1250W XFX PFC power supply refuses to start up. No clue!

[xtv]

Straight to the point:

- Was working before. Refused to start back after a software shutdown.
- Strange oil on the back of the pcb. No appearently leaking cap around this area. [Heard the thermal pads on the case contains oil in its formula and can release it over high temps]
- No smoke, no smell or visually toasted component
- Diode and big FETs in the left side of the pcb seems ok Are OK.
- Standby 5VCC is working
- Right after switching the main power on, the big caps voltage is 170v.
- After attempting to switch the psu on (green-black jumper), the big caps voltage peaks to 400v, but start discharging (slowly), until it reaches back to 170v.
- The relay never switches on.
- No output on 12V, 5V or 3.3v supply
- I've also measued the VCC / GND of the "protection" IC PS232F in the right side of the PSU, and it gets 5v. Probably most control components are being powered.
- PWR_OK (or Power Ready) pin in the ATX connector have near to zero vcc
- PS_ON (Green wire) Atx pin have +3.2v
- No noise that I can notice. (usually, switching PS does a small high pitch noise)
- NiHaoMike suggested his ninja technique to turn a AM/SW radio next to the PSU and attempt to switch it on. We were able to hear it oscillating when the mains switch is turned on. Also a peak noise when attempting to power it on
Record is in here: http://oztix.sdf.org/xfxpwr/PSU_startup_SWRadio_5700KHz.mp3
- peter.mitchell suggested to forcibly power the 12v rails of this PSU using an external regulated PSU using a low value resistor in series to check a possible short and if the DC-DC conversion circuit works. I've got this PSU to work doing the following: (1) Powering it. (2) Power the 12v rail using the external PSU. (3) Ground PS_ON. The relay closes here. (4) Remove Aux. PSU. All outputs working: 12v; 5v; 3.3v.
--------------------------------------
Hi everyone. I'm from Brazil and I have spent the money I had on a litecoin mining rig about 1 and half months ago. Turns out that, after this time running, the XFX 1250W Pro Black Edition PFC power supply is not starting anymore. Not even hard gounding the green wire of the ATX connector. As computer equipment is around 40% cheaper in Paraguay, I've bougth everything there. So rma would be a problem, but I didn't expect having problems so early!

This happened after a soft windows shutdown. When I tryed to start the machine again, nothing happened. No smoke, no noise. So I decided to open the case, in hope to find a leaking capacitor, toasted transistor or diode, check the fuse, and so.

I found nothing visually strange, except of some components in the back of the board soaking on oil. These components are pressed by a thermal conductive rubber that is fixed in the botton plate of the case.
I'm wondering if this is commonly used to help thermal conduction from the components to the backplate (this area gets real hot while working)

I have checked every electrolytic capacitor over this region to make sure it wasn't a leaking cap. All of them not only looks good, but also show correct values in a capacimeter.

This looks an advanced PFC power supply, and I have no good idea on what and where to start investigating.

Any help on diagnosing this will be much appreciated!













The leakage in the back plate is barely seen in eye, but the camera seems to capture it quite well..
The rubber is already clean in the picture, but was a pool of oil when I opened it.

[mariush]

This is a Seasonic OEM design. Quite a high quality one.

The main transformer looks like it was overheated (i see something that looks like melted plastic). I would check the main transformer to see if it didn't have its fuse popped due to heat.

The relay is (if i remember correctly) there to disable the 5v standby section and switch it to the main dc-dc 5v converter once the power supply runs, to increase the efficiency (the separate 5v standby is less efficient than generating 12v and then getting 5v using dc-dc converter). Or it could be just disconnecting the main power supply and leaving just the 5v standby when working, again for efficiency numbers (to have less losses when turned off)

[xtv]

Hey Mariush, thanks for having a look.
If you mean the white thing over the transformer coil, looking like melted plastic, it's a volume glue used to fixate components. It's used all around the pcb.
If you mean by the resine drop over it, yeah I may agree with you. Although I belive it would have some smell if it was overheated that way, and it hasn't, at all.
I don't know these transformers used in ATX PSUs very well, but the coils are not open (my hand multimeter says zero ohm between terminals, not sure how low that should be), but primary and secondary are isolated. Are these fuses internal?

