My Mac addiction started in 1987 with an SE to which I added an 020 accelerator. Next was a IIci in ’92, adding 040 accelerator in ’96. On to G3 desktop in ’98 and G4 accelerator in ’02. And on and on. I still own the SE which works well after an analog board recap and reflowing. Recently, I’ve restored an IIcx, IIci’s, SE/30, IIsi, LCIII, assorted G3s but I’d never had a Classic.
So I picked up a reasonable-looking, non-working Classic II for $50 on eBay. The listing said it powered-on but had bad video like this:
When it arrived, I powered it up briefly. There was a rather load, ominous buzzing from the speaker and the video was indeed bad. Hmmm.
I tore it apart. The case was in good shape: very few nicks and most scuffs came off with household cleaners and, in particular, Magic Eraser. The CRT was pretty clean. The logic board had the tell-tale signs of leaky caps but was otherwise reasonable. The analog board didn’t look too bad but there were signs of electrolyte leakage on the chassis itself in the area of the secondary side caps:
After removing all caps, the main source of the leakage was plain:
Note the corrosion on the metal frame .. with the analog board mounted vertically, leakage drains down off the PCB (upwards on the photo) on to the frame and chassis.
I scrubbed the analog PCB with IPA to remove the goop and replaced allcaps. Even the two large primary high voltage caps had leaked slightly. About half tested with high ESR.
I tested all diodes good. Many checked out in-circuit but some needed at least one leg lifted.
At this point I wanted to test the power supply sections on the test bench without CRT or logic board connected. But I couldn’t find a write-up of a proven procedure for this. Nevertheless, I found this configuration works well:
But sadly I measured only low voltages: ~4V and ~9V which didn’t vary with the adjustment of PP1. Hmmm.
A frequent villan in these units is the optocoupler, QP1. So pulled it and it was indeed bad — the LED side checked out as diode but the photo-transistor side appeared (with LED open-circuit) as a transistor rather than being open-circuit as you would expect.
I didn’t have an exact replacement on hand but I did have a Sharp PC111 left over from repairing a LCIII PSU. The package was the same and the specs looked close enough, so I stuck that in . And, lo and behold, that worked .. good voltages which adjusted with the trimmer.
That brought me to the logic board. After I removed to caps with hot air, the board looked like so:
There was some mask/trace damage from leakage around C4/C5 and C8/C9. I beeped out the suspect traces and all seemed good so I cleaned them up and applied solder mask (see later). Soldering new caps yielded:
I treated all connector blocks with switch cleaner and DeoxIT, and I removed corrosion from grounding sections of the back-panel connectors.
The next step was to partially assemble and fire it up with the logic board on extension cabling. I with greeted with a load chime, good video and flashing floppy:
Note (removed) electrolyte leakage damage on the surface of the chassis below where the meter probes are attached. The machine had come with 2MB+2MB addition SIMMs and I upped that to the maximum of 4BM+4MB.
The floppy drive worked but not reliably .. it first needed cleaning and lubing. The hard-drive was a Quantum ELS 160MB which had originally showed signs of life but went comatose after running a random seek/write/read test connected to another machine. So I attached an external ZuluSCSI Mini and booted into MacOS 7.5. MacTest Pro and Snooper couldn’t find anything wrong.
But the reset switch didn’t work at all and the programmer’s switch was iffy. Luckily though, a blast and soaking with switch cleaner and DeoxIT did the trick.
When I came to test the serial ports, there were further issues arising from the C4/C5/C8/C9 damage. It took several iterations to track down and fix the follows trace breaks:
Finally, I decided to replace the hard-drive with a ZuluSCSI Compact RP2040 from Rabbit Hole Computing .. I’ve had good luck with Rabbit Hole who deliver fast and have higher quality products than equivalent BlueSCSI units.
Voila .. a nice looking Classic II built in June 1992, now back in full action!
So I picked up a reasonable-looking, non-working Classic II for $50 on eBay. The listing said it powered-on but had bad video like this:
When it arrived, I powered it up briefly. There was a rather load, ominous buzzing from the speaker and the video was indeed bad. Hmmm.
I tore it apart. The case was in good shape: very few nicks and most scuffs came off with household cleaners and, in particular, Magic Eraser. The CRT was pretty clean. The logic board had the tell-tale signs of leaky caps but was otherwise reasonable. The analog board didn’t look too bad but there were signs of electrolyte leakage on the chassis itself in the area of the secondary side caps:
After removing all caps, the main source of the leakage was plain:
Note the corrosion on the metal frame .. with the analog board mounted vertically, leakage drains down off the PCB (upwards on the photo) on to the frame and chassis.
I scrubbed the analog PCB with IPA to remove the goop and replaced allcaps. Even the two large primary high voltage caps had leaked slightly. About half tested with high ESR.
I tested all diodes good. Many checked out in-circuit but some needed at least one leg lifted.
At this point I wanted to test the power supply sections on the test bench without CRT or logic board connected. But I couldn’t find a write-up of a proven procedure for this. Nevertheless, I found this configuration works well:
- CRT neck disconnected
- CRT anode disconnected
- CRT yoke disconnected
- logic board disconnected
- disk drive power connected
- fan connected
But sadly I measured only low voltages: ~4V and ~9V which didn’t vary with the adjustment of PP1. Hmmm.
A frequent villan in these units is the optocoupler, QP1. So pulled it and it was indeed bad — the LED side checked out as diode but the photo-transistor side appeared (with LED open-circuit) as a transistor rather than being open-circuit as you would expect.
I didn’t have an exact replacement on hand but I did have a Sharp PC111 left over from repairing a LCIII PSU. The package was the same and the specs looked close enough, so I stuck that in . And, lo and behold, that worked .. good voltages which adjusted with the trimmer.
That brought me to the logic board. After I removed to caps with hot air, the board looked like so:
There was some mask/trace damage from leakage around C4/C5 and C8/C9. I beeped out the suspect traces and all seemed good so I cleaned them up and applied solder mask (see later). Soldering new caps yielded:
I treated all connector blocks with switch cleaner and DeoxIT, and I removed corrosion from grounding sections of the back-panel connectors.
The next step was to partially assemble and fire it up with the logic board on extension cabling. I with greeted with a load chime, good video and flashing floppy:
Note (removed) electrolyte leakage damage on the surface of the chassis below where the meter probes are attached. The machine had come with 2MB+2MB addition SIMMs and I upped that to the maximum of 4BM+4MB.
The floppy drive worked but not reliably .. it first needed cleaning and lubing. The hard-drive was a Quantum ELS 160MB which had originally showed signs of life but went comatose after running a random seek/write/read test connected to another machine. So I attached an external ZuluSCSI Mini and booted into MacOS 7.5. MacTest Pro and Snooper couldn’t find anything wrong.
But the reset switch didn’t work at all and the programmer’s switch was iffy. Luckily though, a blast and soaking with switch cleaner and DeoxIT did the trick.
When I came to test the serial ports, there were further issues arising from the C4/C5/C8/C9 damage. It took several iterations to track down and fix the follows trace breaks:
- under C8 in the modem RxD- from U3 pin3 to U8 pin16
- to pin16 (+5V) of U3
- printer HSK0 path from J3 to RC1 pin2
Finally, I decided to replace the hard-drive with a ZuluSCSI Compact RP2040 from Rabbit Hole Computing .. I’ve had good luck with Rabbit Hole who deliver fast and have higher quality products than equivalent BlueSCSI units.
Voila .. a nice looking Classic II built in June 1992, now back in full action!
