Mac IIsi Repairathon

An embarrassing long time ago, I set off on an effort to restore a collection of Mac IIci. I had much to learn. I knew enough to replace capacitors and clear corroded traces, but I lacked the knowledge to tackle serious problems. I may still lack that knowledge, but I'm going to try again. This time, with a collection of Mac IIsi. (I will get back to the IIci, but I want to start with these IIsi to clear out mental cobwebs and space in my workshop.)

Even longer ago, I successfully restored my first Mac IIsi, and I documented my efforts here. This provides me with one known good Mac IIsi, which is a decent place to start. I've now accumulated six non-working IIsi in my workshop. Two are battery bombed, so I don't expect much of them, but the other four have hope.


To keep things straight, I'm labeling them A through G, using the phonetic alphabet. That makes Alpha the first Mac IIsi I restored back in 2021 (not in the picture above). It will be pressed into service when necessary for testing. Bravo will be another system I partially restored back in 2021. Looking at my old work caused a bit of cringe, but the cap replacement looks passable. However, I attempted some fancy bodge work which I now suspect is unnecessary. So that's where I'm starting.

Beginning with Bravo:

This is the sister to Alpha, which was successfully restored in early 2021. I acquired both from a Texas seller who said he picked them up at Computer Reset. I started off repairing both, but when this unit didn't work after the cap replacement, I set it aside.

It seems I lifted a pad on C10 back in 2021, but that was an easy fix as the via is adjacent. I vaguely recall the system booting to a gray screen, and at some point, I somehow zeroed in on the RAM bus transceivers. Both UE5 and UH5 were quite corroded. This is where I got cute, and ran some fancy bodges when I couldn't restore connectivity to the nearby DRAM. But when things didn't go well, I again set the computer aside.

So, this time I started with removing my bodges and removing UE5. (I lacked the courage to remove surface mount ICs in my earlier repair attempts.) The chip and its pads were disgusting. After lots of flux and heat, I was finally able to get new solder on the pads. I then discovered the bad trace from pin 3 to a via located under the chip. Once I fixed the trace, that bodge became unnecessary. I'm glad, because I also realized I ran my earlier bodge to the wrong DRAM! Pin 3 of UE5 should be connected to pin 2 of UF7, not UF6. Cleaning the pads restored connectivity from pin 9 to UF6 pin 19, making that bodge unnecessary.

I then worked on UH5. I found three pins were no longer connected to their pads due to corrosion. After removing the IC and applying lots of flux, heat, and working it over with fresh solder, I finally got the pads and chip legs clean. I then reattached the UH5 and carefully checked all connections. It looked good, so I prepped the system for testing.

I was happy to see it spring to life! After running a full suite of MacTest Pro diagnostics, it looked to be good to go. However, I noticed the absence of sound. I plugged in some headphones and could hear. Though it wasn't great sound, and the volume slider didn't seem to change anything.

I went through and tested all connections from the Sony sound chips (UB7 and UB8) to the speaker. It was tricky to trace all the way across the board, but I eventually confirmed connectivity. When I probed the speaker with an oscilloscope, I saw noise, but no identifiable signal. I probed the output pin (11) on UB7 and UB8 and saw a lot of noise, but I could also make out a faint square wave poking through when I generated a beep sound. Something was causing noise and a weak signal. While the signal was apparently decent enough to make the headphones work, once the signal traveled across the board, passing through a number of resistors and capacitors, there was not enough for the op amp to pick up for the speaker.

I eventually traced every line in the sound circuitry. Once I was sure everything was connected, I probed each line with the oscilloscope and measured the results. I then pulled out unit Alpha (my working IIsi), and did the same. Everything matched, until I got to UB7 and UB8. It was soon obvious that I was missing about half the voltage on shared pins 12, 13 (getting about 4V instead of 8V) and shared pin 17 only had about 3V instead of 5V. It was then that I realized that pin 16 was connected to something. I say this as if it's a revelation because the BOMARC schematics don't show what either pin 16 is connected to. I was ignoring pin 16, and then realized the drawings did show a line that went off the page and onto a page I don't have! Probing Alpha's pin 16, I could see it is a clock signal humming at 22.5KHz and 5.5V. However, on Bravo, pin 16 is providing 22.5KHz, but at only 2.5V, and the resulting amplitude is much lower.



