Old Compaq Armada internal PSU repair

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Well, okay, I'll have to order more parts. I'll post back when I get around to building the PWM prototype.

[stj]

Re: Old Compaq Armada internal PSU repair

optoisolator looks like a 4pin chip - it's on the edge next to a big diode

[Th3_uN1Qu3]

Re: Old Compaq Armada internal PSU repair

Where is the optoisolator? What even is that? You'll have to excuse me... I'm completely self-taught and a total amatuer, this is just a hobby for me so I'm not well-versed. I would really appreciate it if you could explain this in a bit more detail.

Little 4 pin rectangular black box the size of half a DIP package bridging primary and secondary, also called an optocoupler. Most common one is the EL817, if you see 817 anywhere that is it. Although a power brick that old may have different optos maybe in white DIP packages, some are 6 or 8 pins.

AT and ATX power supplies with bipolar transistors use current drive transformers, the EE-19 ones in the middle between the 5VSB transformer and the main transformer are what drive the bridge and also provide current monitoring and limiting. Your power brick, although old, should be a flyback topology using an opto so go look for it.

[yuuki47]

Re: Old Compaq Armada internal PSU repair

So you're saying that I should get a UC3843 and wire it up in a prototype board with the basic configuration the datasheet has. Ok, I can do that, I've made prototype boards before for a Raspberry Pi Modem Project.

But uh, this is the part I don't get quite understand:

"Lift the gate pin of the power MOSFET and the two primary side pins of the optoisolator off the board and also connect these to your board. The source pin and current limiting does not really matter for now - use the series lightbulb trick for testing as this power supply does not have PFC."

Where is the optoisolator? What even is that? You'll have to excuse me... I'm completely self-taught and a total amatuer, this is just a hobby for me so I'm not well-versed. I would really appreciate it if you could explain this in a bit more detail.

[stj]

Re: Old Compaq Armada internal PSU repair

if the black is smooth then it's epoxy and you can use heat
if it's rough then it's cement and i have read that oven cleaner with disolve it - not confirmed that though.

[Th3_uN1Qu3]

Re: Old Compaq Armada internal PSU repair

But I also noticed that upon disconnecting the PSU from the wall, that voltage starts to fluctuate between 11.3 volts and 9 volts before it settles at 9 volts.

11.3 is too low. The PWM controllers these use start up at about 13-15 volts. Either the controller is dead, or there's not enough current feeding it. I suspect the former.

Weird... I'll have to look for that resistor you talked about.

It looks like this PSU is toast though if the PWM controller is dead. I don't think those black-coated PCBs are repairable, unless I can remove that black paint somehow?

Worth a check. And yes, that black paint is removable somehow, either chemically, by heat, mechanically (scraping/sanding) or a combination of the three methods. But if I were you I'd do something easier.

I'd take a perfboard and wire in a UC3842 in a basic configuration. Or wait, make it a UC3843, that one starts up at 8.3 volts. Go download the datasheet and it is not important to understand the circuit, just the connections, which part goes where and the inputs and outputs to and from the rest of the power supply.

Lift the gate pin of the power MOSFET and the two primary side pins of the optoisolator off the board and also connect these to your board. The source pin and current limiting does not really matter for now - use the series lightbulb trick for testing as this power supply does not have PFC.

See if you can get any output this way. You should. If it works, we can elaborate about making it permanent with current sense/limiting working and everything.

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Huh, you were right! The voltage in the startup cap stays at a constant 11.3 volts. But that also seems low. It's a 32v capacitor, why would they use something this high for 11k volts?

But I also noticed that upon disconnecting the PSU from the wall, that voltage starts to fluctuate between 11.3 volts and 9 volts before it settles at 9 volts.

Weird... I'll have to look for that resistor you talked about.

It looks like this PSU is toast though if the PWM controller is dead. I don't think those black-coated PCBs are repairable, unless I can remove that black paint somehow?

[yuuki47]

Re: Old Compaq Armada internal PSU repair

If correct replace PWM controller. Should run then.

Yeah I'm guessing the PWM controller is inside those black-coated PCBs, in which case, I'd have no idea what chip to replace it with. If its the PWM controller, I'm guessing its impossible to revive that PSU.

I'll do those measurements today and tell ya, have to re-assemble the PSU though.

[Th3_uN1Qu3]

Re: Old Compaq Armada internal PSU repair

Leave the primary capacitor alone. Measure voltage on startup capacitor that you replaced. If (whatever voltage) but consistent, you have a dead PWM controller on the primary. Since that PS is so old it is likely a jellybean part, see what the numbers are on it. Replace it.

If (whatever voltage) but pulsing, it could be either the controller or the resistor feeding the startup cap. More commonly the resistor in this situation. See the numbers or color bands on the resistor, take 1 leg out of resistor (make sure to discharge that big capacitor again before you do that!), measure resistor - if higher value replace resistor. If correct replace PWM controller. Should run then.

