Mac IIsi Repairathon

[RetroViator]

Alpha is A-Okay​

After spending time on Bravo's audio problems, I decided to check out Alpha's speaker problem while the subject was still fresh in my mind. Alpha was working perfectly after I restored it back in 2021, but I was using it as a known-good system for testing, and along the way, my sloppy probe work resulted in the speaker dying.

To get things started, I verified the signal coming out of the Sony ICs and followed it to the bottom of the board. Everything was making sense up to the TL071B op amp at UK3. The signal was coming into pin 2, but the output from pin 6 was garbage. It was coming out a -8.8V and there was just the faintest hint of a signal. All the other pins looked normal, so it seemed something had gone amiss with the op amp.

I ordered replacements, so I installed a new TL071B--and the problem remained. Maybe I ordered the wrong part; so I scavenged one off the battery-bombed Foxtrot--and the problem remained. I was now scratching my head, because all the signals and voltages coming into the op amp were correct, but output was wrong.

I pitched the question to ChatGPT and it suggested the problem was coming downstream of pin 6, keeping it from producing good output. I followed its recommendation and removed R41 to isolate the downstream circuit, and sound came out of the speaker! The sound was a bit hissy, but it was something.

So, I reinstalled R41 (not easy given how small it is) and removed Q11 (an NPN). The probem was back. So, I reinstalled Q11 and pulled Q10 (a PNP), and the speaker was working again. So, apparently I had damaged that transistor with my probe. I had ordered new MMBT3906s from Mouser, but I remembered the issue I had with Bravo, and I supected the pin orientation did not patch the footprint on the board. Since this circuit was easier to follow, I was able to figure out that collector and base were reversed on the new MMBT3906 compared with how the board is oriented. Luckily, I was able to scavenge yet another part from Foxtrot, and with its Q10 installed, the speaker sounded great.



In the end, I replaced UK3 (twice), and I had to reinstall nearby C41 when it came off while replacing UK3. I also pulled and reinstalled R41 and Q11, then finally replaced the faulty Q10. I was distracted by the odd output coming out of UK3, but after working my way down the cirucit, I found the shorted transistor.

 

[RetroViator]

Bravo is Beautiful​

Bravo is the system I've worked on the longest (technically, I started working on it in 2021), and it is finally good to go.

It's last problem was bad serial communication. TechStep was reporting a Serial Communication Controller test failure. Most specifically, it said there was a problem with the handshake input. So, I followed the handshake input pin on the jack to UA6, which is a 75175 (a quadruple differential line receiver). This IC is right next to C9, C10, and C11, and the legs did not look great. After some continuity testing, it looked like several of the IC's pins were not making good contact.

I decided to remove it, and yep, it was pretty dirty. It took quite a while to get the corrosion off the pads and fresh solder in place.



Once I reattached the IC, I connected the TechStep, and for the first time, Bravo passed all its tests!



What work remains?

• Figure out why Charlie won't shut down properly with the Power-Up IIusi replacement PSU
• Cobble together one more working floppy from the remaining parts
• Build four more Power-Up IIusi PSUs

 

[RetroViator]

Charlie is Copacetic​

Charlie was working well after the recap, except it would not properly shutdown with Caymac's Power Up IIusi replacement PSU. I followed the power signal on the PSU's pin 9 (J15) back to the Egret chip, a 68HC05, at UB1. The connection seemed good, so the issue was either with the Egret or the PSU.

With an original Magnetek PSU, I was getting 12.2V, 5.11V, and -12.07V. With the replacement IIusi PSU, I got 12.2V, 4.9V, and -11.2V. One interesting thing is I got a flash on all the power rails when I plugged in the PSU.

When I scoped the signals coming from the new PSU, pin 9 had a jaggy sawtooth signal.



I had harvested the wiring harness off an old PSU, and the pins were a bit green. So I depinned all the wires coming from the PSU and cleaned them with a fiberglass pen and some IPA. I also cleaned up J15, which looked rough too.



After clearning, I tested again, and while the shutdown still didn't work, the power rails no longer flashed when I plugged the PSU into J15 and the sawtooth signal was gone. While cleaning helped, it did not solve the problem.

When triggering a shutdown from the Mac's menu, the signal on pin 9 dropped very briefly but then jumped back up. I confirmed that the computer would shut down immediately when pin 9 was grounded, but would immediately power back on when the ground was removed. After plugging this problem into ChatGPT, along with the pertient schematics, it recommended installing a 100k resistor between the base of Q1, a 2N4401, and ground, providing a better base pull-down. As with my eariler post on the IIusi replacement PSU, all the issues seem to flow back to Q1.



