creative personal new musics: why not?

1. Guidelines for circuit design? In many cases, someone will say: I think such and such module would be nice to have in the system. At that point, the wheels start turning: can such a circuit he done? is it truly needed: i.e. will it possibly lead to interesting musical possibilities? etc. If the answer is yes to these questions, then I start doodling on several sheets of gridded paper, preferably light green. It's at this point that the process gets interesting. Usually, the basic circuit function gets sketched out in a matter of minutes. In this, I apply the standard guidelines which all dependable circuit designers apply: I shoot for the maximum accuracy, minimum parts count, & lowest parts, labor and testing costs. It gets interesting, however, just after I have designed the basic circuit: I ask myself the following question: what else can be done - i.e., in addition to the main function of the circuit, granted that such and such circuit uses such and such concepts, etc... I always begin this procedure of letting my imagination loose just after having designed the main part of a circuit. Very often, I have come up with quite unexpected results, in fact with circuits that do far more than the original idea.

for example: Just a couple weeks ago I was told by a Belgian musician that it would be nice if our Keyboard -Sequencer had quantized outputs. I agreed, & set to work. In two days, I had the basic circuit & had it prototyped. Happily, I hadn't made too many mistakes, and the prototype worked exactly as expected. At that Point, I saw several possibilities to simplify the design. Further, I had been mulling over the possibility that the A to D converters be not fed directly to the output D to As, as in a standard Quantizer, but rather, through a read only memory partitioned to read out (i.e. map out) a large number of separate musical scales in response to the input, (i.e.: a ramp at the input might result in the pentatonic scale, the major scale, the dorian mode, the Indonesian scales of pelog slendro, the nine-step scale, etc. etc.)

Alas, the complexities of the resulting circuit-cum-EPROM was too great for the design goal of low cost & small space. Thus, I had to scratch the idea. However, I realized in the process that by switching some of the lines to the D-to-A, I could easily switch over between the chromatic and three other scales: the whole-tone scale, a three note "scale" (the augmented triad scale), and a six note scale (c, c#, e, f, g#, a). Thus I built in two gate inputs to the quantizer, so that the user could select between those scales. The quantizer, by the way, does quantization for eight channels (8 in to 8 out) in 6 milliseconds. With the second prototype, playing with this unit, I saw that one of the nicest effects of quantization is doing what in musical theory is called "figuration": passages of notes around a central note. So at this point, I decided that each of the channels should have two inputs, one which would have the entire 5 octave range; the other, only a 7 semitone range. Inputting a triangle wave into this latter input causes a figuration which does semitone runs up and down around the tonal center determined by the main input: a very nice effect. At this point, I felt that I had done as much is I needed to do for this module. What has resulted, however, is a PC board which can be packaged In three separate versions: one, as a built-in into our Keyboard Sequencer to allow easy semi-tone tuning (the original purpose); two: a straightforward 7 in 7 out quantizer module occupying two inches on our standard PANELS, with the two special scale selecting gate inputs; and a third, full-blown version occupying 3 or four inches of Panel space: with figuration inputs as well. Thus a serge module is born. As you can see, no commercial manufacturer could afford such horseplay with a circuit. Because such manufacturers have to design a mass producible item which, to warrant development costs, have to selI a minimum large number of units. We can afford such modules, because we are happy to fill the needs of the few who have the imagination to use our modules, even though only few of the modules get sold... (Our standardized packaging system helps, in this respect.)

To continue on your question concerning guidelines:

Quite often, we have re-designed our modules to expand their function or refine it. Thus it was with our NTO which replaced a workhorse VCO designed in '73. The guidelines for the NTO was that I would never have to re-design a VCO: it had to be the ultimate in stability & accuracy, and also offer expanded possibilities. Our previous VCO had a voltage controllable wave-shape output: in the NTO, I designed an even more elaborate variable wave output.

We built a VC portamento into the NTO. We put in a highly regulated "oven" to keep the temperature sensitive elements always at a constant temperature: the only way, ultimately, to ensure the best accuracy & stability. We put in a VCeable linear FM input, because our oscillators tracked so well that a modified (single quadrant) Chowning FM synthesis method was possible due to the accuracy of the tracking between two oscillators.

The DSG was a redesign of the old serge modules: the NEGATIVE SLEW and the POSITIVE SLEW generators; here the guideline was to design a module which did everything that these previous serge modules did, and more. These previous modules were the cornerstones of the old system: a shot in the dark, at the time, since no other designer had made modules like them. In a nutshell, the negative slew was exponentially VCeable as to its slewing rate (the "decay" segment of an envelope) in the downward direction, whereas in the positive direction, any input was followed instantaneously. A little thought will reveal that such a circuit can form the basis for nice functions such as rectifier w/ decay rate, positive going sawtooth wave generator, etc. The POSITIVE SLEW was the complementary function, able to acquire an input instantaneously in the negative direction, but VCeable slew limited in the positive. For many reasons, I added several auxiliary functions (in addition to the straight input output function) Among these were a pulse input to trigger one cycle of saw; a sustain input: which caused one cycle of saw & held it high at five volts (our VC standard voltage) until the gate input returned to zero; and several pulse outputs. Thus the entire circuit had many possibilities: audio & low frequency saw-wave generator; triggered saw wave generator; positive going portamento; Vceable pulse monostable. And in combination with the NEGATIVE SLEW, the two formed a "Vceable envelope generator; a Vceable portamento generator; sub harmonic generation, etc. etc. A shot in the dark at the time, since I didn't know if anyone would be able to learn how to use such highly electronic (i.e. non apparently musical) functions, let alone put them to good use. Time proved that my fears were unwarranted, & many serge users will swear by these old modules: one of my initial customers (in 75) had not a single VCO in his fairly large system, but about 16 such Negative & positive slew generators (plus filters, VCAs, etc.) which he used in lieu of oscillators. In 76, I developed a new "kernel" circuit for accurate transconductance functions. This kernel, it became apparent, permitted a much simpler realization of the NEGATIVE/POSITIVE SLEW functions. Thus I quickly re-designed the functions of these older modules into one single module appropriately called the DUAL UNIVERSAL SLOPE GENERATOR (there are two such circuits) which performs all of the functions of the older circuits in a smaller package. By the way, I think, that the above description of modules describes also quite well our attitude: i.e., let the electronics suggest possible musical uses, rather than vice-versa: where someone says: I think I'll design something that sounds like an oboe. Oboes & violins exist already: there's no need to duplicate their sound: what does a transistor sound like when overdriven, etc. etc. that's much more interesting.

