Another Indigo2 psu died...

marmotta

Member
Feb 22, 2021
60
6
8
Is a very strange… without pin 3 (yellow wire with 0,9v) the psu start and I have a perfect 5v, 3,5v but the 12v start at 3v and slow raising at 5v :unsure:

I have tried to put the trimmer and various resistors to this wire but I no see any difference… but now I have always 0,4v and the psu always start! With limited 12v at 4/5v

Perhaps this wire need more voltage and not less?
 

marmotta

Member
Feb 22, 2021
60
6
8
Yes! I near to goal!

I have tested the pin 3 and 4 on working 8001

The correct voltages is:

Pin 3: 17v
Pin 4: 15v

On not working 8002 I have:

Pin 3: 0,4/0,9v
Pin 4: 15v

This voltage start from LV board and go to HV (only the 18v red wire is start to HV to LV from a little transformer). I have a issue on 12v feedback (pin 4 is 5/3,5v feedback, pin 3 is 12v feedback)
 
  • Like
Reactions: Elf

marmotta

Member
Feb 22, 2021
60
6
8
Now the hard part is trace the track of pin 3… the signal start from the fully populated part of pcb. Multilayer pcb not help.
 

Elf

Storybook
Feb 4, 2019
469
118
43
I would use a multimeter on continuity mode with one side held on that pin and brush the other probe across things.
 

marmotta

Member
Feb 22, 2021
60
6
8
I would use a multimeter on continuity mode with one side held on that pin and brush the other probe across things.
Yes, is the only way. But the track is very winding. The signal pass across some transistors, IC, and sometimes splits.
 

marmotta

Member
Feb 22, 2021
60
6
8
I try to rebuild the logic of this psu... I have an issue on 12v rail or on your feedback.

The 12v wires is connected to the fully IC populated zone (I think for feedback), and is connected to the two MC34166 voltage regulators directly to the pin 4. Is strange because in the data sheets the pin 4 is not the output but the input :unsure: ....then the 12v is connected to the big 3300 capacitor near the 5v regulator and square HV connector. In not clearly understand the logic, but I think the next pass is verify the two MC34166.

The other suspect behavior of psu is the slow raising of 12v.... why it start at 3,5v and after 1 or 2 minutes reach at about 6v?
 

marmotta

Member
Feb 22, 2021
60
6
8
This psu drive me crazy! I have tried to disconnect from 12v the two mc34166. Nothing changed… what are they for?

The 12v line is taken directly to HV board, no regulation on LV board… one of four wire of square connector go to a BYv32E rectifier on the big heatsink, 3300uf cap, and to 12v output.

I think I understand that the problem is the lack of driving the 12v from the LV card to the HV. But I don’t know how it happens. Reduce or raise the voltage on pin 3 don’t help. Trace the path is very hard, to many components, hidden tracks on pcb and I not understand where start…
 
Last edited:

marmotta

Member
Feb 22, 2021
60
6
8
I have a question: the current of 12v is taken from a transformer of HV board. This current is already a DC… what is needed another rectifier on secondary stage? (LV board)
 

Elf

Storybook
Feb 4, 2019
469
118
43
Hm, I'm not sure I understand the question exactly... But, reading back through the things you have been tracing, it sounds like the 12V might be a bootstrap voltage from the HV side to power the switching logic? It is common for these sorts of designs to have a very low current, non-switching supply that is active from the start and only roughly rectified and filtered (e.g. a diode and capacitor) which powers the switching logic before the rest of the supply starts up.

I can feel the frustration though. As you get into these supplies at the troubleshooting level you will need a strong understanding of switch mode power supplies (forward converter topology), good test equipment (i.e. a good scope with a high-voltage differential probe, maybe a few channels worth of HV diff probes), and a lot of patience tracing out a very large number of discrete parts.

