[CB_HK]

UPDATE

I have solved my power supply issue by replacing the faulty components with a donor PSU. If you are experiencing issues with your Crimson power supply, have a read through this thread and see if your problems are similar. In the end my main issue was a power supply that no longer provided the necessary voltage to function, along with capacitors that had aged and begun to leak. I needed to find a donor PSU--a PowerOne SPM5 with an A2 module (to replace the faulty 5V 170Amp module in the original one, along with the backplane and rectifier). Ultimately the swap solved my issue. Even so, combined with heavy cleaning (taking the Crimson down to the metal chassis and cleaning EVERYTHING), checking for any small issues such as unseated cables, loose RAM, broken wiring, etc. has helped restore my Crimson to excellent working order.

Thank you to everyone who helped me solve this problem. I am incredibly thankful for this community and its wealth of knowledge!


_____________________________________________________________________________________
Original Post

I'm in need of assistance with my Crimson. I went to power on the system and noticed a few issues. This led me down a path of troubleshooting, which I'll list below. I'm hoping that this issue can be easily resolved but I'm also concerned that there's a bigger problem that's hiding in there.

1. Turned rear circuit breaker to on. Front power switch set to off.
2. System turns on while front power switch is set to off.
3. Fans do not spin in cooling tray. Power supply fans spin.
4. IP17 board shows 8 solid red lights.
5. No video output.
6. Power and Fault light remain lit.
7. Pulling out PS3 and examining Hex Code shows "F". Overvolt lights are lit on 12v and 5v. Overtemp light is lit without temp circuit breaker popping. Undervolt lights come on whenever I toggle the front power switch to On. Changing from On to Off and back again doesn't seem to affect anything--the computer stays on until the rear circuit breaker is turned off.
8. Swap IP17 with other boards known to work. Same exact results.

Since the machine was booting prior I'm concerned that something related to the power supply has decided to go out. No magic smoke as far as I can tell, but that doesn't mean there isn't something wrong regardless.

Considering I've had the machine for a very short period of time I'm extremely keen on getting this issue resolved. I'm looking for any insight into what the 8 red lights mean (I can find a few posts on the net that refer to other light combinations but not eight), and if anyone has had to troubleshoot a similar issue.

Thank you all for any help you can lend!

[jan-jaap]

I'm in need of assistance with my Crimson. I went to power on the system and noticed a few issues. This led me down a path of troubleshooting, which I'll list below. I'm hoping that this issue can be easily resolved but I'm also concerned that there's a bigger problem that's hiding in there.

1. Turned rear circuit breaker to on. Front power switch set to off.
2. System turns on while front power switch is set to off.
3. Fans do not spin in cooling tray. Power supply fans spin.
4. IP17 board shows 8 solid red lights.
5. No video output.
6. Power and Fault light remain lit.
7. Pulling out PS3 and examining Hex Code shows "F". Overvolt lights are lit on 12v and 5v. Overtemp light is lit without temp circuit breaker popping. Undervolt lights come on whenever I toggle the front power switch to On. Changing from On to Off and back again doesn't seem to affect anything--the computer stays on until the rear circuit breaker is turned off.
8. Swap IP17 with other boards known to work. Same exact results.

Since the machine was booting prior I'm concerned that something related to the power supply has decided to go out. No magic smoke as far as I can tell, but that doesn't mean there isn't something wrong regardless.

Considering I've had the machine for a very short period of time I'm extremely keen on getting this issue resolved. I'm looking for any insight into what the 8 red lights mean (I can find a few posts on the net that refer to other light combinations but not eight), and if anyone has had to troubleshoot a similar issue.

Thank you all for any help you can lend!

So, (#4) you've got LEDs lit on the IP17 board with the front on/off switch in the 'off' position ?? That's unexpected. This is strictly from memory, but I'm pretty sure that shouldn't happen. There's no such thing as as a standby or aux power supply to these systems.

There are two types of PSUs used in PowerSeries/Crimsons. The older systems used PowerOne SPM5 modular power supplies ("silver brick"), the newer systems used a Cherokee ("black brick"). They have the exact same specifications (a total output of 1050W), even if UseNet archives can claim different. This is because the PowerOne chassis is modular, and the PSU chassis has a limit of 1500W, but the modules installed total only 1050W.

