IIsi power supply won't turn off, I'm deep in schematic land

[Garrett B]

Had a chance to look at this further. Agreed that this section of the circuit involves audio. I see a lot of 12V and -12V rail triangles in the schematic at that point. I'm still at a loss here. The "45" transistor is only involved with turning the power supply on or off. When it's on, the power supply is on. When it's off, the power supply is off. Like you said, it only involves /PFW and 5VSB. I don't see how adding a resistor to ground right there would do anything catastrophic like change voltages (which I'm assuming is what happened if something fried). Is there any chance that something was accidentally shorted together (or maybe left open) when re-soldering the riser board back onto the main PSU board?

I also had wondered if I shorted something out from the underside of the logic board, but my desk was empty in that area I was testing. It must have been something I missed, or a complete fluke, because the board itself still works just fine. Power supply is fine too. I have no idea what happened. After more head scratching, I just replaced the NPN and PNP transistors and things are back to normal.

 

[dougg3]

I performed the resistor modification on two other problematic power supplies, and can happily report they're powering on/off as intended. You were right, the original bodge I made was just fine, except on my other two units, I got smart and just piggy-backed the 10k resistor on the ceramic capacitor.

Awesome and congrats! That's a clever way to put on the 10k resistor, I like it!

I just want to be clear that I have no idea whether this 10k resistor is a good long term mod to make, or if there might be any bad side effects. I was mostly just doing it for diagnostics to get a feel for where the problem is coming from. With that said, it does seem to help because it causes the transistor to require more "oomph" from /PFW to actually turn on.

I'm just super confused about what the actual underlying problem is because all of these power supplies did work fine at some point in the past.

Some other interesting pieces of data - the measured resistance across base/emitter of transistor 45 varied a bit between units. One measured 8.4k, one was 6.4k, and my last was 8.2k.

It's definitely interesting to see that much variance in the values there. Transistor 45 has a couple of built-in resistors: an internal 22k resistor from the base pin to the actual base of the transistor, and a 47k resistor from the the actual base of the transistor to the emitter pin. I wonder if the cap goo can get inside of there and muck things up and then be almost uncleanable?

My replacement 45 transistor measures 63.6k across base/emitter pins (in circuit), and the original one I pulled off the board (out of circuit) measures 65.8k. Those measurements make sense, because theoretically there should be 47k + 22k = 69k of resistance from the base pin to the emitter pin.

I'm guessing the measurements you took were after adding the 10k resistor? Because it looks like 69k in parallel with 10k gives you a theoretical combined resistance of 8.7k which is close to what you saw on two of them. Of course, in-circuit measurements might be throwing things off a bit too.

I'm just happy to have three working units. These had been sitting disassembled in a box for years... until now!

I'm glad it helped, and I hope the mod is actually good/safe to do! I still want to figure out the root cause of why so many of these PSUs do this after being recapped.

Power supply is fine too. I have no idea what happened. After more head scratching, I just replaced the NPN and PNP transistors and things are back to normal.

Very weird. I'm glad everything seems all good at least!

 

[dougg3]

Here's a closer look at /PFW and +5V when the "turn back on" happens. (They're both perfectly overlapped so you can't see the marker for +5V, but it's yellow). The turned-off 5V rail has only dropped to about 1.2V when it powers back on. /PFW has gradually crept its way right to about the same level, which is enough to turn on the 45 transistor. I'm assuming really low current means the diode drop across D3 on the logic board is also really low. But...is the diode drop low enough that the yellow line seems to get all the way up to the pink line in my capture?

If the diode drop is really supposed to be larger than what I'm seeing, maybe something else is accidentally leaking power onto /PFW?

 

[Garrett B]

This fascinating issue may be just the thing to convince my wife it's critical I spend $$ on a scope. I would love to see what my +5V line is doing. My suspicion is that it's trending the same as yours. Did you ever try a bleeder resistor? I still think that there's some electrolytic fluid somewhere that's either wicked inside a component, or in a hard to clean place, that's causing extra conductivity. Could also be that source of power leakage onto /PFW?

A year or two ago I did a test swapping daughtercards between three PSUs to further isolate the problem. From what I remember, the problem did not follow the daughtercard. I also don't think it's the logic board's fault, as some recapped PSUs would work, and some wouldn't.