After putting that in place, the shutdown worked perfectly! I tested the new PSU on all five of the IIsi boards, and they all worked just fine.

 

[Byte Knight]

I'm really enjoying reading your updates from the BBS using the TinkerDifferent telnet server. I did a ton of BBS'ing with my IIsi, so it seems very appropriate. I can't comment from the BBS since the telnet server is read-only for now, but I'm following with great interest!

 

[RetroViator]

Software Slows Things Down​

I've reached the point in this project where I need to get the software squared away. As usual, this slows things down.

I don't want to give the wrong impression — this is so much better than it used to be, thanks to the great work of so many in our community. The good folks on the E-Maculation forum, who keep Basilisk II and SheepShaver up to date; Mihai Parparita, who pours so much into InfiniteMac.com; Thomas, for his work with the impressive new Snow emulator; @fogWraith, who generously cultivates the incredible Macintosh Garden; and @eric and the amazing BlueSCSI team have all significantly improved the quality of vintage Macintosh software.

I took this opportunity to clean up several messes I had made. I started by gathering together all the various volume and device images I had created and assembled over the years. After getting the latest version of Basilisk II and Snow running on my systems (Mac, Linux, and Windows), I set about opening each image and pulling together the best of what I had collected. What I discovered is that most of what I had pieced together was not as good as what @eric provided here.

I then updated my external BlueSCSI v2 with a new image that incorporated System Picker, allowing me to switch between System 6.08, System 7.01, System 7.1, and System 7.5.3 as needed. I also added the Utilities from the MacPack, along with other useful system software I've scavenged over the years. Of course, the Apple Legacy Recovery CD was as invaluable as always.

I then set about building a solid BlueSCSI image for the IIsi. I decided that System 7.1 was just about right. Even though the system originally shipped with System 6.07, it ended its run after System 7.1 was released. So I gathered era-appropriate software and copied it onto SD cards for insertion in my BlueSCSI. I had two older BlueSCSI v2s with Raspberry Pi Pico boards, lacking WiFi. So, I ordered two updated BlueSCSIs to play around with WiFI transfers. I haven't decided yet which systems will get BlueSCSIs and which will keep their spinning-rust drives. In any case, I am finally ready to get the hard drives squared away and then build the remaining replacement power supplies to wrap up this project.

 

[RetroViator]

Hero System​

I pulled together the best of my working components and assembled a hero unit. I went with the Echo logic board because it looks clean and doesn't have any bodges. I put in four 2MB RAM SIMMs, to get the system up to 9MB. Then, I selected a clean non-yellowed case, picked a good floppy drive, hard drive bracket, fan, and the first of the Power Up IIusi replacement power supplies. I also installed the latest BlueSCSI v2 with a Pico W, providing WiFi capabilities.



To add a cherry on top, I installed my one-and-only direct slot adapter card, with its floating-point co-processor (FPU). This card extends the PDS slot at a right angle. Apple released another card (also with an FPU), to accommodate NuBus cards. When I installed the card, I heard the dreaded chimes of death, so I knew something was wrong. After looking at it for a second, it was obvious. The card's one and only electrolytic capacitor had leaked.



I replaced the cap, and I also pulled the nearby tantalum capacitor, to re-flow its solder. After replacing the electrolytic and reinstalling the tantalum, I cleaned up the nearby vias and removed any visible corrosion. I put the card back in the IIsi, and it chimed happily. However, I ran the FPU tests in Snooper 2, and the FPU failed. So, I pulled the card again, this time paying close attention to the FPU. I added flux and went around the chip with my iron, keeping a bit of fresh solder on the tip. The side of the IC next to the leaky cap was dull and questionable. I worked on all the vias until they were silver again. Then I checked for shorts on the FPU. I had to bring up a datasheet because there are a number of ground pins, but it toned out properly--giving me hope the chip had not shorted internally. So, I put it back in the IIsi, and this time Snooper reported good test results!

I now have a tricked out IIsi, or at least tricked out enough for me. This computer will never be a speed demon, but it is working well. I may play around with overclocking it, but I might not, as that might be an issue for the PDS-connected FPU.

