Random Macintosh Classic issues: (Not) The solution (Part 2)

A simple fix brought the 1991 Macintosh Classic back to life.

Or did it?

The Classic operated nearly flawlessly from the time I restored it back in 2020 until recently, when it randomly decided to begin acting up. After having the machine running without problems for hours, it randomly crashed to a striped pattern. Rebooting the machine was only briefly successful before it crashed again. Sometimes the machine would just crash with a Sad Mac icon (displaying error code 000000F 0000000D), while other times it would go “haywire” with all kinds of sounds and displaying strange patterns.

I immediately suspected a loose memory expansion card, which has caused similar problems in the past. After reinstalling the card, ensuring it was tight, I powered on the machine to find it made no difference.

The next step was to look at voltages. After ordering a multimeter (to replace my old dodgy one), I checked all voltages on the logic board molex connector and they all were well within range — 5.04 volts on the 5-volt rail, and 12.16 on the 12-volt rail. The power supply/analog board was working properly, so the issue was most likely somewhere on the logic board. Fun.

Voltages on the Macintosh Classic seemed fine, with 12.16 volts on the 12-volt rail and 5.04 volts on the 5-volt rail.

TinkerDifferent users suggested checking the reset and interrupt switches, as a stuck switch could cause strange behaviors and the Sad Mac code I received. However, the switches were quickly ruled out with a continuity test — both were good, testing closed only when pressed.

It was time to get a schematic and start tracing the signal.

The reset and interrupt (non-maskable interrupt; NMI) switches and UA5 sound chip seen in the Macintosh Classic logic board schematic.

The next item to check was UA5, or the sound/reset chip. Upon closer inspection of the board itself, the likely culprit was revealed: A couple of the legs on UA5 appeared dirty and possibly oxidized. After a close inspection of the rest of the board, I still suspected UA5 was the issue. It was the only problem that revealed itself to the naked eye.

Some minor oxidization is seen on a couple pins for UA5 on the Macintosh Classic logic board, as seen in the red circle.

After sending the board off to Thomas at Amiga of Rochester (who recapped the Classic’s logic board), I was happy to discover that my hypothesis was correct — the oxidization on UA5 was preventing the machine from functioning properly. He removed UA5, cleaned the leg, reinserted it and tested the board. It was stable.

While the board was away, I also used the opportunity to replace the original, flimsy 3D-printed BlueSCSI bracket for a sturdier “PotatoFi” one Eric sent me shortly following my BlueSCSI “first impressions” post. This bracket is not only made of thicker material, but also has four mounting holes (opposed to two on the original bracket) that mate perfectly with the holes in the stock drive carrier.

I also swapped the standard flimsy BlueSCSI mount for a sturdier PotatoFi mount, which is made of thicker material and mates perfectly with the four holes found in the Classic’s stock drive carrier.

Once I got the board back, I immediately got it inside the Classic for testing. It works! I had lucked out, as I was afraid the issue may have been deeper than a dirty leg on an IC.

(At the time of finally getting around to publishing this, a couple weeks later: The Classic is still stable, regardless of time powered on. A simple problem seems to be what brought this machine down.)

After using the Macintosh “SuperSE” for weeks now, it is interesting to return to the 68000 world. While the Classic isn’t as snappy as the SuperSE with its ‘030, the Classic is far more easier to work with. The Classic runs System 7.1, and can run some newer versions of System 7 (thru System 7.5.5 — although that would likely be slower than molasses in winter). The Classic can also access high-density floppy disks (and images) and HD-20 volume images on my FloppyEmu. Although I love the SuperSE and its speed, it has many limitations the Classic doesn’t have. It allows me to finally continue testing serial connections, including with a new toy of mine: the TheOldNet modem emulator. I can finally try accessing the web using a very archaic web browser (MacWeb), instead of using text-only ASCII/ANSI terminal emulators connecting to BBSs and other telnet services. More to come on MacWeb to see if it works.

Also coming soon is restoring the 1994 Macintosh PowerBook 165. It’s a project that has kind of concerned me, given my bad history with servicing laptops. It needs a lot of work to ensure it can be used for years to come — such as a more resilient hinge, reinforcing plastic screw posts and getting its battery pack rebuilt for portability. Most importantly, however, is replacing the capacitors on the LCD driver board and finding a modern replacement power supply. (I’ve been warned to not use the original power supply I purchased from a fellow vintage computer hobbyist, as bad caps and other age-related issues can cause fluctuations in voltage — enough to possibly cause damage to the computer. I haven’t used it since receiving that warning.)

Update: Aug. 25, 2023

There’s some good news and some bad news regarding the Classic…

First, the machine has been working. I’ve been tinkering around with serial communication and transferring files between it and my M1 Mac mini. I’ve had some issues with files not going through and it timing out. It seems Kermit and Zmodem are reliable for small transfers, which is the bulk of it.

I also have used Excel (successfully opening a comma-separated value file from the modern web earlier today), PageMaker, FreeHand and SuperPaint. Being able to use those TrueType fonts and having a more modern version of FreeHand is so nice. I also played some Lemmings the other day.

Now time for the bad…

It hasn’t been very reliable. It will run perfectly for hours, just to act up and reboot into one of its infamous “chime cycles.” (See video above.) Tonight it took it a step further, after giving me the “000000F 000000D” Sad Mac code, by displaying a white pattern and making strange buzzing sounds. After letting it sit for a few minutes, it booted right up like nothing ever happened.

I’m suspecting there might be a heat-related problem, since it seems to only fail lately when it has been running for a while. (Sometimes just sitting idle with the After Dark screensaver running.) Perhaps it could be a bad solder joint, either on UA5 or another chip, causing the issue. Or maybe something else is intermittently failing. It sucks I don’t have a laser thermometer or thermal imaging camera to see if there are any hot spots on the logic board.

I’ve also been having problems with screenshots not opening at all on the Mac mini. Previously, System 7 screenshots — in PICT format — would open just fine in Lemke Software’s GraphicConverter 11. From there, I would export the PICT to a JPG file that can be viewed more reliably on the web in pretty much every browser. However, the PICT screenshots I’ve sent over the wire have not opened at all.

Error message displayed by Lemke Software’s GraphicConverter 11 when trying to open a screenshot PICT file from System 7.1 on the Macintosh Classic. The software used to never have problems opening the screenshots.

So it appears I will need to take another closer look soon.

Original post written on a 1991 Macintosh Classic using Microsoft Word 4.0.