About that same time, I tried to find datasheets for all the OEM caps installed to compare impedance/ESR with the ones I had replaced. I found that the Nichicon UPS/UPM/UPW caps I used were almost identical in terms of electrical properties. This leads me to believe it's not the capacitor selection either.

 

[jmacz]

@dougg3 when you said you took every component off, cleaned the board, and put them back on, was that the daughtercard only or also the main board? I don’t have a IIsi anymore but two of them I worked on had the same problem you are describing and both went away when I let the main PSU board soak in IPA for half an hour. I did not submerge the PSU but just had enough IPA to cover only the PCB of the main board. I assumed this flushed out some remaining capacitor juice under one of the components. Never figured out the root cause but both PSUs were perfect after that.

 

[dougg3]

I wonder if mine sometimes briefly powering on and right back off when I initially plug it into AC power is a clue? That seems like it could potentially point toward some interaction between /PFW and the standby power, since I would think the caps would be fully drained at that point.

Did you ever try a bleeder resistor? I still think that there's some electrolytic fluid somewhere that's either wicked inside a component, or in a hard to clean place, that's causing extra conductivity. Could also be that source of power leakage onto /PFW?

I haven't tried a bleeder resistor yet. Leaked fluid somewhere is definitely a possibility. I'm not seeing anything with a low enough resistance to /PFW that would cause this, but I definitely could have missed something. /PFW doesn't go to very many places on the power supply board, so if something is bridging onto it, there are only a few possibilities for where it could be happening.

Good to know about your tests with the problem not following the daughtercard, and same with your cap choices being almost identical.

@dougg3 when you said you took every component off, cleaned the board, and put them back on, was that the daughtercard only or also the main board?

I only did that with the daughtercard. I did pull all the larger components off the secondary side of the main board so that I could thoroughly scrub everything in the vicinity with a toothbrush and IPA. That whole side of the board did get a good cleaning. But admittedly, I didn't go so far as to give the entire PSU main board an IPA bath. If I don't discover anything with my research soon, I may try it just to eliminate possibilities, thanks for the data points!

 

[dougg3]

Because I'm crazy, I cobbled together a IIsi power supply tester. Probably good to have anyway if I'm planning on designing a replica riser board. It'll help me test and calibrate a new riser board without potentially damaging a motherboard.

I just used a 20-pin ATX extension cable and cut it up to turn it into a 10-pin Molex Mini-Fit Jr breakout cable. To make things easier to look at, I moved the wiring around in the connector so the colors would make sense.



I set up a microcontroller to precisely control /PFW. I don't have anything sitting around that has 5V GPIO and also runs on low enough power to only need the 1 mA that PSTRICL is rated for, so I just used one of my old AVR-based ROM SIMM programmers and powered it over USB for controlling /PFW. I leave /PFW floating most of the time. I can temporarily drive it high for 200 ms to turn the power supply on, or low for 100 ms to turn it off. Those pulse times match the captures I took earlier. BTW, I also added the 3.3k pullup from /PFW to the power supply's 5V rail through a diode just like the IIsi logic board has, for keeping the power supply turned on after the 5V rail comes alive.

I'll eventually find a super-low-power 5V MCU and add that to really finish this thing up and remove the dependency on my SIMM programmer.

The power supply wouldn't stay on at first and I quickly realized I needed to add a load to give it something to actually power, so I used a bunch of 50W, 6 ohm power resistors. Got a kit from Amazon with way more than I needed. I put 3 of them in parallel across the 5V output rail (2 ohms = 2.5 amps = 12.5 watts) and 4 of them in series across the 12V rail (24 ohms = 0.5 amps = 6 watts).

I haven't scoped it yet, but interestingly enough, the power supply behaves perfectly fine in my tester. It turns on with the 200 ms high pulse, and off with the 100 ms low pulse. When I turn it off, it stays off.

So now the big question is: is there something wrong with my IIsi's logic board, or is my tester inaccurate because it doesn't add all the extra capacitance onto the 5V rail that the logic board has? Maybe I should pay closer attention on the logic board to where exactly /PFW goes and look for any board damage or leakage along the way.