[stj]

Re: Old Compaq Armada internal PSU repair

no - dont risk the tester - those caps can probably re-constitute some charge after you short them.
the black one may be 100uf btw, if it was 10uf it would be a lot smaller i'm sure

[yuuki47]

Re: Old Compaq Armada internal PSU repair

something thats been bugging me,
whats the large cap that is on it's side?

I think its the main voltage cap? Is that what its called? I always see these super big capacitors in old supplies like this one. Its at the primary side so I'm guessing it acts as a buffer.


Its another KMG rated for 400v 65uf. I'd say it works because every-time I discharge it, it makes a pretty big bang. Like a firecracker! But I also measured the voltage across it when the PSU is on, there are 350v there which explains the bang :P

Its not the only high voltage cap though, there is a black nichicon rated for 400v 10uf so I guess those sparks could come from that instead since they are on the same line. And by that I mean, I probed them with the multimeter and there is a direct path across both of their legs.

Should I take both out and put them on the M-Tester?

[stj]

Re: Old Compaq Armada internal PSU repair

something thats been bugging me,
whats the large cap that is on it's side?

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Good morning!



I went ahead and did more testing today. The disc cap was hard to pull out because it was behind the transistor, so I had to take this out as well. When I did, I discovered that this was actually the coil, and this was actually what had the short across it, not the ceramic cap the ceramic cap is fine.

Here are some markings to help you out:

Blue --- Transistor (by the output caps, which is the secondary side I think)
Red --- Coil windings connector 1
Light-blue --- Opposing coil windings connector 2
Brown --- Actual location of the disc cap
Green --- Black-coated PCB near transistor
Purple --- Startup Cap

Ignore the brown mess, its flux not a burn. The solder is really old so it needs tons of flux to actually melt. That's where the diode I pulled out to test was.

And now that this big heatsink and MOSFET are gone, here is a better view of the output caps.



You can see this white-looking plastic thing. That's part of the coil, it looks like the connector for the windings. So it makes sense that its shorted. But even if probe between the opposite windings, there is still a short! Is it supposed to be that way? It has two windings, gold and red, so I'm guessing not. But its probably not the coil itself causing the short, maybe another component.

That said, since I happened to take out the MOSFET, I tested it. It's good.



Mamoka --- it looks like the ceramic cap is connected in series with resistors like you said, and its right next to them too. You can see them (R42).

Strangely two of those resistors are connected to a big blob of solder. That big blob of solder connects the MOSFet's heatsink. That's weird... what's going on there? But anyway, there's no short at the disc cap after all.

Speaking of the coil, its connected in series to the black-coated PCB. But the black coated PCB is fine, no shorts and there are 1.5k ohms across the pins. I'd say its good. The white component continues to be shorted though...

About the diode directly connected to the transformer. I found it and no short!



The red markings are the connection between the cap and the diode and the green ones are between the diode and the transformer. I was able to read the model number and its a n-channel mosfet (NEC K2141)

It has 1.5k ohms between Gate and Source. I tried to discharge the mosfet to see if the drain closes properly but probing between source and drain the resistance slowly goes up... likely because of some capacitor in the line :/

There's no short though so is it reasonable to say that its okay? I really don't want to take it out, I have no de-soldering braid and the way I desolder is by stripping thick gauge wires and coating them with flux to suck the solder. Weird but works. Its a huge pain though.

It was also covered in glue underneath (you can see it in picture). Should I re-cover it?

[momaka]

Re: Old Compaq Armada internal PSU repair

Something is not quite right by that description. It sounds as if you're measuring a short-circuit on the output, but the fact that there was no short-circuit between the positive (+) and negative (-) marks of the output toroid choke suggests there is no short-circuit on the output. Therefor, I doubt the output caps would be shorted either (and you measured them already with the transistor tester, if I remember correctly.)

I did check the transistor besides the output caps with the multimeter. Don't those fail shorting the gate to one of the pins? Anyway, the short isn't coming from there because the Gate and Source pins have 10k ohms between them. No short at the Drain either so I'd say its good.

They can fail short-circuit from Gate to Drain or Source, but the most common failure mode for MOSFETs is short-circuit between Source and Drain. So I guess check for short-circuit between any of its 3 pins.

About the filter. You mean the one with the black ring? I'm not sure how that thing works but thankfully it has positive and negative markings on the board so I probed across those. No short at all.

Yup.
It's a common mode choke in this case, as it connects the negative (-) output wire of the PSU to the negative (-) side of the output caps and the positive (+) wire of the PSU output to the positive (+) side of the output caps.
Essentially, it reads like a piece of wire - i.e. short-circuit between the two "positive" ends and same between the two "negative" ends. It should not read a short-circuit between positive and negative, though, which you confirmed. This means the output of the PSU isn't shorted.

