Commodore 64 motherboard revisions

Last Updated on March 26, 2024 by Dave Farquhar

Commodore made the C-64 for about 11 years, so it’s probably no surprise they went through several Commodore 64 motherboard revisions during that time. Collectors enjoy the challenge of trying to get a machine with each type and revision of board, and knowing the characteristics of each board can help someone puzzle out the history of a machine, such as whether it had been repaired in the past.

Commodore made a lot of changes to the 64’s outward appearance over the years, but they made a lot of changes internally as well. This helped them drive the price down from $595 to $149 over the course of about three years.

Caveats with Commodore 64 motherboard revisions

Commodore 64 motherboard revisions
This motherboard, assembly 250407, dates to 1983, and it was the board in use for most of the C-64’s peak popularity. That makes it the most common of the various Commodore 64 motherboard revisions.

Sometimes you will find a board someplace it shouldn’t be. By that I mean it’s paired up with parts from a different generation. In these cases, provenance is nice to have. Once when I bought a boxed C-64, I found repair records inside the box along with it. To some people, this may give a bit of legitimacy to whatever anomalies they may find in the system.

Also in the early 1990s, Commodore offered a trade-in program called “A C-64 for $64.” You could send Commodore a broken 64 along with a check for $64, and Commodore sent a refurbished unit in return. This can explain some 64s with mismatched parts, such as old boards in newer cases, or boards with date codes years apart. An oddball 64 with paperwork proving it came from Commodore or from a dealer during this period is the only way to distinguish it from something a hobbyist cobbled together by combining multiple broken 64s. C-64s sold in 1993-94 through Software Hut and the Grapevine Group, with the paperwork to prove it, would be especially interesting because they likely represent the last machines Commodore made.

Silver label boards

Commodore sold somewhere between 150,000 and 300,000 C-64s in 1982, which was a lot of machines for its time but isn’t a lot by C-64 standards. These early machines also had fairly high defect rates, which led to a lot of these boards being replaced. So there were fewer of these boards to begin with, and a fair number didn’t survive.

326298 Rev A – C. 1982. Schematic # 326106.

These boards turned up in silver label C-64s, and later versions of it turned up in the first rainbow-label breadbins. These boards are notable for their 5-pin video ports. They are scarce and very collectible. This generation of boards has compatibility issues, except, notably, the CP/M cartridge. Making matters worse, the video quality on these boards was subpar due to the values of some of the resistors in the video circuit. I have heard stories from former Commodore service center technicians that when they swapped a newer board in for a 32698, they would remove any socketed chips and discard the board.

KU-14194HB

This board is poorly documented, and KU-14194HB is a designation that Commodore used on other things, so that’s not the proper name for this board, but it’s the only prominent marking on it. This board appeared in European silver labels. It was an interim board, dating to 1982. It had characteristics in common with the 326298 and with the later 250407. Notably, it has the 8-pin video port we associate with later boards. A lot of sites say this board was PAL-only, but some NTSC variants have turned up in North America. This board is rare.

Breadbin boards

By 1983, Commodore had its act together in a big way, and Commodore may have sold as many as 8 million machines from 1983 to 1985. It was one of these machines that inspired me to go into IT, personally, and the same is true of hundreds of thousands of other GenXers. These boards are anything but rare, but they’re treasured.

250407 – 1983. Schematic # 251138

The consensus seems to be this is the most common C-64 motherboard, and it was the board Commodore was building at the peak of the 64’s popularity. It has 8-pin video. Even though cost reduction was one of its design goals, it’s much improved over the early 1982 boards. The PLA is nearest the serial port, and the SID is by the VIC-II. It had three revisions, A-C. If you or your friends had C-64s in its heyday, there’s a very good chance your 64 had this board, or someone you know did.

250425 – 1984. Schematic # 251469

This board is also very common. Commodore sometimes referred to this as the 64B. This board is more integrated and cost-reduced than previous boards. The most visible difference between this and earlier boards is that the VIC-II isn’t in a metal can anymore. Notably, the PLA switched spots with the SID on this board. The PLA is now by the VIC-II, and the SID by the serial port. So if you’re swapping parts between multiple boards, pay attention to where the chips go so you don’t blow up a SID by plugging it into the PLA socket. This board is desirable because of its excellent video output quality. If you want the classic C-64 experience, this is a good board to look for.

