E&MM interviews Dave Rossum, the man behind E-mu designs
Dave Rossum, E-mu Systems designer, talks about his involvement in the Electro-music industry.
For the last 10 years, Dave Rossum has played an active part in the design of some of the most important developments in Electronic Music technology. Recently, he was in this country promoting the release of his latest design, the Drumulator.
"My schooling is in Biology, I was a graduate student studying the Biochemistry of Ribosomes, at the University of California. My advisor walked into the lab one day and said, 'Have you ever seen a Moog synthesiser, they're unpacking one in the music room, do you want to go and help them?' So I went to the Moog room and sure enough they were unpacking a model 12. By the end of the day I was teaching everybody how to use it even though I'd never seen it before. It just kind of clicked with me so I spent the whole quarter not doing biology but teaching synthesisers and by the end I was just so intrigued by the whole thing. We got schematics and I figured out how the thing worked. My training in Physics and Computer Science gave me a basic understanding of what a transistor did and plenty of maths to go with it. So I was able to pick it up pretty easily, checking up on a few things in the library.
Over the summer of '71 a group of six of us from Caltech and the University of California got together and decided to build a synthesiser. We started out with some very primitive uni-junction type oscillators and so on, but proceeded fairly quickly to the type of VCO design you see today, with dual-transistor, temperature compensated exponential convertor and op-amp integrator with a discharge transistor. We used the universal filter approach, the state variable, back then with a Moog ladder type network. Transconductance amplifiers weren't available then, it was the end of that summer that RCA announced the release of the CA3080. We used the 3060 which was similar, but pretty non-linear, for our amplifiers. My first patent was in fact how to make the 3080 into a Voltage Controlled Filter.
We built the first unit as a sort of modular system similar to the ARP2600 with a single front panel, pre-patched behind the panel, but which could be externally patched from the front. We learned a lesson from that, every patch you did you wanted one more oscillator or a different kind of filter or another input to the mixer, it just made you realise how much you wanted a true modular system. We had two custom front panels made so we built two of them which we then sold."
"When we went into business in November '72 we decided to go for the high end of the market and designed a modular system which is still in production today. It's built like a Sherman tank, front panels ¼" thick, walnut cabinet made by my brother and has very expensive pots and stuff which make it last forever. They go into universities and don't fall apart even although the students seem to play them with hammers!
We really tried to get the circuit quality better than anybody else's. There's a wonderful story which happened during the development. We had as reference the ARP 2500 manual which had a thing called the multi-mode resonator. We wanted a standard of excellence, basing our designs on the best of each system available, be it Moog's filter, ARP's multi-mode resonator or whatever and studied the spec sheets on each of these things. I worked for weeks getting our State-Variable filter to have a Q of 500 at 20KHz, which requires a 10MHz gain bandwidth, learning a tremendous amount about filters, high frequency amplifiers, how to get a VCA to run to 10MHz and all these wonderful things. We laid out the circuit on the double sided PCB and had to recompensate the whole thing for trace capacitances from one side to the other, since at those kind of frequencies they are very significant!
It ended up that the last 'tweek' on the E-mu modular system active filter is that you take a piece of superstrip, which is stuck on, and clip it to a point where the filter goes 'bing'. Then you look at it on the scope, running at about 18 or 19KHz, watching it ring down, then you clip it again and again until just before it goes into oscillation. So it's 'tweeked-up' to the limit to get to the specifications we read on the ARP spec sheets. About 2 years later I actually had my hands on the multi-mode resonator for the first time and it didn't come anywhere near those specifications! Still it was a wonderful way to really push yourself and learn electronics. Once you know something can be done you'll figure out a way to do it. That's really the company attitude towards electronics and engineering.
I've been, up until very recently, really the only engineer in the company. Scott Wedge, our president, is capable of digital design but I do the current hardware and software design.
I had some formal courses in software, even in Biology you use computer simulations, and I was always interested in such things at a machine level. At college we had an IBM1800 which was a process control computer which you could play with at 'bit' levels. I would spend days there writing programs in assembly or machine code language. I then used a 360 model 20 which was a little less powerful than an 8080 microprocessor but still a good computer back in those days. It was simply used for listing out punched card decks, but it was a multi-purpose computer. So we worked out how to punch an entire operating system in assembler! It was all on punched cards, we didn't have a disk drive in those days. That was one project at college I learned a lot of my software training from as well as studying the more sophisticated computer techniques.
