The distinctive thing about the Commodore 64 was its custom chips. And while the VIC-II chip provided competitive graphics, the 64’s secret weapon was the sound interface device, also known as the SID. The Mastermind behind the SID was a young chip designer named Robert Yannes, who went on to found the synthesizer company Ensoniq. But the chips frequently fail today and the problem is getting worse. Here’s why 6581 SID chips can go bad just sitting on a shelf.
10 times as good as the rest and 5 times as good as it needs to be
Yannes was not impressed with the other sound chips available at the time. He said it was clear whoever designed them didn’t know anything about music. He ran out of time and budget to build what he wanted to build. His dream would have been better than the OPL3 in the 16-bit Sound Blaster cards that came out nearly a decade later. What he ended up building was much more modest, essentially an analog synthesizer capable of producing three tones at once. But it could produce more complex waveforms than anything else available at the time. When he was finished, his colleagues said his chip was 10 times better than anything else on the market at the time, and five times better than it needed to be.
But making a sound device that was 10 times better than it needed to be was why the machine had such staying power. In 1989, 7 years after the machine was released, it was still possible for a game to impress you with its theme song. Yes, it was a chiptune sound competing in an era when newer computers could play back sampled instruments. But the SID was the king of chiptunes.
King of chiptunes
The SID is valuable and sought after today because of this. It has capabilities other chips of the era can’t match, so musicians make synthesizers with them. The ability to only play three sounds at a time is limiting, but you solve that by doubling or tripling up. That’s not cheating, there were add-ons for the 64 that included a second SID for music composition and playback. And there were PC sound cards that doubled up on OPL chips.
The problem, many Commodore fans observe, is that the chips are failing at an alarming rate. And in some cases, someone will test a batch of loose chips, mark the good ones, store them, and come back a year or two later and find one or two of them went bad just sitting in a drawer. And the problem isn’t storage, other vintage chips of comparable age stored right next to them are doing just fine.
MOS Technology
The secret of Commodores success was its chip making subsidiary, MOS Technology. When they wanted to design a computer, they designed a custom chipset for it, and then made it themselves. This gave them a big cost advantage, at least early on.
Unfortunately, Commodore wasn’t the best run of companies. And MOS wasn’t run very well as a result. In the 1970s, they had technology no one else had, which reduced their number of defective chips. This increased yield allowed them to severely undercut everyone in price.
The problem was they waited too long to modernize, and they only did it twice. So while other chip companies eventually figured out how to match their yield, they were also figuring out how to make their chips smaller. In effect, MOS was using 1978 technology in 1982 when the C-64 was released. And they used 1980s technology into 1992, while its competitors had more current technology.
Why SID chips are failing: Modernization, or lack of it
The problem was that more modern methods allowed ships to be smaller, which made them cheaper. It also made them run cooler, so they were more reliable. But Irving Gould didn’t understand the business he was in, so he didn’t invest in it. He was more interested in his jet setting lifestyle. Being in the computer business wasn’t much more to him than an excuse to fly to Japan and add to his collection of Japanese art.
The SID chip ran hot. This thing was all about pushing the limits of the technology that was available to him. That’s why it was so great. But running so hot wasn’t great for longevity. And not modernizing it meant the later chips didn’t really run much cooler.
Why SID chips are failing: Chemistry
This will seem irrelevant, but what did Andy Grove major in in college? Yes, that Andy Grove, the legendary CEO of Intel. He majored in chemical engineering. Chip making is a chemical reaction.
And when you talk to former Commodore engineers, they admit that MOS wasn’t as good at chemistry as other companies at the time, like Intel. And that chemistry is starting to cause SID chips to break down just like other earlier MOS chips. Gases used to help the silicon flow during the manufacturing process slowly react with the aluminum in the chip over time. This makes the aluminum corrode. The corrosion explains why the oven baking trick that can sometimes bring Atari chips back from the dead doesn’t generally help SID chips.
Other infamous MOS chips that self destruct
There are infamous examples of unstable MOS chips that self-destruct. The PLA chip in the Commodore 64 is the most notorious example. Those chips started failing in the mid-eighties, when they were only around 3 years old.