P.S. You can see in the photo above that the status light is both green and blue. I installed a blue LED on the BlueSCSI's hard drive indicator. However, I kept the original green power-on LED next to it. I was suprprised that the light pipe is wide enough to cover both LEDs, so I see a bit of green when the computer is on and bit of blue with the hard drive is accessed.

 

[Byte Knight]

Looks great! I added a TechWorks Ethernet / FPU PDS card and a DayStar IIsi cache adapter / 50 MHz 68030 accelerator to mine - it's pretty amazing what you can cram into that one shallow PDS slot.

 

[RetroViator]

Looks great! I added a TechWorks Ethernet / FPU PDS card and a DayStar IIsi cache adapter / 50 MHz 68030 accelerator to mine - it's pretty amazing what you can cram into that one shallow PDS slot.

You just reminded me that Golf, the horribly battery-bombed IIsi I got from a nice retired couple who were downsizing, came with a DayStar 601 Processor Card. This jumped the lowly 20MHz 68030 IIsi to a 80MHz PowerMac. I had planned on putting this 601 Processor Card in one of my IIci (a restoration I will get back to soon), but it just hit me that it might be a PDS card and not NuBus. I checked, and indeed it is a PDS card. So, I will have a hero and a hero+ system! One as described above, and the other sporting the 601 Processor Card. I need to find the software, but I'm hopeful that Macintosh Garden will come to my rescue.

Here are a few photos of the 601 Processor Card.

 

[RetroViator]

PSUpalooza​

This weekend I finally put together four additional Power-Up-IIusi replacement power supplies. This was pretty uneventful, as I figured out the nuisances with my earlier single PSU build. This time, I built each PSU up in an assembly line, simlar to my floppy restoration.

I started by soldering on the connector for the PicoATX, along with the transistor, which I knew to install backwards from the silkscreen due to the placement of the collector and emitter pins on the transistors I ordered from Mouser.



I then soldered on the 100K resistor on the back, thanks to the discovery I made earlier to ensure consistent shutdowns.



Then I harvested four power connectors from the old PSUs. I depinned each to clean up any corrosion and added ferrule connectors to the ends of cables. I then soldered them onto the PCBs.



Then I tweaked the mains plugs to wire the monitor receptacle so it was always hot, since the new PSU could not switch it on with soft power.



Finally, I put each into one of the original PSU cases and tested them. I was pleased that each provided good voltage and the IIsi passed all of Snooper 2's diagnostics while running on each of the PSUs.



When putting these together, I realized the nylon M2.5 6mm screws I ordered to secure the 12V power supply to the PCB did not have large enough heads. Some of the screw heads popped through the holes on the PCB. So, I ordered new screws, and will install them once they arrive. Then, I'll finally be able to assemble all the parts into complete IIsi systems!

 

[RetroViator]

PSU Redux​

I received the washers so I could button up the Power Up IIusi replacement PSUs, but something was bothering me.

CayMac Vintage had updated the design significantly from the open source version I built, and the new design is better in many ways. First, you don't have to tweak the mains plugs, you simply transfer the connector over to the new PCB and plug in the mains outlets as-is. You do need to move over the fuses, but that is easy to do. Just like in the earlier design, you also need to reuse the wiring harness going to the IIsi. The biggest change with the new design is it uses a combination of Mean Well units to generate the voltages. The earlier design used a WX-DC2412 to convert the AC to 12V DC feeding the PicoATX. The combination of the low cost AC converter and the PicoATX introduced a significant amount of ripple into the power rails. I have not done a deep dive, but it looks like most of noise is coming from the AC to 12V DC converter.

Here is a screenshot from my oscilloscope looking at the 5V trickle with no load on the PSU.


It shows a 3.28V peak-to-peak swing in those noisy bursts.

So, I ordered up a pair of CayMac Vintage's new Power Up IIusi kits. @Fizzbinn has already done a post on the new PSUs here, so I won't repeat that, but I was a bit challenged because I had already built up five of the earlier Power Up IIusi units. Therefore, I had already modified my mains outlets and cut off the connector. I had kept the connectors and the spare wire that you don't need after the modification, but I will need to find the proper terminals to plug the wires back into the connector.

However, I did have one unmodified outlet in a PSU from one of the battery bombed IIsi. The trouble was it was badly corroded, I could not remove the screw to release the PCB from the PSU's case. So, out came the Dremel, and I cut away the metal holding the PCB in place.