BTW, I've misinformed something above. PWR_OK are not +5VCC. It's near zero under mV scale. The one with 5VCC is the purple wire, that is the standby voltage (VSB). Correcting it...

[SeanB]

No standby 5V means U1 is dead, and it looks like it has popped it's top as well. When replacing it change the small capacitor next to it as well, as that could be the cause along with a dry joint on the transformer, or a period of high mains voltage. It could have happened a while ago but you will only see it on power off after the main power rails go away.

[peter.mitchell]

I haven't gone through anything in your post or the posts above, however, i do congratulate you for providing one of the most informative and best picture taken posts on this forum relating to a problem in the last 2 years. Thank you, this post should be an example to others.

[xtv]

No standby 5V means U1 is dead, and it looks like it has popped it's top as well. When replacing it change the small capacitor next to it as well, as that could be the cause along with a dry joint on the transformer, or a period of high mains voltage. It could have happened a while ago but you will only see it on power off after the main power rails go away.

Hey Sean, thanks for having a look.
Actually, Standby is 5VCC. What is not (and should be) is Power_OK (or Power Ready), which tells the motherboard the PSU is ready to power up and everything is ok.
The little mark over the U1 that made you think it's popped is a ink pen. I've visually inspected it and it seems ok.

Also, PS_ON (the pin that switches the PSU on when "lowed") is at 3.2v, what makes me think the circuit that controls the PSU startup may be working, but something is wrong to start delivering power..

[xtv]

I haven't gone through anything in your post or the posts above, however, i do congratulate you for providing one of the most informative and best picture taken posts on this forum relating to a problem in the last 2 years. Thank you, this post should be an example to others.

Thanks a lot, Peter.
I belive if you're taking somebody else's time, you should at least give the best to help them help you.

Also this PSU costs 1k bucks in my country.. So my options are to fix it, or fix it. Or keep trying... 

[NiHaoMike]

Set an AM radio next to the PSU, turn it on and tune it to a blank station, then do the green to black jumper. Let us know what you hear since that can give a hint if it the PSU is attempting to start up at all.

[Monkeh]

What is not (and should be) is Power_OK (or Power Ready), which tells the motherboard the PSU is ready to power up and everything is ok.

No, it tells the motherboard when the PSU is turned on properly, not that it's ready to. It should remain low when the PSU is on standby.

[Monkeh]

The relay turns on the main supply. Pull the relay and test it, then start investigating the circuit controlling it.

[Monkeh]

I'm pretty sure the relay shorts out the vertically mounted heatshrunk resistor, inrush limiting.

I think you'll find that's a fuse. Notice the gigantic inductor for inrush limiting.

[NiHaoMike]

The PTC that gets bypassed is that heatshrinked green bit next to the relay.

[Rerouter]

The capacitor next to U1 has leaked, and flowed over the edge, that is the residue you see in the image of the casing, the increase in ESR and drop in capacitance probably then caused U1 to oscillate and pop, replace it and the chip, though you may have to find the datasheet on the cap as these things can get upset by changes in capacitance and ESR

if you can find more info on the cap, replacing it with a higher voltage model with similar ESR and the same capacitance would be the best bet,

[Monkeh]

Does the relay close?

[xtv]

Does the relay close?

The relay never closes.
I've monitored the standby 5v while attempting to switch the psu on, to check if the voltage drops. Not even a little.

[Monkeh]

Does the relay close?

The relay never closes.
I've monitored the standby 5v while attempting to switch the psu on, to check if the voltage drops. Not even a little.

Now, check if the relay works and/or if it attempts to energise it.