There is a very small disc-like capacitor by the big transistor at the heatsink... not sure you can see it from the pictures. I have a feeling that this is causing the short. It does have 0 ohms across it but other things that are on the same line do as well so that's not exactly a clue.

Barely, but yes.
Looks like a ceramic disc cap. Those can short-out too. But before you remove it, see what it's connected to. Often times these are connected in series with a low-value resistor (typically less than 10 Ohms), and this RC network is typically across the secondary-side output pins of the transformer. The transformer output will read like a short-circuit due to low number of turns, so checking across such ceramic disc caps usually shows the low resistance of the series resistor.

That being said, I can't make it out from the pictures how this ceramic cap is connected, so I can't say if that's what you're seeing or not.

I mean, is there any way to find the short other than to try pulling out components at random? If it wasn't a PSU, I'd feel for heat with my hands to see what gets hot, but it is a PSU so I ain't gonna do that!

Well, you wouldn't be able to feel for any hot components, because the PSU immediately shuts down when it detects a short-circuit on its output.
In terms of not removing components... well, you don't have to remove resistors, unless they test much higher resistance than the one printed on their sleeve (since resistors go high resistance or open-circuit when they go bad.) On the other hand, it's ceramic caps, diodes, and transistors that like to go short-circuit and may need to be removed out of the circuit to test properly.

There are other ways, but it's hard to tell from the two pictures of the straight top and bottom. It could be that all of the "shorted" parts you found are normal for the circuit. Or not. Sometimes when I get stuck on a PSU like this, I go back to checking things on the primary side. In the case of this adapter, if you haven't checked already, see if the primary side auxiliary winding rectifier (diode) is OK. It's usually directly connected to one of the pins on the transformer (or through a low-value resistor) and then feeds into that small "startup" cap you replaced.

But I guess let's finish with the secondary side first to make sure everything is OK. Indeed check the ceramic disc cap tomorrow and also the transistor on the heatsink for short-circuit between Drain and Source. Also, in regards to D7 - just check it in both directions between any two pins. If any two show a short-circuit, not them down. From the looks of the underside, it doesn't appear that any of the pins are connected together, so I don't think you should find any of the pins to read short-circuit between them. But do correct me if I'm wrong on that.

[stj]

Re: Old Compaq Armada internal PSU repair

maybe one of the electrolytics is shorted,
it's rare but i happens

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Yes, I checked the zener diodes on the secondary side --- all good!

There are two diodes besides the output caps (should have mentioned) a regular black one and a zenner one at the edge of the board. No short on the zenner, but a short at the black diode. It's not the diode causing the short tho, its something else.

I did check the transistor besides the output caps with the multimeter. Don't those fail shorting the gate to one of the pins? Anyway, the short isn't coming from there because the Gate and Source pins have 10k ohms between them. No short at the Drain either so I'd say its good.

The ceramic C16 cap has 19k ohms across it so certainly not it.

About the filter. You mean the one with the black ring? I'm not sure how that thing works but thankfully it has positive and negative markings on the board so I probed across those. No short at all.

There is a very small disc-like capacitor by the big transistor at the heatsink... not sure you can see it from the pictures. I have a feeling that this is causing the short. It does have 0 ohms across it but other things that are on the same line do as well so that's not exactly a clue.

I mean, is there any way to find the short other than to try pulling out components at random? If it wasn't a PSU, I'd feel for heat with my hands to see what gets hot, but it is a PSU so I ain't gonna do that!

I'll pull out that disc cap tomorrow...

[momaka]

Re: Old Compaq Armada internal PSU repair

That black output toroid by the output caps... check if the the windings from the two wires on it aren't shorting out. I believe that's a common-mode output choke (meant to suppress HF EMI/RFI.) I also see a bunch of Zener diodes (Z2, Z5, Z7, and etc.) on the secondary side. Check those too (just like you would a regular diode.)

Another one: ceramic cap C16 - check for short-circuit. I don't think that's connected to anything high power, but check in just in case. Feel free to check the other ceramic caps too, though I doubt they would be the problem.

Found a short at the diode besides the output caps

Is that the small red one?

If so, that might be a protection diode. If it's connected across (in parallel) with the caps, then the main rectifier might be bad.

Did you check the diode/transistor mounted on the heatsink by the output caps? That one should be the main rectifier. It's the one in the upper left corner of this picture you posted:
https://cdn.discordapp.com/attachmen...126_220859.jpg

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Found a short at the diode besides the output caps

But even after removing it to double check, the short is still there. And the diode itself forward biases fine once I removed it, so its not the diode causing the short... its something else.

There is a tiny SMD diode underneath the PCB (D7) where the output caps are. But I'm not sure how to test that.

[stj]

Re: Old Compaq Armada internal PSU repair

check all the diodes with a multimeter for shorts

[yuuki47]

Re: Old Compaq Armada internal PSU repair

Definitely not. It might be the camera making it look that way but the colour looks pretty consistent IRL. I don't see any burns, BTW There's two of them...