Both the Emulator and the Drumulator use the Z80, which is an extremely good processor because of its interrupt structure. You can go into one routine, interrupt to another and still interrupt with higher priority routines. In the Drumulator it does everything except the audio path, since it's not fast enough to do digital audio with the required fidelity. However, we're doing a lot with that processor, which makes the Drumulator so cheap."
"I met. Ron Dow, an engineer wanting to start a company to make his design for a VCA chip (which became the SSM2000), but at the time he wanted someone to help finance the integration. We could have scraped together the money but unfortunately it was on a 24V process and our modular system had ±15V supply rails as standard. He then went off to EXAR to get a quote on the integration and they offered him a job because he had done such a good custom integration design! He took the job but kept working on his designs and eventually found a company called Solid State Music (making computer boards and selling surplus parts to hobbyists) who were willing to put his design into chip form. Two years after my first meeting I went to SSM to buy some parts and ran into Ron who told me about his new design using a 36V process, with all of the original problems solved. So we then started designing chips together, Amplifiers, Oscillators, Filters, ADSRs and came up with an amazing amount of technology for what was really second generation linear design.
When Dave Smith, a good friend at the time, started to design the Prophet he asked our help as consultants on a royalty basis. So our help with the design using SSM chips really got Sequential Circuits off the ground. When CEM (Curtis ElectroMusic) brought out their temperature coefficient resistorless oscillator design, which SSM had not developed yet, SCI changed to CEM chips. CEM was probably the best thing to happen to SSM who have just released eight new products."
"After the May '80 AES, at which we had seen the Publison and the Fairlight, people seemed very interested mainly in the sampling capabilities. I knew that the Fairlight had individual memories for each one of the sound generators, so the idea that we could share memory and cut the cost down led to the Emulator. A group of us got together one afternoon and figured out basically what the Emulator would do. So I crawled into the back room, was chained to the bench and started work.
Two months later we had pretty well proved that the sound quality from companding was acceptable. The next trick was managing to get the bus bandwidth fast enough so that you could get all the sounds out. When you pitch shift up you get a pretty high bandwidth. For example, sampling at 30KHz, then shifting up an octave, giving 60KHz, in eight channels is about ½MHz which is a lot on a microprocessor bus. I didn't want to have a separate sound bus and processor bus, so getting that system to work was a little tricky. We still use a 2.5MHz Z80."
"As far as future products are concerned, I don't know what we are going to do next, we have a lot of ideas but I still have a month or so of work with the Drumulator. The software has been developed but there are bound to be production headaches.
We're still producing modular systems but the sales come in little waves. A few years ago we went for 6 months without selling anything, around the time when the Fairlight and Synclavier were being introduced. When people saw them, realised they weren't God's gift to men but had their own particular sound, they came back and started to buy modular systems.
As far as the future is concerned I think everything will be digital. By the year 2050 you'll probably have something identical to the Prophet 5 but completely digital, a total simulation. The big thing you'll see at the maturity of digital technology, which I think is a long way away, is the Filter knob coming back! I mean the filter is a real nice thing, you've got to admit. The ear just knows what that's going to do, because it works along the same lines as the ear does. You can't do that with digital, you turn a knob on almost any of the synthesis algorithms right now and it doesn't quite do as 'nice' a thing as a filter does. The beauty of digital filters is of course that you can program them to do anything. Unfortunately, they have to be done to at least 24 bits of linear precision or go into some kind of floating-point technology to do it. It won't be until the digital filter becomes easy, which is about 10 years off, that a strong analogue filter IC will become obsolete. The digital oscillator, however, is coming of age - essentially that's all the Emulator is, its a fancy digital oscillator. As times goes on we'll find other interesting things to do digitally along with the analog stuff.
The next big step in electronic music is the human element, getting the expressiveness into the controllers, the sequencers and the sound in many ways. After all that's really what music is all about."
For further details of products from E-mu Systems, Inc., contact Syco Systems Ltd., (Contact Details).