Anyway, what's supposed to happen depends on the PSU type installed. You's have to remove the plastic from the back and the metal sheet underneath. Look for a silver or black PSU above the power inlet. The PowerOne SPM5 will spin the fan inside the PSU when you engage the breaker switch, the Cherokee doesn't. Sounds like you have the PowerOne. At this moment, the PSU does not supply *any* voltage to the system. The power switch on the front works like an inhibit signal to the PSU. Toggle that and the system comes alive (or it should at least).

If you have overvoltage LEDs lit, I would switch off the system immediately to avoid damage to the electronics. There's no software controlling that stuff, it's all hardware and there's probably a PSU fault. To diagnose, you could either remove/disconnect all electronics from the chassis and try again, but personally I would remove the PSU from the system and do any testing on a workbench and not put the internals of the system at risk until you've got confidence in the PSU. It is possible to start up the PSU outside the system.

There are also voltage measurement points on the controller PCB with the hex display. IIRC, the hex display counts down from 'F' to '0' during POST, and then alternates between '0' and '1' when the system reaches the PROM monitor. The codes are somewhat described here: http://archive.irix.cc/thisoldsgi/#profledcode  'F' probably means the CPU is stalled.

I've had to do some repairs to PowerOne PSUs over the years. I posted some photos here: https://www.vdheijden-messerli.net/sgist...wersupply/ and https://www.vdheijden-messerli.net/sgist...-powerone/  At some point I figured out that since these things are modular, you don't necessarily need to repair them at the component level, you can simply replace modules. Search eBay for 'powerone spm5': it's full of them. Of course they will have different modules, with the wrong voltages or current specifications, installed. However, they all share the same input module (300VDC rectifier/filter) and backplane, and that's where mine developed faults.

[CB_HK]

So, (#4) you've got LEDs lit on the IP17 board with the front on/off switch in the 'off' position ?? That's unexpected. This is strictly from memory, but I'm pretty sure that shouldn't happen. There's no such thing as as a standby or aux power supply to these systems.

There are two types of PSUs used in PowerSeries/Crimsons. The older systems used PowerOne SPM5 modular power supplies ("silver brick"), the newer systems used a Cherokee ("black brick"). They have the exact same specifications (a total output of 1050W), even if UseNet archives can claim different. This is because the PowerOne chassis is modular, and the PSU chassis has a limit of 1500W, but the modules installed total only 1050W.

#4 is almost more concerning to me than anything else. It happens whether or not there are any boards installed (I pulled them to avoid damage earlier) and whenever power is switched on in back. The main power LED and fault LED will light as though the system is booting and then just hang there with a fault code. I pulled the PS3 out and there doesn't appear to be any damage to it that I can see. The voltage LEDs on the PS3 are not solid either, they have a faint glow to them almost as though they're lighting up to indicate an issue but at the same time it seems ambiguous. I hope this is related to the PSU acting wonky and ignoring commands from the PS3, and not a faulty PS3.

Luckily it appears I have a PowerOne PSU. I already have the system stripped down because I know this is going to be one of those things where I'm going to have to pull parts. Looking on eBay as you suggested showed a lot of them. Obviously not specific to the Crimson (a lot of different power boards installed it appears) but you mentioned that any SPM5 should work if it's the backplane having the issue, correct?

I'm going to basically remove most everything from the case and make my way through cleaning everything in order to be sure that all parts are in good working condition. That PSU is going to come out and I'll have to check what's going on with it. Do you have any documentation from when you repaired yours?

Thanks, jan-jaap! Always appreciate your insight.

[jan-jaap]

Do you have any documentation from when you repaired yours?

What I have is mostly in that second link to my site: https://www.vdheijden-messerli.net/sgist...-powerone/

It has some links to material on the PowerOne site which is long gone, but archive.org should have it. Otherwise I still do. This is the functional diagram of the PSU:

As soon as you engage the breaker, the input module will put 300VDC on the internal 'bus'. The sync & bias supply is always active and the fan inside the PSU spins. "system control" is connected to the front on/off switch, and will trigger the control input to the output modules.

Now, in your case the system control circuitry is definitely malfunctioning because the outputs don't listen to the front panel on/off switch. This is on the backplane. I would also check the 300VDC bus for excessive ripple (dried out rectifier caps), though you may need to test the PSU under load to make sure. If the rectifier is half dead and is feeding the PSU bus with a 300V saw-tooth I can imagine that the backplane and output modules will misbehave.