BTW, I think I'm pretty much done with the IIsi riser board schematic. I still need to determine the values of all the ceramic caps, measure the trimpot to make sure I have the right resistance, and try to identify a suitable replacement for the =B (mine's marking looks more like _B) diode that goes between the 12V and 12V-always-on rails.

 

[dougg3]

I haven't had time to tinker further with troubleshooting, but I thought I'd share the progress I've made on a IIsi power supply riser board recreation. This is probably the weirdest PCB I've ever laid out. I took a picture of the original and then traced everything on top of it. The pad and trace sizes are very inconsistent, so I had to do a lot of manual manipulation of the pads. Each trace ended up just being a polygon copper fill that I drew by hand with the original picture underneath as a reference. It's not perfect, but it's pretty darn close to the original design. Once it's all finished, I'll share the KiCad schematic for it too.

I don't even think my riser board is the problem at this point...but oh well. I'm sure this will be useful for someone eventually.

 

[dougg3]

Update: My IIsi power supply riser board recreation project is pretty much complete. I soldered up the first one today using a stencil, solder paste, and hot air. (Added the two big caps afterward by hand). All parts on it are brand new from Digi-Key except for the obsolete ones: M51977FP (eBay) and M51944BML (transferred from the original board). I'm still waiting for some M51944BML chips to arrive so I can confirm it's really the correct part, but I'm 99% sure I have it right.



Yes, this is a brand new replica of the IIsi power supply daughterboard. Probably the first new IIsi daughterboard manufactured in over 30 years!

This will be made available for anyone who really wants to replace a nasty leakage-damaged board. I will put the PCB design and schematic up on GitHub once I've fully validated it, and for convenience I will also offer them for sale totally populated, as well as bare boards for people who just want to transfer their components over to a clean board. I may also consider doing boards that have everything except the M51977FP and M51944BML populated, and people can just transfer those over from their old board. That might be a little tricky if you don't have a way to test the power supply independently though, because you need to adjust the trimpot so the 5V rail is good and not accidentally overvolt your IIsi.

After all that...my IIsi still doesn't power off correctly even with my brand new replica board. Haha. So I've definitely eliminated the riser board as the cause! I still need to find some time to see if I can find anything funky going on with the logic board.

 

[croissantking]

Update: My IIsi power supply riser board recreation project is pretty much complete. I soldered up the first one today using a stencil, solder paste, and hot air. (Added the two big caps afterward by hand). All parts on it are brand new from Digi-Key except for the obsolete ones: M51977FP (eBay) and M51944BML (transferred from the original board). I'm still waiting for some M51944BML chips to arrive so I can confirm it's really the correct part, but I'm 99% sure I have it right.

View attachment 90252

Yes, this is a brand new replica of the IIsi power supply daughterboard. Probably the first new IIsi daughterboard manufactured in over 30 years!

This will be made available for anyone who really wants to replace a nasty leakage-damaged board. I will put the PCB design and schematic up on GitHub once I've fully validated it, and for convenience I will also offer them for sale totally populated, as well as bare boards for people who just want to transfer their components over to a clean board. I may also consider doing boards that have everything except the M51977FP and M51944BML populated, and people can just transfer those over from their old board. That might be a little tricky if you don't have a way to test the power supply independently though, because you need to adjust the trimpot so the 5V rail is good and not accidentally overvolt your IIsi.

After all that...my IIsi still doesn't power off correctly even with my brand new replica board. Haha. So I've definitely eliminated the riser board as the cause! I still need to find some time to see if I can find anything funky going on with the logic board.

Beautiful. Thanks so much (in advance) for contributing this.

 

[David Cook]

Update: My IIsi power supply riser board recreation project is pretty much complete.

Wow! Amazing. By the way, there was a revision to the board. They moved a capacitor. Maybe to help pick and place of the IC? Very questionable to run a trace under a tiny SMT, though.

 

[David Cook]

So I've definitely eliminated the riser board as the cause! I still need to find some time to see if I can find anything funky going on with the logic board.

Fascinating.

 

[dougg3]

Beautiful. Thanks so much (in advance) for contributing this.

Thank you! The Bomarc schematic (which is fantastic and I heavily relied on for this project) has a few mistakes so it will be nice to have a fully correct schematic available for it.