[xtv]

The capacitor next to U1 has leaked, and flowed over the edge, that is the residue you see in the image of the casing, the increase in ESR and drop in capacitance probably then caused U1 to oscillate and pop, replace it and the chip, though you may have to find the datasheet on the cap as these things can get upset by changes in capacitance and ESR

if you can find more info on the cap, replacing it with a higher voltage model with similar ESR and the same capacitance would be the best bet,

What an eagle's eyes, uh?
Actually, that cap leak you see going to the edge (with dust over it), is a minor part I haven't cleaned.
The big region with a leak was this one:
http://oztix.sdf.org/xfxpwr/region_oil.jpg
My main suspicion was the bigger cap in the center of the smudge.
As I reported, the MOSFETS on the back of the board was submerged in some sort of oil. I've asked here if that was used to help dissipate heat.
IF that oil is meant to be there, a small soldering fail of those heat sinks left a hole open in the pcb, allowing the oil to submerge to the surface of the board.
I've searched internal pictures of this PSU in the Internet, and I haven't seen this sort of oil in any of those. Reviewers never commented anything about oil in the back of the board.
But again, I have removed those electrolytic capacitors in this region, none have any signal of leaking, neither shows abnormal capacitance in a LRC meter. (Wish I had a scope to check the charge/discharge curve)

I tryed to measure the resistance of this oil (maybe it's stupid) but, it seems unconductive. At 2 Megaohms, less than half millimeter one probe from another, no signal of any resistance. I don't know if the liquid inside an electrolytic capacitor is conductive, but I wanted to check if that could have caused damage to the mosfets..
These fets in the back of the board are wired in parallel (S-D). I haven't removed them to test individually, but none of them are shorted.

Now that "solid state caps", can't have leaked?

[Monkeh]

No, the oil won't have been anything to do with the FETs, and no, solid caps can't leak (the clue is in the name). It's come out of one of the wet electrolytics somewhere, unless it's some byproduct of a defective silpad.

[xtv]

Does the relay close?

The relay never closes.
I've monitored the standby 5v while attempting to switch the psu on, to check if the voltage drops. Not even a little.

Now, check if the relay works and/or if it attempts to energise it.

Oh, I guess we have some progress.
The relay (component) works. 9v dc arms it gracefully.
On a switching on attempt, it's voltage peaks from 0v to 5v, and comes back to 0v. (I was able to hear a little "click" inside this time. It's trying to arm, but it wont.
I'll take a look at the relay driver transistor, but if it was bad, there was no reason to fall back to zero, right?

[xtv]

The capacitor next to U1 has leaked, and flowed over the edge, that is the residue you see in the image of the casing, the increase in ESR and drop in capacitance probably then caused U1 to oscillate and pop, replace it and the chip, though you may have to find the datasheet on the cap as these things can get upset by changes in capacitance and ESR

if you can find more info on the cap, replacing it with a higher voltage model with similar ESR and the same capacitance would be the best bet,

I've removed and tested the cap you mentioned.
unfortunately, it looks ok..
http://oztix.sdf.org/xfxpwr/47ucap.jpg
http://oztix.sdf.org/xfxpwr/47ucap2.jpg
http://oztix.sdf.org/xfxpwr/47ucap3.jpg

[Psi]

I've not read this thread, so if this has already been said sorry, but might be a good idea to locate the bootstrap resistor and check that.
It's usually like 100K or above and will be throughhole. If it looks to have got hot at all, remove and check it.

It wont start up with that resistor failed open. It's required to bootstrap the flyback and it's a common failure mode due to being underspec'ed most of the time.
(although i wouldn't expect that in a good quality psu)

[xtv]

I'm pretty sure the relay shorts out the vertically mounted heatshrunk resistor, inrush limiting.

Have you checked the bias supply? These bigger supplies usually have a dedicated supply for powering all primary side circuits. It may be an isolated flyback so it can power secondary side control and/or supervisory circuits as well. It will be a small transformer particularly when compared to the other magnetics. It looks like a quality supply. Did you try to talk to the manufacturer? They may a least give you an idea of what and where to look.

Bummer dumping a grand on a PSU and it craps out on you though. Hope you get it going.