A note about the video quality: This admittedly has more to do with the VIC-II chips in production at the time than it does with the board. If someone swapped an earlier chip into one of these boards, the video quality probably would degrade. And yes, this does happen. A dealer could have swapped the chip in the 80s as part of a repair because an older chip was all they had. Even more likely, some previous owner could have combined parts from multiple broken units to get one good unit. To get the best experience out of this board, look at the four-digit date code on the VIC-II chip, near the center of the board. If the date code contains an 82 or 83, it’s a mismatch.

250441 – 1985. Schematic # 251469

Commodore made reference to this board in its service manuals, sometimes calling it the 64B2, but this board never went into production. I’ve never found anyone who says they have one. Maybe a pre-production example will turn up someday. The changes were minor, involving changes to a few diodes and moving some resistors into an integrated package.

Commodore 64C boards

If you ever wondered how the streamlined, modernized C-64 from 1986 got its “64C” designation, wonder no more. The name came from the motherboard inside it. The 1984 model was the 64B, and the 1986 model was the next letter up.

250466 – 1986. Schematic # 252278

This board, known as the 64C long board, turns up in some late breadbins and early 64Cs. It still has the 6581 SID. The fuse is horizontal behind the cartridge port if you want to spot one without opening up the machine. The most visible change on this board from its predecessors is the presence of two RAM chips, down from eight, leaving lots of empty space in the lower left. This is a highly regarded board. It has lots of parts commonality with the older boards but with the simplicity of fewer RAM chips. It also has excellent video output quality, like its immediate predecessor.

The same caveats about earlier VIC chips apply to this board. As some Commodore hobbyists have pointed out to me, if you put a mid 80s chip in a 1982 board, the 82 board looks good too. And an 82 or 83 VIC-II chip in a mid 80s board will make its output look like an earlier board.

250469 – 1987. Schematic # 252312

This board is known as the 64C short board, or the 64E board. There’s no date on the board but it seems to have first appeared in 1987. This is a smaller, further cost-reduced version with lots of changes, including a much higher degree of integration. The dreaded PLA was replaced with a Super PLA, which is much more reliable. It has the lowest parts count of any C-64 board Commodore produced. The 64E board used the 8580 SID and 8500 CPU. It had four revisions in total. Rev. 3 and 4 were the earliest. The later Rev. A and Rev. B integrated the color RAM into the Super PLA. Rev. B is rare in the USA.

Some of these boards are yellow and some are green. This board isn’t 100% compatible with its immediate predecessors but has fewer compatibility issues than the first boards from 1982. The fuse is vertical to the cartridge port if you want to identify this board without opening the machine. You can ID Amiga 500 motherboards without opening the machine too.

Modern boards

Sometime around 2015, efforts to re-implement the Commodore 64 and produce new motherboards with new technology become practical. These newer boards solve a number of problems with trying to use a vintage C-64 three decades after its heyday, most notably the scarce and terrible power supplies, the PLA chip that likes to self-destruct, and memory chips that are well past their life expectancy.

Some people enjoy repairing dead C-64s. Others would rather just open up a C-64 that shows a black screen at power-on, pry off the major chips, and plug them into a new board. That’s the target audience for these.

C64 Reloaded

This board resembles what Commodore probably would have eventually built if they had been able to integrate more tightly. It integrates the PLA and all of the glue logic into modern chips while including ZIF sockets for original MOS CIA, VIC-II, SID, and 6510/8500 CPU chips.

Ultimate 64

The Ultimate 64 re-implements all of the C-64 logic and chips onto an FPGA while providing sockets for one or two SID chips for more accurate sound than is possible to implement via FPGA. Had Commodore kept pace with other chipmakers throughout the 1980s, they would have been able to implement the C-64 on a single chip (also known as an SoC) sometime in the late 90s. Except for the addition of the 1541 Ultimate, this hypothetical 64 would have resembled the Ultimate 64.

