[sdiy] Understanding 80s Synth Architectures

[Brian Willoughby]

On Feb 4, 2022, at 02:22, ackolonges fds via Synth-diy <synth-diy at synth-diy.org> wrote:
> Hi Everyone,
> 
> As the local SDIYer I sometimes get asked to try and fix synths from the 80s like various Rolands, Korgs, Oberheims etc. and I generally just try to pinpoint the rough area of the issue and replace logic chips until the issue is resolved...
> 
> Most of these synths have a CPU connected to everything via a parallel address bus and a parallel data bus, with all sorts of glue logic chips doing various things. They also usually just have 1 DAC, time-domain multiplexed to all of the different parameters via more logic chips. 
> 
> Obviously this is very different to the modern ways that microcontrollers and DACs are used in synths, and since I wasn't around in the 80s, these older architectures are very foreign to me. To aid in my troubleshooting efforts, I would love to better understand the details of how these architectures work, so I was wondering if anyone on here would be able to point me to any resources that could explain these types of systems to me, be they websites, articles, or books?
> 
> Thanks a lot for any advice you might have.

This is a great question. I'm old enough to have learned this stuff in the 80s, so the shared context at that time made things a lot easier. I've often wondered, "How does someone starting today make sense of the layers of technology that former generations only had to learn one-at-a-time, simply because they were invented one at a time?"

My hunch is that the best answer is to learn the various layers in an order similar to someone who might have learned them in the 80s. That sure seems like an easier task that trying to learn everything at once.

In other words, I couldn't have learned all of this at once because it hadn't all been invented yet. I think that made it a lot easier for me, because I could absorb one group of concepts before the new concepts came along.

For me, the path was like this: First, I learned simple analog electronics like guitar pedals. Authors like Craig Anderton, who wrote synth manuals (E-Mu EMax) as well as his own book, "Electronic Projects for Musicians," were technical enough to be practical, but accessible to those who did not have a formal training. It takes a lot of skill to write in a way that bridges those worlds. Second, I learned digital electronics when my uncle made an Apple II clone for me from scratch, and I started designing audio peripherals for it. Authors like Hal Chamberlin made digital audio approachable. His book, "Musical Applications for Microprocessors" actually described the fast fourier transform in a way that finally made sense to me (not that it's related to 80s synths, necessarily, but it sure shows his skill). Third, seeing as how I knew that I really enjoyed analog and digital electronics design, I spent four years studying for a BSEE degree. Surprisingly, even though I studied for my degree in the late 80s, my school's labs seemed to be based on 70s technology, making me very glad that I'd taught myself the "modern" components used in 80s arcade games and 80s hybrid analog/digital polysynths.

That's a long journey. So, a more practical approach might be as follows: I recommend starting with a specific 80s synth. Grab the schematics and then grab the data sheets for every chip on the board. Study the data sheets for every chip until you understand how it works, what part of the overall synth design is handled by that one chip, and how it interfaces to the rest of the circuit. In particular, a lot of the "why is it designed like this?" should be answered by the details in the data sheets. Usually, things are done a particular way because there isn't any other choice. Granted, the biggest learning curve will be the MPU or CPU, and you'll have to dig in to the firmware and peripherals to understand how that works and "why" things are connected the way they are.

I'll say one thing contrary to your comments: I don't think that modern designs are that different in principle. Sure, most CPU chips these days have onboard ROMs (using Flash instead of EPROM), onboard SRAM, onboard peripherals, and often onboard DAC or ADC. It would be possible to redesign 80s synths with the same block diagram. The only difference would be fewer chips.

The challenge is that if you just want to repair something, it might not be worth the effort to learn everything about how it works. If you want to design a MIDI upgrade board or otherwise add features or design replacement parts, then the time spent learning everything is obviously worth it.

In some sense, though, once you learn the entire system for one 80s synth, you can probably get a good idea of how the rest work without spending so much time. The recommended reading - Prophet 5 Technical Manual - might be the best way to thoroughly understand one example so the subsequent synths seem more familiar.

Brian Willoughby