Wow! Amazing. By the way, there was a revision to the board. They moved a capacitor. Maybe to help pick and place of the IC? Very questionable to run a trace under a tiny SMT, though.

Thanks! I remember seeing your pictures of the two revisions while I was doing all my research. Very interesting. It looks like the circuit is identical. I wonder what their motivation was for moving it. Those components are pretty close together. I bet you're right. It was super close to the IC. In my 3D models of the board it looked like they were going to overlap. Another possibility is that the capacitor all the way to the right was super close to it. Maybe they were having trouble during manufacturing with the solder bridging those two together.

My next project is to put the IIsi/IIci power supply tester apparatus I cobbled together into a nice enclosure. It's just a mess of stuff floating around right now. I got it working with an ATTINY13A powered by the 5V standby rail and two buttons: one for turning it on, one for turning it off. I'm a little nervous about using my only IIsi power supply for testing all these riser boards as I make them, but designing a special tester just for the riser boards seems like it would be a lot of work.

 

[dougg3]

Here's a link to my final IIsi PSU daughterboard replica design on GitHub. Feel free to do whatever you want with it; most of the credit goes to Bomarc honestly. I just did some sleuthing to figure out a few things Bomarc didn't, and fixed a couple of errors in their reverse engineered schematics.

GitHub - dougg3/iisi-riser-board: Replica of Macintosh IIsi power supply riser board model AP41S

Replica of Macintosh IIsi power supply riser board model AP41S - dougg3/iisi-riser-board

github.com

I'm still working out the logistics of how the heck I'm going to build more of these things and offer them as a convenience for people.

BTW, I confirmed last night that M51944BML is indeed the correct identification for the part marked as 44; it's a reset IC, not a transistor. Here's a brand new one from Utsource populated on my new board and it works great. So I've successfully built a new riser board that uses all brand new parts except for the two Mitsubishi/Renesas ICs which I definitely know the identity of.

Ignore my god awful soldering/positioning on that part; I also sort of burned the PCB because I was being careless. I'll do better on actual boards I put together. But yeah, you can tell it's marked exactly like the factory parts on the original boards so I'm confident I figured it out.

 

[David Cook]

Ignore my god awful soldering/positioning on that part; I also sort of burned the PCB because I was being careless.

I can't begin to tell you the number of times I've taken pictures of some repair that I performed, only to see it up close and think "There's no way I can post that!" So, I appreciate your acknowledgement that sometimes it doesn't have to be pretty to illustrate a point.

 

[dougg3]

I can't begin to tell you the number of times I've taken pictures of some repair that I performed, only to see it up close and think "There's no way I can post that!" So, I appreciate your acknowledgement that sometimes it doesn't have to be pretty to illustrate a point.

Haha, I know the exact feeling! Thank you!

Here are a few more updates on this project. I built up seven more riser boards over the weekend. It was a nice excuse to pull out the toaster oven and Reflowster. Reflowster was a Kickstarter project I backed in 2014. It does the temperature control for the soldering without requiring you to modify the actual oven. You just turn the oven on full blast, and a relay in Reflowster turns it on and off as needed.



I did have to make a few manual fixes for components that moved during reflow, but it was so much easier than building these all by hand. The first two boards I made from this batch tested out fine (pictured on the left), and the third one, when I powered it on, went kablooey! Visible spark from the PSU main board, and the main fuse (F101) blew.



This demonstrates the risk I'm taking on by trying to build more of these. I wish I had an easy way to test these out before putting them into the power supply. I guess I could try to rig something up that provides the rest of the circuit, but it would be a lot of work.

I don't see any obvious defects in the soldering on the board, but I also had to source parts from Utsource, so it's possible it's a bad/fake PWM controller IC. I'm really, really hoping that the bad riser board didn't take anything else out with it on the power supply. Crossing my fingers that the fuse protected everything else. So now, I'm stuck waiting for more slow blow, 3.15A, 250V fuses to arrive from Digi-Key. I don't trust no-name fuses from Amazon.

On a more positive note, this mishap gave me an excuse to finish putting my IIci/IIsi power supply tester into a nice aluminum case. I 3D printed a few brackets for holding things onto the lid, drilled some holes, and used a hand nibbling tool to cut larger holes. I'm pretty proud of how this turned out! It's the first time I've made anything like this.