You've got overvoltage faults on *two* independent modules: the 5VDC is a big "double width" module, and a second module supplies +/-12VDC and -5VDC. I wonder if these faults are real or maybe will disappear with a good backplane and/or rectifier.

When probing around the internals, beware that large sections are not galvanically isolated from the mains. This includes the 300VDC bus! You need an isolated high voltage probe if you want to attach an oscilloscope.

When shopping for a suitable donor PSU, it helps to know how to decipher the product code of the PSU. The type in the SGI is SPM5ADAELBS180, or SPM5-AD-AE-LB-S180. SPM5=chassis type, AD and AE are the modules installed and S180 is the code PowerOne assigned for SGI. The AD module is the 5V@170A module, and the AE is +/-12V and -5V. One of the PDFs on my page has a full list of codes for all modules. IIRC, A1 or A2 is a suitable 5V module, but newer. In short: when choosing a donor, it helps if at least one module installed is usable

At some point I also managed to figure out roughly how old an SPM5 PSU was based on the stickers and serial numbers. They built them for ~ 15 years, from the late eighties to ~ 2004 so given a choice you want the newest. This, I forgot though, I'm afraid.

Finally, there's a bank of dip switches on the backplane which configures the thing. Make sure you set them like the original PSU before you assemble the replacement. Also, the SGI (S180) version includes an extra PCB which effectively sites between the control input and the backplane. You've got to keep that, of course.

---

Oh, one more thing: these PSUs are not auto switching between 110VAC and 220VAC mains. You have to configure them with a jumper wire on the terminal block where the mains enter the PSU. Make sure it's wired correctly. 220VAC input to a PSU wired for 110VAC is certain destruction, the other way around I don't know.

[mchartmann]

I would make sure all contacts are cleaned with a good electronic parts cleaner that leaves no residue. I have to do this to several systems and also 91 percent alchohol on memory contacts and a q tip have worked wonders. This has revived most of my systems that failed to boot. Of course it could be hard failure but do basics first.

[jan-jaap]

I would make sure all contacts are cleaned with a good electronic parts cleaner that leaves no residue. I have to do this to several systems and also 91 percent alchohol on memory contacts and a q tip have worked wonders. This has revived most of my systems that failed to boot. Of course it could be hard failure but do basics first.

I'm sorry, but "basics" in this case means make sure you don't fry the electronics with a faulty PSU. The voltage and temperature monitoring in these systems is completely solid state, no software involved. Same for the way it's supposed to react to the on/off switch. None of it can be fixed by cleaning the memory contacts.

FWIW: A PowerSeries/Crimson can do the initial parts of it's POST without touching main memory. It starts with a brief CPU and L1 and L2 cache test and initializes the UART of the console, all using only ROM and CPU registers. This should change the hex display from 'F' to a lower value and some of the LEDs on the CPU card changing state. None of this happens here.

[CB_HK]

I would make sure all contacts are cleaned with a good electronic parts cleaner that leaves no residue. I have to do this to several systems and also 91 percent alchohol on memory contacts and a q tip have worked wonders. This has revived most of my systems that failed to boot. Of course it could be hard failure but do basics first.

While I agree with jan-jaap that this is a power supply issue, you make a good point about ensuring everything inside is clean as well. There is a good layer of dust with even a few small dust bunnies inside. I'll be cleaning everything with compressed alcohol, compressed air, and for the contacts I'll be layering in some de-oxit. Even though the issue is not necessarily with just the contacts, I definitely think it's a very good idea to keep the inside clean and well maintained. At the moment this Crimson is in need of some attention with that regard.

Thank you for pointing this out. I definitely think this will solve some issues for folks who are having problems. Just not my specific case unfortunately. But your advice will be heeded and my Crimson will be shiny and clean inside and out!

---

Do you have any documentation from when you repaired yours?