The other notorious example was the CPU and the TED chip in the Commodore 264 series of machines. These ill-fated machines never stood a chance. They were designed to solve a problem that solved itself and ended up not having anything to compete with. But they were also unreliable.
Those machines ended up being door prizes that you got for going to a hear a boring sales pitch. Offer people a computer with a suggested retail price of $299, and they’re a bit more willing sit and listen for an hour. So people did. Then they took the computer home, realized it wasn’t a Commodore 64 but rather some other weird incompatible machine from the same company. And then they realized they needed a $200 disk drive and printer to use even the built-in software. Most of them just ended up being boxed back up and put in a closet. I have one such machine with that backstory. I’m not sure which of the two chips failed. But the thing went bad sitting in a box in a dark closet.
And then there are the junk 74LS series chips that Commodore used in the 64 when supplies of glue logic chips from companies like Texas Instruments were low.
The chemistry MOS used in the 6581 SID chip was more stable, because a fair number of those chips have survived. But they are starting to go bad. The later 8580, made using a slightly newer process, seems to fare a bit better. But it uses different voltages so it’s not a drop-in direct replacement. Early C-64 longboards can’t use an 8580 unless you use a stereo SID board or adapter that changes the voltages for it.
More trouble at MOS Technology
In other news that seems irrelevant at first but becomes more relevant once you think about it, the MOS plant was one of the first EPA superfund sites. They weren’t disposing of their industrial waste properly, so the EPA had to come in and intervene in the 1990s. Commodore was out of business by then. But some managers from MOS scraped together enough money to buy the remains of the chipmaking operation and go back into business. They had even managed to eke out a small profit, but then the EPA came knocking. The cost of cleaning up a quarter century of improperly stored toxic waste proved to be a problem. The only other company still in business that had any connection was Allen Bradley, so the EPA went after them as well.
Why we can’t just make new SID chips
A common question among vintage computer enthusiasts is why someone can’t just fire up production again and make a new batch of SID chips. The problem is that the process for the chips is hopelessly obsolete, some of the necessary equipment was probably gone as early as 1994, and even more of it was likely a casualty of the EPA cleanup. Further, the plant sat empty and vacant for a quarter century. It was cleaned out a few years ago, and some relics were still in the building. There are some wild stories about things like Commodore 65 prototypes being thrown away before the people doing the cleanup realized the place contained valuable artifacts.
But it’s not like there was a binder containing all the secrets of making the SID chip that was recovered and could be scanned and sent to TSMC to get production going again.
Modern SID replacements
The SID is a curious mix of analog and digital. This is what makes it extremely difficult to try to re-implement using modern silicon. But it also makes it more difficult to emulate than its contemporaries.
There was huge demand for c64 emulation even in the mid-80s. Emulating the SID chip was by far the most difficult part of that task. Eventually it became doable, but it took 486 class performance to do it reasonably well.
Modern microcontrollers are up to the task of course. So there are some modern SID replacements that plug in to the SID socket and are basically a microcontroller on a PCB emulating the original hardware. Depending on how powerful the microcontroller is, the emulation can range from complete to 80%.
Weighing the modern options
How good you need it to be depends on who you are. Those of us who grew up with these things aren’t getting any younger, and our hearing isn’t getting any better. There are certain songs that expose the difference between, say, the SwinSID and ARMSID. Some of the more ambitious songs just don’t play right on the Swin. If you never play the games that have those songs, you may not be able to tell the difference, especially with aging ears that went to too many loud concerts. Maybe that’s just my friends and me.
The better and more expensive options, such as the ARMSID, usually sound pretty close, and they implement all of the features. It may not be perfect, but then again, there was variance in how the originals sound. They do sound cleaner, but that’s a side effect of the way the originals were made. Emulating that noise is trickier.
None of the modern replacement options is necessarily perfect, but for many uses, they are good enough, and much better than no option. It’s likely that better options will eventually surface, but in the meantime, emulating it with a microcontroller gives us something.
David Farquhar is a computer security professional, entrepreneur, and author. He has written professionally about computers since 1991, so he was writing about retro computers when they were still new. He has been working in IT professionally since 1994 and has specialized in vulnerability management since 2013. He holds Security+ and CISSP certifications. Today he blogs five times a week, mostly about retro computers and retro gaming covering the time period from 1975 to 2000.