Implementing the 64 on an SoC would have increased profitability and also would have allowed them to continue reducing the 64’s price. At $149 the C-64 was a hard sell in 1991 or 1992, but it’s interesting to consider what might have been if Commodore could have sold the 64 for $50 and still turned a profit.

But is FPGA still retro? Is it Commodore?

There’s a lot of debate about FPGA and whether it’s retro, but if nothing else, an FPGA implementation of a C-64 board makes an interesting might-have-been. FPGA is also letting us build modern replacements for chips like the 6526 CIA and 6510 CPU last made in 1992. So you can use a blended approach, with modern replacements on a vintage board, if that’s what makes you happy. There’s no wrong way to like retro.

A note about mismatched motherboards

Every vintage motherboard tells a story. Commodore 64 motherboards are no exception to that. A lot of C-64s had a long service life, and it took a repair or two to make them last that long. Some C-64s, especially the earlier models, had a troubled life as Commodore struggled to deal with suddenly having 1/3 of the entire personal computer market all to itself. It’s hard to go from also-ran to #1 without some quality control issues along the way.

Sometimes you find a vintage board where all the date codes are within a few weeks of each other and everything matches. And sometimes you find a board where the date codes are all over the place.

Is a pristine unit worth more? Probably, especially if you can prove it’s pristine. But if you don’t know what chips are normally in sockets and don’t know how to check the date codes on the 64’s motherboard, I don’t recommend you push it.

Dealer swaps

When a system went in for repair, either in or out of warranty, the technician swapped any bad chips on the motherboard with whatever compatible chip was in stock. Ideally the chip was newer than the unit. But it wasn’t always. Truth be told, the replacement parts might not have been unused parts. When there was a backlog, a tech might swap the whole motherboard, then repair swapped-out boards to replenish the inventory of spares.

Of course, a loose motherboard won’t have a service sticker on it. But if the board is in a case and has service stickers on it, I preserve the service sticker as best I can. It’s part of the machine’s history.

Factory swaps

When a unit got returned directly to Commodore, its technicians would fix whatever was wrong, and then resell those machines through liquidators. Refurbished C-64s turned up even in 1985, though the inventory never lasted long at that stage.

In the late 80s and early 90s, you could send a broken C-64 and $64, and Commodore would send you a rebuilt unit to replace it. Commodore then fixed the broken machines and sold those for $64 plus a trade-in, or sold them to liquidators.

And then, of course, there was the C-64’s long decline. The C-64 remained popular far longer than anyone imagined it would. No one thought in 1982 that Commodore would still be selling these units 10 years later. But at that time, Commodore was struggling. The profit margins on the 64 weren’t good, even at $149, and Commodore was bleeding cash at an alarming rate. They were building machines with whatever parts they could find.

Some may dispute me on this but I don’t believe Commodore sold inventory with used parts as new machines. I think if they had unused chips and motherboards in inventory, regardless of age, they assembled and shipped them. When it was a mismash of parts, it went to deep discounters like Aldi. When the mix looked current, it went into the regular retail channel. And when it was a mix of new and previously used parts, it went to a liquidator. Commodore sold way too many refurbished machines in the early 90s for me to believe they were selling used parts as new.

Hobbyist swaps

And of course, along the way, any number of hobbyists may have dabbled in repairing Commodore equipment. I fixed a few C-64s for extra money when I was a teenager in the early 90s. It was better than working fast food. I even conducted some of my transactions in that fast-food parking lot, though I always did the repairs at home, of course.

As the Commodore repair shops closed, we either had to learn to repair our own machines or move on. Invariably, the people who moved on sold or gave their machines to people like me who hadn’t. We would swap parts around among machines to get working units. And honestly, we probably didn’t pay attention to whether the replacement parts were the proper vintage. Even if we did, a mismatched part may have been all we had. And an early, lower quality VIC-II chip is better than a completely broken one.

Some shade tree mechanics did good work. Some of my early work was sub-par. But the machine worked at the time, even if I would shake my head if I encountered the machine again at my current age.

If you found this post informative or helpful, please share it!