When shopping for a suitable donor PSU, it helps to know how to decipher the product code of the PSU. The type in the SGI is SPM5ADAELBS180, or SPM5-AD-AE-LB-S180. SPM5=chassis type, AD and AE are the modules installed and S180 is the code PowerOne assigned for SGI. The AD module is the 5V@170A module, and the AE is +/-12V and -5V. One of the PDFs on my page has a full list of codes for all modules. IIRC, A1 or A2 is a suitable 5V module, but newer. In short: when choosing a donor, it helps if at least one module installed is usable

Here's what I've been able to locate as a possible donor (it's being sold as refurbished and tested):





From what you've written so far it looks as though I should be able to use the first stage (AC to 300V DC), the backplane, and possibly the A2 module. Would you agree with my assessment? I noticed the A2 only goes to 150 Amps instead of 170 as in the original. I'm guessing that shouldn't be a major issue, but I want to be sure before I decide to use that component.

My plan would be to use those three components to swap out what's in the Crimson's power supply and keep and use the AE board (5/12 volt) that's already in there. Thoughts?

[CB_HK]

Update on the original power supply:

I removed the unit from the Crimson and disassembled it to check the components we've discussed. I don't see any obvious issues on the backplane aside from it being dirty and covered in some sort of potting material. The 5v/12v board seems to be in good condition though I think I may do a re-cap for the three electrolytic capacitors on it, just to be safe since this board can't easily be replaced. The 5v/170amp board on the other hand is showing signs of electrolyte leakage and corrosion that has started to eat into the soldering joints. Hopefully the newer 5v/150amp board is an adequate replacement, though if push comes to shove I could re-cap the original board.

Backplane (I'm assuming the DIP switches you mentioned are the ones located in the upper right section):


Lower portion of 5v/12v board:


Upper portion of 5v/12v board:


Here are the caps in question on the 5v/170amp board:


And this is what they're doing to the other side of the board:




I went ahead and ordered that other power supply. Hopefully it will prove to be in good condition and make it fairly straight forward to repair the Crimson's power supply.

[CB_HK]

Update:

I received the refurbished power supply as well as some replacement caps for the 5v/12v board. I was able to get everything put together, so I now have a power supply that is basically the same as the original, except the 5v board now outputs 150 amps instead of 170.

I put everything back together, triple checked to make sure everything was connected properly in the power supply and inside the Crimson. I flipped the breaker and.... nothing. I checked the inputs and made sure 115v was making its way from the outlet to the power supply and indeed it was. I unplugged the power cord, plugged it back in, flipped the breaker and... the fan started up. So now there's progress and I don't have anything else on except for the PSU. That's better than before.

Next step was to attempt power on with the PS3. I flipped the front power switch on and... nothing. I waited a moment, heard the fan start to warble a bit and then it stopped. Now I am unable to get the fan to start in the PSU no matter what.

At this point I'm guessing the fan decided to quit and from reading your write ups, jan-jaap, it appears that the fan must be in working order for the power supply to start up correctly. I'm leaning toward bad luck and a fan that has decided it's done. Any thoughts aside from that?

[jan-jaap]

I would try to troubleshoot the PSU outside the system, like this:



This PSU only has the 220VAC mains cable attached (green = earth, blue and brown are phase and neutral). If you've got 110VAC you probably have a bridge between terminals '1' and '2' on the input.

With power applied, it should spin up the fan and put 300VDC on the power bus, as shown. In this case the two output modules are installed, but even without any output modules it will spin the fan, and it's easier to reach the power bus with a multimeter (beware: not isolated from mains!!). The outputs should all be 0VDC. The pinout of the power bus:


Like I mentioned in https://www.vdheijden-messerli.net/sgist...-powerone/ if you connect #10 and #11 on the white 3x4 connector block on the input module you switch on the PSU. I use a short piece of solid "VD" copper wire for that -- the stuff you normally put in the walls. This connector is galvanically isolated from the mains, and mains power doesn't pass here, but be careful anyway. The fan will spin faster and the output modules will be activated. The output terminals will have the correct voltages even without a load attached.

In your case, I would start measuring the power bus, because that should always show 300VDC. If it doesn't, replace the input module using the donor.

If you've got 300VDC but no fan spinning, there's something wrong with either the aux/bias supply which steps down the 300VDC to 12V (could be 24 or 48V), the fan or the cables or connectors (this happened to me).

Only if both of these conditions are met does it make sense to start looking at the output modules.

Once you've got a PSU which does all of the above, and switches on and off the way it's intended to do, you can either test it under load (if you have one), or put it back into the system.