*** FriedOrange's ***
(Commodore) PET Project

Part 1: Taking stock

1 June 2024

Before we can even think about repairs or restoration: what exactly do we have here? Unfortunately, I don’t know the history of this machine—the eBay seller said it had come with something else he’d bought and knew nothing about it—but I’ll try to infer as much as possible from what I can see.

P.E.T. or C.B.M.?

I’ve been calling this computer a PET, but as you can see in the photos, it’s clearly labelled CBM Model 4016. The label on the rear further identifies it as a 4016-N. What’s going on?

In the USA and Canada, this machine would have been a PET 4016. Commodore used the CBM trademark for the business-oriented 8000-series (perhaps hoping to steal some of IBM’s thunder) and also for PETs in some export markets. I had believed Australia wasn’t one of those markets, judging by these original advertisements, but maybe there was some randomness involved. Commodore used quite a few variant labels over the years, after all.

The -N suffix to the full model number signifies the Normal or Graphic keyboard, which depicts graphics symbols on the fronts of the keys. See how—unlike the Business keyboard, used in the 8000-series and 2001-, 3000- and 4000-series machines with the -B suffix—no characters are duplicated anywhere on the keyboard; to type numbers and even the full stop/decimal point, you need to use the numeric keypad:

This is one of the early 4000-series models, probably made in late 1980 or early 1981, with the 9-inch monitor and so-called Dynamic motherboard (named for the fact it uses dynamic RAM chips, as opposed to the original PET 2001’s static RAMs which made for simpler circuitry but were more expensive). This makes it identical to the earlier 2001-N models (3000-series in export markets) with the exception of updated firmware. In 1981, the 4000-series would receive the 12-inch monitor and Universal motherboard, both shared with the 8000-series.

4016 indicates that this machine has a 40-column text display and originally came with 16kB of RAM; a previous owner has boosted it up to the maximum of 32kB (note the second row of DRAM chips, which are of a different brand from the originals):

The metal parts of the computer have begun to rust quite significantly, but at least it could have been worse: the upper cabinet (i.e. the off-white painted parts, except the rear cover to the monitor housing) is made from injection-moulded plastic, unlike earlier PETs which were entirely sheet metal.

Fire hazard!

It's a good thing I thought to check inside the fuse holder on the back, because I found a 5A fuse there. This is massively overrated for this machine; I wouldn’t be surprised if it could actually go up in flames before the fuse blows! I’ll replace the fuse with one of the correct rating—reportedly an 800mA slow-blow—before trying to power it up for any significant length of time.

Of course, the next question is: what caused the original fuse to blow in the first place? And, was that fault ever fixed? It’s too early for me to tell.

Franken-Firmware

The PET’s operating system and Microsoft BASIC programming language are contained in five ROM chips. There is also a character generator ROM, which holds the font for the display (but isn’t actually accessible to the CPU), and an additional two Option ROM sockets. In this machine, one of those sockets is occupied by a mystery EPROM! How exciting. My PET contains the following ROMs; one of them looks unmarked in the photo, but I was able to make it out with a bright light at an oblique angle:

These ROMs are what we would expect from a 4000-series PET. It’s nice to see that I have the updated version— sometimes referred to in the literature as “BASIC 4.1”—that fixed some bugs in the original release.

There is one oddity, though: the screen editor ROM, 901498-01, might not be the original. According to the archive on zimmers.net, this ROM is for machines with the CRTC chip (found only on the Universal motherboard) and a 50Hz display. This machine should instead have 901447-29, since it is an earlier model with no CRTC (and therefore a 60Hz display). The date code, indicating the 19^th week of 1981, is also an outlier here: it is the only chip on the whole board that was manufactured after 1980 (besides the EPROM and aftermarket RAM upgrades).

This evidence suggests that the original editor ROM was replaced at some point. I wasn’t sure if I could expect it to work properly with this replacement, but it seems to work fine in VICE, with two noticeable differences compared to the original ROM: the good news is that it now makes an audible chirp on boot-up; the bad news is that the internal clock runs too fast. I guess it doesn’t matter much.

The EPROM present in one of the Option ROM sockets, simply labelled “TK 4.0”, is interesting. I thought I would have to wait until I could get the PET working before I could see its contents, but then I noticed dumps of one Toolkit 4.0 on the aforementioned zimmers.net archive. It apparently adds some utility commands to BASIC, similar to the Programmer’s Aid cartridge for the VIC-20. This information has very likely spoiled the mystery early, but at least I may still get to see if it matches any of the alternate versions online.

Modifications

As mentioned above, this machine has had its memory maxed out by a previous owner. But from the two switches installed haphazardly in the front of the case, we can immediately see that some other mods have been made to the PET. Inside, a loudspeaker has also been added; this was an unofficial mod, developed so that games could have sound (naturally), that later became a standard feature in subsequent PET/CBM models:

But what about those two switches? One is a DPDT toggle and the other is a SPDT momentary switch. Inside, we can see wires going off to a mass of electrical tape, bundled around the existing wires that go up to the monitor. Let’s unwrap the tape and see what’s going on here...

That’s right, those switches are wired up to nothing at all. It was already disappointing enough that holes were drilled right in the nameplate on the front of the case, where the textured plastic makes it impossible to invisibly patch them (why couldn’t they have been discreetly added to the lower cabinet, like the speaker mute switch?), but the fact that the switches do nothing only adds insult to injury! Oh well; what can you do. I’m still elated that I finally own this amazing piece of history.

What's next?

The motherboard is so dirty that it’s hard to see what’s going on there. This makes component identification, soldering and desoldering more difficult; plus, physical damage could be hiding under the dust. I think my first real task, then, should be to give the PET a thorough cleaning.