I could see that the trace for pin 16 hits a via to the other side of the board where it contacts R85. That resistor is reading a correct 33.3k Ohms. Then it hits another via and disappers.

• Does anyone know what pin 16 on UB7 and UB8 goes to?
• Does anyone have the missing page from the BOMARC schematics?
• Can anyone point me in the right direction for the datasheet to the Sony sound chips?

Byte Knight said:

Good to see you restoring Macs again! Where on earth did you get six IIsi's from?

Click to expand...

Well, they sort of grew over time.

I got my first from a retired couple who sold me all their old Macs when they downsized. Unfortunately, this is one of the battery bombed units. I then went on the hunt for a replacement, and I ended up getting two because the seller was slow to ship the one I ordered so he graciously sent along another. Finally, I got four more (for a total of seven) from someone who had picked them up, years ago along with some IIci, from an auction. I was interested in their collection of IIci, but the IIsi were part of the package.

I'm not looking for any more!

Charlie:​

The Charlie system needed a recap, a good clean, and it worked perfectly. It's nice to have one easy one!



Though, some of my problems with Alpha and Bravo were due by my poor recapping skills when I tackled them in 2021. Now, I'll sometimes use the snip and remove the legs method, but for serious recapping, I use my Hakko hot tweezers. I had trouble getting the hang of it at first, but this time, everything went smoothly. I've noticed the tweezers struggle when dealing with serious pad corrosion.

This was not leaking badly. However, I decided it try to fire it up before the recap, and I was greeted with a hissing power supply and a bad video signal. So it really needed that recap.

PL212 said:

I will say, as the owner of a IIsi in 1990, the speakers on these are quite prone to losing their connection -- it's just pressure pad fingers, which can easily oxidize. If you're getting headphone audio but no speaker audio, try cleaning both the speaker assembly and the place on the motherboard where it mounts...

Click to expand...

Yes, I did clean them, and even added fresh solder to see of thickening them up a bit would help. Unfortunately, something more troublesome is going on. Also, unfortunately, I damaged the speaker amplification circuit on my working Alpha unit while trying to diagnose what was going on with Bravo. I suspect I shorted something with my sloppy oscilloscope probe handling. I've ordered replacement op amps and transistors, and I'll get back to the audio problems on Bravo (and now Alpha), after the parts arrive.

Delta:​

Another straightforward recap and clean for this one. I will say I make less dumb mistakes as I go when I recap several machines in a row. This logic board is working well.

Delta had corrosion from leakage, but it was not terrible. No traces were eaten away and the IC pins were okay. There was blue nasty corrosion in spots across the board. I believe that was coming from the power supply. (I’ll deal with terrible Sony PSUs soon enough.) This corrosion seemed to travel through the air and attach to larger metal pieces, such as the testing connector.

Echo:​

This IIsi was sending me mixed messages. It was the dirtiest of the non-battery bombed systems, so I tackled it last, assuming it would likely have issues.

However, when I started recapping, the old caps came off pretty easily, and even more surprising, I was able to get the pads to take fresh solder with little drama. After the recap, I cleaned the board with IPA and a good scrub of dirty looking pins and decided to give it a test. While I got a solid bong and a happy Mac, the boot process stopped with a bus error. I was using the system's original RAM, so I swapped them out for some known good SIMMs, and everything was fine. It passed all the diagnostic tests. However I suspicious...

Some of the pins, especially around the Memory Decode Unit (pins 50-80) and UJ5 ROM IC looked bad. Actually, all the resistors and ICs nearby were not great. Once I hit the pins with flux and heat, I was greeted with a strong smell--not the fishy smell, but urine. This makes me think the board was suffering not from leaking electrolyte but rodent pee. This IIsi came from the same lot as a IIci that was destroyed by serving as a mouse house. While this system was dirty, it was not that dirty. So the mouse must have just passed through, and settled in the more spacious IIci.



I checked the SIMMs in my Retro Chip Test Pro, and they came back good. And after I thoroughly cleaned the MDU, ROM, and nearby components, I tested again with the original SIMMs, and everything is fine.

Logic Board Summary:​

I started with one working IIsi, and I now have five functional IIsi logic boards!

