IIcx and IIci PSUs failing to boot a IIcx

[nurrwick]

Hi folks!

I have a IIcx that refuses to boot with its recapped Astec AA15830 power supply, or the AA15831 that I borrowed out of a IIci. After painstakingly rechecking all of the caps I replaced for accidental shorts between pads and traces, feeding the board a current-limited input on +5 and seeing if anything got warm (nope), and then learning that 16 ohms between +5 to ground is not a sign of trouble, I took the plunge of putting a recapped IIsi APS06 on it, and it booted. We had a little instability on our first trip around, but I am inclined to blame that more on my uhhhhh imperfect bench setup:

than on the PSU, board, or graphics card. It froze when I opened the Monitors control panel the first time, but no such issue the second time. Anyway, hooray! Board at least mostly works, yet another old boy cared for.

So the question of the power supplies sets in... the AA15830 that was in this machine, when installed and regardless of how much stuff is plugged into the board (all of it? none? doesn't matter), will just click/chunk somewhere between once and twice a second - I'd guess about 700ms. The IIci's AA15831 does basically the same thing, but faster - somewhere between two and three times a second. That AA15831 *does* power the IIci on; I can check if the AA15830 works that machine, but have not done so yet.

Since I know the board can respond to voltage by working, I decided to throw a scope at points on +12 and +5 on the IIcx board with the AA15830 installed:

Yellow trace is the +12; it just does the nice even ramp down about 8ms after charging, what I would assume is just cap discharge following the supply entering internal protection. The blue trace is +5. A quick blip up to 5v, then 5ms later it starts that rapid charge/discharge behavior down to 3V, and after about 25ms, it drops into cap discharge territory. I have not run this test with the AA15831 yet. The fans in both supplies do spin while the click disco is taking place.

In anyone's experience, is it going to be worth even trying to troubleshoot these supplies beyond this? It seems like these two supplies are going to be a phenomenal pain to run while open, and while it appears whatever drama is happening is taking place on +5, beyond looking for a visible short, I've not had enough experience with power supply diagnostics to have any idea what would be causing the supply behavior beyond "idk I guess a voltage regulator has failed?" And then there's the complication of the AA15831 powering the IIci's board, RAM, and network card...

Thanks!

 

[volvo242gt]

Got a IIvx, Centris/Quadra 650, or Power Mac 7100 power supply lying around? Could try one of those with the cx board. The last few IIci machines I've owned have received one of those, replacing the original GE or Astec unit with the removable fan module. A little newer, and slightly more powerful.

 

[nurrwick]

Sadly, the Mac hoard skips from the IIsi to beige Power Mac G3s, so it looks like I have some reckoning to do: track down a spare or replace innards with one of the Meanwell-based kits. I guess I can also probe components on the top board before giving up! Thanks for the suggestions on where else to look for options!

 

[nurrwick]

Okay, so the AA16870 arrived and did not solve the problem. In spending a ton of time digging through atrocious search results for a how-to on disassembling this model of PSU, I kept seeing the recurring theme that the soft power circuit on the IIcx and IIci are frequent problem areas on these machines. I guess the IIsi's PSU is different enough that whatever is troubling the IIcx doesn't cause the supply itself to go into protection and reset.

I guess the next step for this box is to get comfy with the schematic and see what other components may need attention, since it's not likely to be a capacitor problem following the cap replacement.

I'm also going to describe the teardown process on the AA16870 so the next poor sap doesn't spend any time wondering if you really are supposed to bend the sheet metal out of the way...

 

[nurrwick]

Having finally had a moment to poke at this machine again, we have some forward progress, at least I assume so. It still doesn't boot, but after studying the schematic, scratching my head and making a face, and then metering across all the caps to check to see if I had shorts, I noticed a short from the positive sides of C15 and C16 to ground.