Thanks, AcHmed.
You're right about the relay, it just shorts the thermistor. It's driver is feeded by a zener; when the voltage at C209 (the brown cap near U1) rises, the relay bypasses the PTC.

About the bias supply, aint it the same supply as 5V standby? I'm getting +5Vcc in ATX 5V standby.
I belive most controlling circuit is up, because I can hear the oscilation noise in a shortwave radio near the PSU. A peak noise while attempting to power it on as well. (audio file's in the main thread)

I've mailed Seasonic tech support sharing this thread. Let's see if they can be helpful in some way...

[crmaris]

The relay is only to bypass the NTC thermistor and allow it to cool it down for the next restart (and save some energy lost on it).

Firstly you have to be extra carefull while you apply power to this unit since the current on the first side is dangerous! That said do the following.

Connect it to power and switch on the on/off switch. Then short the green and black and check if the relay activates (it will make a sound). If it does then measure the voltage on the large APFC caps. It should be 380 VDC (DC bus voltage of the APFC converter). If it isn't so high then you have a problem although some PFC controllers can drop it down to 320 VDC when there is no load applied, to increase efficiency.

If the DC bus voltage is ok then the problem is either in the secondary side or to the main switchers/transformers so things get more complicated and I cannot do a diagnosis from here.

From what I know some batches of this unit showed similar problems but with a specific procedure you were able to start them again (this had to do with the on/off switch but I don't remember how). A user had reached me again with a similar problem but what ever he did he couldn't make the PSU operational again so I think you should try, before you open it, your luck with XFX's RMA (which isn't from the good ones as I hear, at least here in the EU).

However Seasonic's RMA is top so better next time buy directly the Seasonic unit (X-1250), although it might be a little more expensive.

[xtv]

The relay is only to bypass the NTC thermistor and allow it to cool it down for the next restart (and save some energy lost on it).

Firstly you have to be extra carefull while you apply power to this unit since the current on the first side is dangerous! That said do the following.

Connect it to power and switch on the on/off switch. Then short the green and black and check if the relay activates (it will make a sound). If it does then measure the voltage on the large APFC caps. It should be 380 VDC (DC bus voltage of the APFC converter). If it isn't so high then you have a problem although some PFC controllers can drop it down to 320 VDC when there is no load applied, to increase efficiency.

If the DC bus voltage is ok then the problem is either in the secondary side or to the main switchers/transformers so things get more complicated and I cannot do a diagnosis from here.

Hello crmaris, thanks a lot for having a look.

Here are the results:
- Right after switching the main power on, the big caps voltage is 170v.
- After attempting to switch the psu on (green-black jumper), the big caps voltage peaks to 400v, but start discharging (slowly), until it reaches back to 170v.
The relay never switches on.
I've also measued the VCC / GND of the protection" IC PS232F in the right side of the PSU, and it gets 5v.

From what I know some batches of this unit showed similar problems but with a specific procedure you were able to start them again (this had to do with the on/off switch but I don't remember how). A user had reached me again with a similar problem but what ever he did he couldn't make the PSU operational again so I think you should try, before you open it, your luck with XFX's RMA (which isn't from the good ones as I hear, at least here in the EU).

However Seasonic's RMA is top so better next time buy directly the Seasonic unit (X-1250), although it might be a little more expensive.

That's true, this psu since the beginnig had some problems powering on.. I had to switch it off and on, wait at least about 30 seconds and push the power button (case) for it to turn on. Now you made me wonder if this could be a capacitor that was operating right in the limit and now that it got worse, wont start at all...

[xtv]

I've not read this thread, so if this has already been said sorry, but might be a good idea to locate the bootstrap resistor and check that.
It's usually like 100K or above and will be throughhole. If it looks to have got hot at all, remove and check it.

It wont start up with that resistor failed open. It's required to bootstrap the flyback and it's a common failure mode due to being underspec'ed most of the time.
(although i wouldn't expect that in a good quality psu)

Hi Psi, thanks for the suggestion.
I've checked every resistor in the pcb (without desoldering them), except in the PFC / PWM small pcbs.
Do you have an idea how can I locate this resistor?