Here is a run down:

• Alpha - Recapped and working in 2021, now has an audio issue *
• Bravo - Recapped and failed repair in 2021, now working except for an audio issue
• Charlie - Recapped and working
• Delta - Recapped and working
• Echo - Recapped and working
• Foxtrot - Battery bombed
• Golf - Battery bombed

* I was using Alpha to help repair the audio issue with Bravo, and I foolishly touched my oscilloscope probe to the wrong thing and now the speaker has distorted audio. I have ordered spare parts and will revisit the audio issues with both Alpha and Bravo.

Next: In addition to fixing the remaining audio issues, I will clean up the floppy drives and see if the original hard drives are salvageable (though I expect to use BlueSCSIs). Then I'll tackle the infamous IIsi power supplies.

Thankfully, I've had one working Magnetek PSU for testing, because it looks like all the Sony PSUs have issues. Of course, the daughtboard on the PSU is a known problem, and there may be others. So, I've decided to try building replacement PSUs. I took advantage of the replacement PCBs developed by CayMac Vintage. I am using the older design that is posted to GitHub. After I reached out with a question, Will offered to send me some new PCBs with his latest design, but my order was already in with JLCPCB. Truth be told, this is the first time I've ordered custom PCBs. I'm looking forward to the parts arriving so I can get to work making five functional IIsi PSUs.

Tonight was floppy night​

I worked on all seven floppy drives, following @Branchus excellent restoration tutorial. It was like spending an evening with the Bob Ross of vintage Mac repairs.



I nearly stumbled on the eject gears. While disassembling the eject mechanisms, I noticed that two of the seven had a different kind of metal covering the gear box and the problematic gear inside was white, not cheesy yellow. That prompted me to flip them over to see that two were made by Sanwa and the rest Omron.



In the back of my brain, something was telling me that mattered. I vaguely remembered buying two sets of gears at some point in the past. Sure enough, my past self was smart and ordered replacement gears for both the common Omron and the less common Sanwa eject mechanisms. Even though the existing Sanwa gears looked okay, I replaced them anyway since I had the parts, but I kept the original gears just in case.

In the end, four of the seven drives passed testing.

Two of the drives came from the battery bombed systems, so I knew all seven was a long shot, but I hope to get at least one more working from the parts of the three non-working drives so I have five floppies to go with my five functioning logic boards.

Hard Drive Archeology​

Tonight, I plugged in seven hard drives from the Mac IIsi to see if anything was salvageable.



There were three Quantum LPS drives (two of which were 80MB and one was 230MB), a Quantum ELS of unknown size, a 40MB Conner drive, a 120MB Maxtor, and finally a 200MB Maxtor drive.

In the end, one of the 80MB Quantum drives, the Maxtor 120MB, and Maxtor 200MB drives were accessible.

I attempted to access the drives by plugging them into one of the Mac IIsi, and booting from my external BlueSCSI. For the operable drives, I copied their files over to the BlueSCSI. I then created an appropriately sized HDA images using Disc Jockey and opened my BlueSCSI image in Basilisk to copy the files over to the new HDA images for archiving.

• The Maxtor 80MB was the most interesting. This was apparently used by the news department at WUSF in the 1990s. WUSF is a public radio station broadcasting from the campus of the University of South Florida in the Tampa Bay area. Most of the files were written in WordPerfect, though oddly, WordPerfect was no longer installed on the drive. There were hundreds, maybe thousands, of news scripts from the mid to late 1990s and a few other journalism files. The shocking thing is this drive came from a battery bombed IIsi. The drive cage is rusted and the drive looks pretty bad; I never expected it to work!
• The Quantum 80MB was moderately interesting. It looked to be from a well loved system with great applications, such as Adobe Illustrator 5.5, Photoshop 2.5, Claris CAD, and other drawing programs. There were a few games and a copy of Netscape Navigator 3.0. The bookmarks revealed that the user was likely Japanese and enjoyed mid '90s chat rooms. All the personal files had been removed.
• The Maxtor 200MB was boring. It was a wiped drive with an install of System 7.5.

I may come back to one or two of the dead drives and try pulling data from them using the BlueSCSI initator mode, but most of them clearly had mechanical problems.