I have only tried to characterize the supplies externally because being older designs with a lot more discrete components they require a lot of reverse engineering. My feeling was that replacing them with a functional equivalent was easier than troubleshooting them :p
 
Last edited:
  • Like
Reactions: chulofiasco

marmotta

Member
Feb 22, 2021
60
6
8
@Elf

I have some patience and stubbornness, but less switching knowledge and I not have a good English :giggle:

My question is probably trivial… on switching psu the ac current is converted to dc by a rectifier, then filter cap, and go to transformers…at transformer out I have a dc high voltage current. Why on switching psu is present another rectifier after the transformer? …I think is for voltage regulation but I not understand how it work.

On the repair I as on impasse, I have tried all, I have traced all, without success. Too many components, and to many tracks… I have tried to cut a voltage regulation circuit (I presume… it connect the logic to rectifier of 12v on LV) and I have 19v, reconnected the logic to rectifier I already have 4v with slow rising to 6/7v. All test is without pin 3 (presume the feedback). With pin 3 connected the psu don’t start at all.

The psu make me angry because with pin 3 disconnected all voltages are perfect (3,5v 5v -12v -5v) the only issue is 12v and I sure is a stupid reason but I not able to find it :confused:
 

marmotta

Member
Feb 22, 2021
60
6
8
Nothing resolved, but some news.

The pin 3 is not really the 12v feedback, the pin 3 is a sort of safe switch to turn off the HV board! Without pin 3 the psu is free to startup, skipping the overvoltage or undervoltage issue. The key is the chip UC3903, it have a reference voltage of 2,5v, it take four rails (with reduced voltage) and if all four is 2,5v it put the pin 14 to high and the psu stay up, if one of this four input is not 2,5v the IC3903 put pin 14 low, the signal go to pin 3 of brown connector and HV board turn off! This is the explanation because some Indigo2 psu try to start and then off... removing pin 3 make the psu to force on for diagnostic (can be dangerous but is the only way).

Understood this I sure have an electrical and not feedback/logical issue! To be sure I have removed the pin 3 to my working 8001, and 12v is perfect and stable.

On the IC3903 have three 2,5v and one 1v.... and it slow rising to 1,7v because it follow the my 12v rail, which start at 3,5v and slow rising (in many minutes) to the max of 7v.

Now for those with good knowledge on switching can be more easily. HV board is excluded (this board work with 8001 LV board), the rectifier on 12v line of LV board is excluded (I have replaced it without success).... what can it be? What can make the 12v rail start at 3,5v and after 2 minutes is at 5v, and after 4 minutes at 7v? One component sensitive to temperature? :unsure:

IC3903 Datasheets
 
Last edited:

Elf

Storybook
Feb 4, 2019
469
118
43
My question is probably trivial… on switching psu the ac current is converted to dc by a rectifier, then filter cap, and go to transformers…at transformer out I have a dc high voltage current. Why on switching psu is present another rectifier after the transformer? …I think is for voltage regulation but I not understand how it work.
Hm, well, there are some steps missing in there and perhaps some mistaken assumptions. For example, there is usually some MOSFET switch bridge, between the high voltage DC and the transformer, that usually pulse width modulates it according to how much power it wants to generate. The controller provides the PWM signal and MOSFETs and perhaps a gate driver perform the actual switching. This high frequency switching also means that the switched product of this high voltage stage now has non-zero frequency components to it (i.e. is AC) and can pass through the transformer. DC will not pass through a transformer but will only heat the coils. Likewise, anything emerging from the transformer is not DC, only AC components pass, thus the rectification stage.

If you want to know more, study transformers ( https://www.electronics-tutorials.ws/transformer/transformer-basics.html ) all the way through to forward converter topology SMPS circuits ( http://ww1.microchip.com/downloads/en/AppNotes/01114A.pdf ). But this might be considered quite complex without prior electronics knowledge. There are very large textbooks written on understanding power electronics.