 

[shirsch]

Nice load box!

 

[theletter0xff]

Amazing work, thank you for your this!

several of those daughter cards caused the population of my currently-beyond-repair heap to grow; do you have any completed boards available?

What would be the minimum circuit needed to test the cards without a power supply? Does the spacing match any ZIF connectors?

Do you have a write up of exactly what you did to create the soft-power on-off emulator? I want to copy your PSU tester, it would be a much nicer solution than a paperclip.

I made a stack of ATX to apple motherboard adapters. Changed the end of ATX extension cables to the correctly wired mini-fit jr and KK series.
To test the PSUs, I built wiring harnesses with the matching mini-fit and KK receptacles and standard molex connectors to use old drives as load.

 

[dougg3]

Amazing work, thank you for your this!

Thanks! Unfortunately, I don't have things fully figured out yet to the point I'd be comfortable selling completed boards. I was able to eventually repair my power supply briefly, but it's dead again right now. Q101 and D104 had blown when I installed the faulty riser board. Q101 got really hot when it blew and it also melted some of the heatshrink on one of the nearby transformers, but as far as I can tell the transformers are still good.

I replaced Q101 and D104, and had the power supply working correctly with my tester box, and then I installed it in the IIsi and it would go into protection immediately when I tried to power it on. That is still a bit of a puzzle to me. I stupidly tried another known good riser board, and it blew again. I think maybe all the riser card pins weren't in the socket properly, but I'm not sure. I haven't had the motivation to continue playing with the power supply after that, but I've come to the realization that it's impossible for me to offer completed boards for sale without creating some kind of tester device that can prove all the different parts of the circuit are working correctly, and given the high voltages it's controlling, it may be too much of a risk for me to ever want to take on offering completed boards for sale.

If you would like any bare boards to move components onto, I would be happy to send a few your way though. I have way too many of them. I'm confident that the PCB is a good replica of the riser card revision that was in my IIsi PSU's (there is also a different revision that moved a cap or two around I think).

What would be the minimum circuit needed to test the cards without a power supply? Does the spacing match any ZIF connectors?

This is exactly the question I've been asking myself, and the complexity is another contributing factor that kind of put the project on hiatus for me for now. The card looks at all three voltage rails for overvoltage protection, as well as looking at both the +12V and +5V rails together for voltage regulation. There's also an interesting current limiting circuit, as well as another weird circuit capable of forcing the power supply to shut off based on certain feedback that I don't fully understand. That little board is responsible for quite a bit of stuff, and I think in order to be fully confident about a newly built board, it would need a tester capable of simulating the power rails and varying their voltages.

The spacing is 0.1" pitch, which is pretty standard. ZIF connector would likely be more robust than the simple pin header socket I soldered onto my PSU's main board.

Do you have a write up of exactly what you did to create the soft-power on-off emulator? I want to copy your PSU tester, it would be a much nicer solution than a paperclip.

I don't have a write-up yet (maybe I'll do a blog about this whole thing eventually), but it does involve a simple microcontroller program for a super-low-power AVR that powers itself from the 5V standby rail and looks at two buttons for on and off control. I'd be happy to share the code. I've been meaning to put it up on GitHub anyway, so thanks for the reminder!

Basically, what I mentioned in this post is a decent summary about the power resistors I put on the 12V and 5V rails for a dummy load. Then just some cheap voltmeters from Amazon (being sure to wire the -5V rail's voltmeter backwards). The /PFW pin is pulled up to the 5V output rail through a 3.3k resistor and diode just like on the IIsi's motherboard (R115 and D3), which keeps /PFW pulled up after the power supply has turned itself on. Then, I have /PFW also wired up to the AVR so it can drive it high or low briefly to turn the supply on or off, similar to how the Egret does it.

I made a stack of ATX to apple motherboard adapters. Changed the end of ATX extension cables to the correctly wired mini-fit jr and KK series.
To test the PSUs, I built wiring harnesses with the matching mini-fit and KK receptacles and standard molex connectors to use old drives as load.

Nice! Yeah, you could probably just use old drives instead of my crazy power resistors. Honestly, I think that having a higher load attached to my PSU might be dangerous for initial testing of the cards. I should probably be using a minimal load that is still capable of allowing the regulation to occur.