The VCM is also an evolution of an older module. The older module, in this case, exists even now. The triple WAVESHAPER was specifically designed to permit a wide variety of VCeable wave-shaping of the basic sawtooth waves from the POSITIVE and NEGATIVE SLEWS. Thus a user could input a saw into a wave-shaper and obtain a smoothed waveform much like a sine which, under voltage control as two separate VC inputs, could transmute into a warped saw, or a progressively clipped triangle (similar in sound to a PW wave, but with greatly attenuated high frequencies). The triple WAVESHAPE? uses the non-linear characteristics of semiconductors as the basic tool to warp an input wave. The WAVE MULTIPLIERS, the new version, expands greatly upon those early concepts: and here too, the guideline was to explode the possibilities: by hook, crook or moxie to create new timbres through exploiting non-linearities. Thus the WAVE MULTIPLIERS were conceived to create three basic forms of wave warping, multiplied n-fold through VC-ability. In this case, however, the new module merely expands the palette of timbral transformations, rather than render obsolete the older WAVESHAPERS. Both modules have their distinctive sound capabilities.

3. "Even the most beautiful woman serves only to frighten the fish when she jumps into the water"; thus neither fully digital, nor fully analog synthesizers really do it. What does it is the musician utilizing the equipment. As of now, as Don Lancaster of much fame among do-it-ourselvers has said: 97% of the applications for micros has yet to be evolved. This is not the case for analog systems: I'd say that about 85% has been done (and some of it, if I can say without false modesty & some pride indeed, has been to 99.99% of the state of the possible art by us...) in the analog world. Thus the analog world offers an incredibly well developed and expressive set of instruments, whereas the digital offers an embryonic potential needing much creative thought and work on all of our part. To my mind, very few musicians have explored the true frontiers that digital technology has to offer: microtonalities and accurate temporal relation ships. Thus our D to A will offer tuneability to 3 cent (hundredth of semitone, and just about the threshold of human pitch discrimination. This obviously can't be done with just a 12 bit DAC (don't ask about how we'll do it: that's for now a secret/..) The characteristics of our digital system will be 1) real-time expressivity: i.e. parameters will be as in a VC system, fully accessible to the performer during a performance: it won't be locked into pre-programmed units, but accessible to the pressure of a foot, finger, etc... 2) extreme finesse of tuning (exploiting the accuracy of the NTO and PCOs, and foremost, ease of programming: we're basing the operating system on the Fig-FORTH model, allied to total user extensibility: i.e. the musician, if he wants to, can work with his or her computer to organize functional relationships involving automated algorithms, etc... use of matrixes, etc... Eventually, it seems to me that the language LISP'S fundamental operations and concepts may be the kernel for a new musical theory, Music is queen of the arts, much as Math is queen of the sciences: because both are based on entirely abstract concepts. Even the most doped-up guitar player improvising loosely something that sounds vaguely like some vague tune vaguely perceived over the jukebox makes sounds which resonate with precision in the Great Hall of Numbers up there in the Javanese Heaven: or is it vice versa: does the Great Hall resonate, and our vague puppet plays accordingly?

Thus we are proceeding with some slowness into the computer field: because what is to be done is to form part of the yet unknown theory of new music. My first interest will be to develop and work with our new multi-channel A/D and D/A (with the aforementioned accuracy) and see where that leads us. Thereafter, we'll keep on fishing (not jumping into the water): we have some plans for computers in signal modification (re Vocoder like equipment) and equalization.

4. Did I make a list of the new modules which are done or almost done? Here's the list: the Quantizer; spin-offs of our Keyboard-Sequencer (several different hut related modules), a triple slope compander/expander module; an 8 channel VCeable light dimmer; a new envelope follower based on a new design with an extremely wide dynamic range (better than 70 dbs), a new equal power panner, a truly perfected Ring Modulator (yes there was a need for this module!), a companded reverb with VC-eable effective decay rate; a transformer coupled low noise wide dynamic range preamp; a. frequency shifter, an analog delay; a warbling vocoder (new type of vocoder); the 32 channel D/A + 8 channel A/D; the Synthaphone; a new Phaser. About half are nearly in, production; the other modules are closely following. As you will note, many of these modules are very traditional synthesizer modules: I must say that I have always been interested in making things as good as possible: thus it is that we have spent quite some time making new versions of older modules!

back