The keen IIcx afficionados will already know that those two 10uF caps are installed with opposite orientation to the 47uF caps right next to them. This being the only IIcx I've ever seen or worked on, I made the critical mistake of trusting the silk screen. Yes, they are missing the + polarity indicator of the four other caps, but if you're like me, you didn't take a before photo because you trust the screens, the shoulder indicators on the two smaller caps are pointing in the wrong direction. They should be going toward the burn-in edge connector, not the floppy and SCSI connectors.

So anyway, now the machine no longer clicks endlessly as soon as power is applied. Unfortunately, I'm now at a point where it'll click a couple times if I press the power key or the button on the back, and then nothing else happens. Press and hold power on the back? Endless clicking. This happens regardless of how much stuff is plugged into the board, is consistent across all three IIcx-shaped power supplies, and it happens with or without a PRAM battery.

I know this symptom is documented elsewhere on the forum, so I'll start combing those threads and see where troubleshooting takes me!

 

[nurrwick]

Hey folks, doing some thread necromancy as I've gotten the IIcx to chime, no clicking, with an Astec AA16870!

At some point, I had a long, hard think about what was going on with this board because someone, somewhere, said that the power circuit for the IIci was well understood... and at some point after that, I had gone on a tear trying to understand that board to see if any of the lessons would carry over to the IIcx.

It turns out they all do. These two machines use the same circuit for soft power on and off.

I found writeups on the circuit at least two places, one of which I know is somewhere on preterhuman, and the other escapes me, lost to the sands of browser history. The second one was the key - it showed me at least one of the Gamba circuit diagrams, with the power on and power off halves clearly marked out on the schematic. From there, I found the archive of Gamba's work, and loaded up the IIcx and IIci schematics and there we go. That meant I did have a working machine I could probe and scope, and it appeared that the problem existed downstream from the POWEROFF signal from UH6, and component tests indicated it was upstream of the diode chain. I took a guess that the first component in the way, the 74HC132 at UK2, was the problem. I am still not clear on, and probably never will be, exactly what part of the chip let go, but it was the most corrosion-damaged of the three 74HC chips in this area. Attempts to even reflow solder on all but one of this chip's legs failed, so I made the decison to mechanically remove and replace it.

I tried using fine-point snippers to chew on the legs, and the very first one I touched just kind of pulled away from the pad. I didn't want to run any more risk of breaking this board further, so I desoldered the battery holder and got out the dremel (and damaged this board further). Pins 8 through 14 went fine, but the chip went flying when pins 1 and 2 let go similarly to 14, and it was just enough of a slip to cut the trace off pad 2, and then the thin trace between pads 2 and 3. I was able to tone out where each end of that trace went - pin 13 of UH6, which is POWEROFF, and pin 1 of UL2.

I got all my pads cleaned up, got the new chip mounted and cleaned, then ran bodges from pin 2 to pin 8 of UK2 and to replace POWEROFF. I learned my lesson on not reinforcing my bodges with my IIsi, which lost one of its memory transceiver bodges when I moved it, and put small dabs of hot glue at a few points along the wires.

To bench test this, I decided not to do the extremely janky setup I did with the IIsi's PSU in the first post and instead verified that the first 5x2 pins of an ATX connector fit the 10-pin plug and socket, and then dremeled an ATX extension to act as an extension for this. Works great! Set up the speaker and blammo, we have chime.

Next step will be to get it back in the case, get it under a monitor, and make sure both the power on and off circuits are working correctly. At least with it powering on now, it should be so much easier to narrow down faults. I could probably stand to replace the 74HC74 given how hard it was to get the bodge to stick to pin 1, and I've definitely learned that to cut these off, first use tape to hold them down, and second come in at a very shallow angle.

If I'm able to refind the writeups that helped me, I will add links to this thread. Things that were helpful that I do already know where to find:
https://68kmla.org/bb/threads/macintosh-iicx-schematics-bom.51317/ - the official Apple schematic helped me figure out the broken POWEROFF trace
https://www.vintageapple.org/gamba2/schematics.html - the Gamba soft power schematics.