This article explain my test procedures and equipment for the Commodore 64. I also take a deeper look at the different versions of the C64 and explain why you need to apply slightly different acceptance criteria when testing these machines.

A bit of background

I buy most of the computers that I sell through listings on online marketplaces, auctions and the likes. Many sales listings lack information on the condition. Sometimes, the seller doesn’t have the means and knowledge to test. In other cases, the seller thinks the items are working because the power LED turns or because the machine worked in 1992 when it was put on the loft.

For clarity, none of this provides any bearing on the actual, current condition. A quick count of my latest 50 untested Commodore 64-purchases shows that 15 were defective, albeit including minor defects such as defective keys, dirty power switches and blown fuses. Many of these machines were thought to be working when they were listed, though.

As long as the price is right, I prefer to buy machines sold as either untested or defective, as this allows me to add value through repairing or by selling them with a warranty. But in order to provide a warranty, I obviously need to test the machines first.

Test equipment

My go-to test equipment when testing a C64 is listed below:

• A known good aftermarket power supply
• A known good video cable, upscaler and monitor
• Diagnostics cartridge
• Diagnostics test harness
• A cartridge with a single-part demo with a known tune which allows me to test the SID chip
• My own keyboard tester

Acceptance criteria

As a starting point, any C64 that lands on my desk, has to pass the following tests before being listed with a warranty:

• The machine must produce a normal video output.
• The machine must PASS all Diagnostics Cartridge test procedures with the external parts of the test harness connected.
• Every single key must work when pressed with a normal amount of power.
• The machine must be able to provide normal output on all channels when playing a SID tune.

No test procedure is flawless, and there are certain types of faults that I don’t test for:

• As discussed in my article Testing floppy drives, I don’t perform burn in testing. If the machine has an error which only shows up after 30 minutes, the above tests won’t detect it. This is an accepted risk. After having sold hundreds of C64s, I haven’t seen more than 1-2 examples of errors that didn’t show up right away.
• The diagnostics cartridge is know to produces some (but few) false positives and quite a few false negatives.
• I don’t test the video output of the RF modulator. If the RF modulator has failed completely, the video output won’t work either, and I will notice it. But if only the components producing the RF output have failed, I won’t detect it. I haven’t head of more than 1-2 examples of this, though.

All the test procedures above are simple, and 90% of the machines will either fail or pass based on the clear criteria above. And then there is the other 10%, where things are a bit more complicated.

Differences in behavior

Although the Commodore 64 widely is regarded as the highest-selling single computer model of all time, not all machines work the same despite having passed Commodore’s quality assurance process, and some bugs are more common with certain revisions.

Some of the variation stems from Commodores many design updates between the first release in 1982 and the final production run in 1994.

If we dive into the electronics inside, an even bigger contributor to the variation is that Commodore developed minimum seven different motherboards for the Commodore 64, provided we only count the assembly numbers and ignore the revisions within each assembly number. If we add revisions, include various PCB manufacturers and variation in parts, I am about to say that every Commodore 64 is unique

Mainboard variants

• The first breadbins came with the now rare and highly error-prone assembly 326298 or, in Europe, the PAL-only KU-14194HB boards. I note that KU-14194HB isn’t an assembly number (94HB is the flame retardant rating). But the name is the only designation found on a range of boards produced around 1982/1983 without any other means of identification.
• As sales numbers started picking up during 1983, Commodore released its perhaps most common board, the assembly 250407.
• The next major release came in 1984 with the also common assembly 250425, which differs from its predecessor by having a simplified video compartment.
• It took more than 2 years before the next major hardware update. With the introduction of the Commodore 64C in august 1986 came the fifth board in the family, the assembly 250466, which was the last longboard. It came with a simplified RAM compartment, but is otherwise largely similar to the 250425. It only remained in production for around one year, and during that period, Commodore 64Cs continued producing computers with 250425 boards, presumably from a remaining stock from when the breadbins were produced.
• Finally, in 1987, Commodore launched the most radical revision, the “shortboard” assembly 250469. Assembly 250469 was produced from 1987 and until Commodore went bankrupt in 1994. It was released in 5 different revisions, including a revision B, where Commodore integrated the color RAM into the super PLA.

Learn How to detect the mainboard from outside without opening the case.

Related products

Outside the main line of Commodore 64 computers, Commodore released two related products.

• The C64GS was a game console without keyboard and connectors for peripherals, but technically, it was based on a completely standard assembly 250469 board, which however came without the components needed to drive the rear expansion ports, keyboard connector and obviously with a vertical cartridge connector instead of the angled connector in the original C64.
• The SX-64 came with a completely separate motherboard design, namely the assembly 251102, which is the seventh entry on the list, although I would consider it a remote cousin rather than a sibling.

I have to add that the service manual mentions an assembly 250441 as well, although it remains an open question whether it was released. I have so far not been able to confirm its existence.

But here is the thing: With each assembly number came some slight differences in behavior, and within each revision, you will find significant variation in the choice of components, all leading to a great deal of variation in the behavior of the machines.

Known variants

Determining whether a C64 works as intended or not may in some cases depend on the context. There are some known behaviors that would have been considered bugs by 1993, but made it past Commodore’s internal QA when the machine was sold. I have listed some examples below:

Symptom	Not a bug if…	Could be / is a bug if…
Blurry video output	VIC II revisions R1 and R2	VIC II revisions 3+
The SID chip sound different / has varying volume levels	SID revisions R2 and R3	SID revisions 6581R4AR / R4 and 8580R5
Samples (think “Stay a while, stay forever”) are played at a very low volume	SID is an 8580	SID is any revision of 6581
Easyflash 3 etc. doesn’t work	Assembly 326298	Any other revision than 326298
Diagnostics cartridge fails CIA test	Board has a 8251 CIA-chip instead of 6526	The board is equipped with 6526 only
Diagnostics cartridge reporting the control ports as not working	You can confirm that they are working (remember to test analogue input)	You can’t confirm that they are working.
Restore key won’t always trigger an NMI interrupt, i.e. “doesn’t always work”	Assembly 250407	Any later board than 250407

The inevitable consequence is that a behavior, which would be considered normal in some cases, would be considered an anomaly in other cases.

There probably are many more items to add to the list above (feel free to contribute in the comments).

Attention points for certain models

Some variants of C64 are more prone to certain types of hardware failure. I have collected some examples below:

Variant	How to dentify this variant	Attention points
Breadbins in general	By looking at the case	Have a much higher error ratio than earlier variants. Pay particular attention to the SID chip and the keyboard.
Breadbins with type 2 keyboards	The space bar has a more square shape than normally.	Check that all keys work with only a light touch.
Commodore 64C (early)	The PETSCII symbols are on the front of the keycaps.	Have a slightly increased error ratio relative to later variants.
Commodore 64C (later)	The PETSCII symbols are on top of the keycaps.	Power switches tend to get stuck. If the power switch is tight, spray contact cleaner into it and wiggle it until it either moves freely or can’t be moved, in which case you should repair or replace it.
Commodore 64G	The label says “C64G”. The case is a breadbin case, but has the same color as a C64C.	Check the diode as they tend to fail often. Same issue with power switch as with the C64C.