Once that was done, I pulled the parts I needed and finished assembling a new Power Up IIusi PSUs.



Once I plugged it in, I saw a clear difference on my oscilloscope.


A mere 320mV peak-to-peak! Essentially no ripple at all.

CayMac Vintage is well aware of this, and that is why he moved to the new design. I should have bought the new kits from his shop at the beginning of the repairathon, but I decided to take my first stab at ordering custom PCBs for an open source project. Also, the Mean Well units are not cheap, especially if you're buying enough for five IIsi.

Now, I would like to know the type of terminal ends (and crimping tool) I need if I want to reattach the mains wires to the connector. The connector and pictures of the terminals ends are below.



Update: Thanks to @btb keen eye, terminal connectors appear to be JST V Connectors, specifically SVF-01T-1.5A.

 

[RetroViator]

I built the second of the updated Power Up IIusi kits I ordered after the JST V connectors arrived, so I could transfer the mains plug I had modified in the earlier Power Up IIusi design. I also needed to remove the Mac IIsi wiring connector. It was fiddly, but I got it done.



The 5V trickle power tested good, and I confidently plugged the new Power Up IIusi into one of the IIsi, and... nothing happened... it wouldn't power on.

I pulled the PSU apart and checked every solder joint on the PCB and every connector. I could test if it would power on by sending 5V from pin 10 (Trickle) to pin 9 (PFW), which triggers start-up. From my new working Power Up IIusi, I knew I should hear the relay click and the LED on the PCB light up. After a considerable amount of time redoing most of my work, I was down to the transistor and resistors. I swapped the Q1 transistor with another 2N4401, but still nothing. So I pulled each resistor, which was a pain once the wiring harness and Mean Well units were installed. I checked each resistor's value, and they were fine. As an act of desperation, I decided to flip Q1 backward from the silkscreen since that had worked on the earlier IIusi design, and... it worked. The relay clicked, and the Mean Well units sprang to life.

I had tried earlier to test the transistor in my component tester before installing a replacement, but because of the close spacing of the through-holes on the PCB, the legs were too short to fit into the tester. However, after my backward replacement transistor worked, I soldered legs on the kit-provided transistor to test it. When I did, I found it had the same configuration as my replacement transistor. That means the placement was also backward from the silkscreen. This issue bit me again! I was not expecting this since I was using the transistor that came with the kit, and the other Power Up IIusi kit was just fine. Apparently, I received two different transistors with opposite leg configurations.



The lesson is, if you have an issue with the Power Up IIusi (either version), the problem is likely with the Q1 transistor.

 

[RetroViator]

Hero+ System​

Earlier (here), I built up the Echo board and outfitted it as my "hero" model with my best accessories. However, when @Byte Knight shared that his IIsi came with a DayStar accelerator, I realized that the original Alpha system I acquired from a nice retired couple in 2020 came with a DayStar card installed.

So, I picked the Charlie board and outfitted it with a good case, one of the new Power Up IIusi PSUs, a BlueSCSI v.2 with WiFI, a working floppy, and 4 x 4MB SIMMs, bringing the system to an impressive 18MB of RAM. I then installed the DayStar card.

It took a few minutes to determine which card I had. DayStar made a number of processor cards, and this was the DayStar Turbo 601 with a 256K processor cache and running at 66MHz. Thanks to the Macintosh Garden, I found the installation disk I needed. I dropped the HQX file into the "shared" directory on my BlueSCSI SD card and used the handy BlueSCSI Toolbox to "download" it to the IIsi. After extracting the archive, I excitedly clicked the inviting Install button, but I was quickly told I needed to install System 7.5. So, I used the Snow emulator to build a custom System 7.5.3 install. It took a while, but when done, I copied over the Utilities folder from my existing IIsi image, skipping the other files, figuring I would want PPC-enabled software if this worked. This time, the DayStar software loaded easily, and I was ready to restart.



What I wasn't expecting was the resonant bong of a Power Mac coming from the IIsi speaker! I checked the DayStar control panel and saw that the system was now running on the PowerPC 601 processor. I ran a few benchmarks, and it was confirmed that the system is now a PowerPC sleeper!



This is now a very fast IIsi--capable for accessing the Internet. For a computer intentionally hobbled by Apple so as not to compete with the mainstay-IIci, this computer finally has its revenge. Though... this card might find its way into a IIci when I finally get back to restoring them, but for now, I will enjoy this fast little IIsi.