Now for those with good knowledge on switching can be more easily. HV board is excluded (this board work with 8001 LV board), the rectifier on 12v line of LV board is excluded (I have replaced it without success).... what can it be? What can make the 12v rail start at 3,5v and after 2 minutes is at 5v, and after 4 minutes at 7v? One component sensitive to temperature?
Maybe something went high resistance and is slowly charging up a capacitor? A scope with high voltage differential probes would make troubleshooting easier, I think.
 

marmotta

Member
Feb 22, 2021
60
6
8
I have two identical psu on the desk, same capacitors, two sure working HV boards… Is it possible that I don't understand what's wrong? Ok I not have good diagnostic tools, but with two psu, it should be easy… what test can i do?
 

marmotta

Member
Feb 22, 2021
60
6
8
On 12v rectifier circuit I see two pnp transistors, one have a hfe of 291 and another is 338 (same model). Is a normal tolerance or can be an issue?
 

weblacky

Member
Jan 13, 2020
91
23
8
Seattle, WA
I know I'm repeating myself, but Elf just repeated the same advice I gave earlier, which I cannot tell if you're ignoring or unwilling to look into. Resistance likely changed in one or more components. You're fixated on Capacitors, I'm unsure why. You had a problematic PSU when you started, simply replacing the caps isn't a guarantee of a fix. Because it starts up, likely it is a small issue along a logic path or along the feedback path.

Either way those heavily USE RESISTANCE. A change in resistance is bad (just as a change in capacitor ESR is bad). You've done everything you can to avoid this. I think it's pretty clear you either need to start measuring resistors (start with LARGE power resistors) or you need to research the data sheets of the original caps and double check your ESR for the new ones you selected. That is to say, either you have resistors out of tolerance or you've changed the resistance yourself with poor replacement capacitor selection.

The easiest way to start would be to start taking DIODE and OHM measurements (drained PSU, while off) of an identical & working PSU and measure several/many points to local LV PCB ground. Then if you confirm the Cap ESR with the originals, you'll likely find your culprit because your impedance is different from your known-working PSU to your problematic PSU.

You need to perform comparison measurements, because something has an incorrect value. To find it, you need to know the CORRECT value.
 

marmotta

Member
Feb 22, 2021
60
6
8
@weblacky

Thanks for your suggestions, but is not so easy… pc switching psu it is one thing, this is another. I have some days already excluded the capacitors, the two psu have the same identical caps (one work, other not). Can be a resistor? Possibly, but this psu have only a two or three classic resistor near the rectifiers (one for 12v) and two panel resistor… seems to be ok, but I try to remove and better verify. Then it has dozens, hundreds of small smd resistors… is impossible to verify all. I verified the voltages on various point of pcb on working psu and on not working psu, and the voltages on 12v zone is totally different! Nothing is the same (some are lower, some are higher), this psu it’s a mess o_O …if only there was a schematic or documentation 😔
 

weblacky

Member
Jan 13, 2020
91
23
8
Seattle, WA
I’ve been in these PSUs, so yeah it’s not a modern PC ATX, not a big deal. All those small resistors? Impossible? I think you’re just impatient. If you had a duplicate you can measure each spot at the same time (two meters, two boards, side by side) checking the readings as you went. I doubt it would even take an hour to check all visible resistors.

I’m not asking you to remove each component, if that’s why you say it’s impossible. Resistors rarely go less then spec. They normally go higher or open.

if you use tweezer probes or just another set of hands. You can touch where you need to get measurements.

granted, my equipment and probes are design to perform this, so it’s easier for me.

The logic board (board on end cap between HV and LV PCBs is likely the first place I’d compare.

I’m literally suggesting you use two boards, side by side comparison. To make first attempt even easier you could clip black to grounding plane if each PCB (one per meter) and then use both your hands to operate the red probes of each multimeter to touch the same place on each board and compare readings (if you don’t want to prerecord values).

Don’t remove things unless you have to. Just use your meter to measure diode and ohms modes and look for differences that are greater than say 10%?

Comparison is the easiest way to find faults, so instead of trying to understand how it works, start measurements to find noticeable differences that shouldn’t be.
 
  • Like
Reactions: Elf

About us

  • Silicon Graphics User Group (SGUG) is a community for users, developers, and admirers of Silicon Graphics (SGI) products. We aim to be a friendly hobbyist community for discussing all aspects of SGIs, including use, software development, the IRIX Operating System, and troubleshooting, as well as facilitating hardware exchange.

User Menu