FSP-350GLN(80) not booting mobo properly

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

My Fortron is still running MSI motherboard well, so I have decided to delve deeper into Seasonic G360.

I have disassembled it to inspect PCB, it had untouched warranty sticker so technically I was first person to open it. I didn't really dig deep, on surface everything was fine but I still cleaned it with alcohol. All caps looked fine, nothing bent nor oxidized.

On the bottom side of PCB, in one place I found some small area where it looked like there was small leakage of something but I've cleaned it all areas around it and sucessfully tested it for continuity. Oddly, after cleaning area I found two-three mini scratches on PCB and there was this brownish-golden color exposed under PCB varnish. But this area had no traces so I left it untouched. It was on the opposite side of VEF20FB11, albeit not sure it if was exactly it or something around. Picture taken from internet, as I didn't make own:

Also LQP332001 was bit loose on top side(bottom part was soldered well, it looks like it's 2-piece component).

Solders on the opposite side of PWR_OK cable didn't look good, color was bit rusty but they passed continuity test. They still didn't look great after cleaning. Oddly, one of tracers near PWR_OK(but not directly linked to it) was also scratched(green varnish gone and brownish-golden color exposed), but everything on that trace passed continuity test. I have tried to check PWR_OK Voltage while touching exposed area with screwdriver but nothing changed, I've decided to leave it as it was and didn't even apply any coating.

Given that I was first person to open this PSU it seems that those scratch damages could come straight from the factory. It wasn't definitely me and sticker looked pretty good. Also to scratch it on that side of PCB they would have to cut one of zip ties holding power cable but it was untouched and original. I had to apply my own after mounting PSU back. Unless it was opened by Seasonic for warranty/refurbishment purposes.

Nevertheless, nothing has changed, PSU still doesn't want to boot unless there's something else connected(and it can't be just 0.1A LED, I think it must be at least 0.5A) drawing power, but after passing PWR_OK and booting everything is fine and stable.

And sadly, I have probably broken my multimeter so now it will take me even more time to toy with this issue.

[CapLeaker]

Re: FSP-350GLN(80) not booting mobo properly

you did pretty good work and it was quite the catch up read.
Here is one thing: Any brown goo you find, always remove it and clean it. After a while it does go conductive and causes problems.

Now this PSU on top looks to me like a few caps ran a bit hotter than others, since the sleeve have shrunk a little. Maybe worth checking it out.

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

Thanks, I paused this for a bit due to more important things. Also, accidentally I dropped my multimeter and it ran out of range and is showing unrealiable results(like scale would go down to lowest digits), so until I swap one or fix it there will be no progress. And the picture I posted was generic one I found on internet, it's not mine. If(when) I will find more time I will send some real pictures and more multimeter readings. But the bigger problem is, if this Seasonic PSU has some issues then I will still need to replace proper components and I'm afraid of using my soldering gun because I don't have much experience and gun itself is of poor quality.

As for this goo, I had only two things in mind:
- either it was leaked electrolyte(the affected area of PCB is directly under VEF2011FB transformer - can be seen on picture, but I doubt it's transformer itself as I don't think PSU would start without it; obviously there are some capacitors, resistors and other elements around it too but all did pass continuity test - last test I managed to do before I broke multimeter)
- flux leftover from factory solder... I heard such leftovers with time can get conductive and short circuits or do other damage

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

Forgot to add that at some point I've decided to test Seasonic PSU in PC again, just with 12V fan as bypass launcher. It worked and everything booted normally. I have successfully stress-tested PSU under Prime95+, Furmark, etc. load and all Voltages were very stable. Everything worked perfectly fine. But next day I have tried to boot PC using 12V fan as bypass and it didn't launch at all. I've noticed that PWR_OK is still 0.6V even with fan connected and running. I've also tried adding more load, ie. more fans, DVD-ROMs etc. but it didn't work. I will leave PSU disconnected for few days, as I want to see how does it behave when discharged.

So I also want to ask, for how long do disconnected PSU capacitors keep their charge in? And also, can I just use screwdriver to instantly discharge them by connecting both ends?

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

I couldn't resist, so I took my barely working multimeter and found out that 2000 Ohm continuity test works(every other continuity test is broken including buzzer one). And then I have decided to follow the PWR_OK trace and it led to smaller PCB board soldered to main PSU PCB board. Then on this smaller PCB board I found 8-pin HY-510N(538 J69A3G) element. It's PC Power Supply Supervisor, essentially responsible for PWR_OK checks. Datasheet is here: https://z3d9b7u8.stackpathcdn.com/pd...0N-HawYang.pdf

Because all I could do were continuity tests, I have decided to test each pin of HY-510N. If you look at the image on page 2 and 5 of datasheet, there are pin descriptions. They all worked with each other except for PDON_N pin. I do believe that PDON_N is a check that is responsible for checking PSU hold_up time(ie. how much power does PSU store in case of sudden power-loss).

On page 3 of spreadsheet, there's a diagram with PDON_N(page 5 also has some basic information) and I can see 'not gate' there. Hence I wonder, according to this diagram, should PDON_N be passing continuity test or not?

[CapLeaker]

Re: FSP-350GLN(80) not booting mobo properly

So I also want to ask, for how long do disconnected PSU capacitors keep their charge in? And also, can I just use screwdriver to instantly discharge them by connecting both ends?

That depends on if there is a bleeder resistor I stalled or not, the size of the capacitor, what kind of a capacitor etc.

Maybe you should get your DMM a situation figured out first and either borrow or invest into a new one. At this point I wouldn’t trust it.

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

I am still pretty busy with other things but for all that time my FSP-350GLN is working fine with motherboard. It's currently Seasonic G360 that is causing issues(exactly same as FSP before), so perhaps topic name should be changed to reflect it.

Nevertheless I got new multimeter but I don't have time to do more extensive tests yet. Moreover it's not only multimeter but general lack of proper tools that is holding me. I don't have proper lightning, magnifiers, accurate soldering guns, neither high-resolution camera to post pictures here. I can't even access certain parts of PSU because they are glued. Moreover I would have to desolder certain parts just to reach the others, and this includes certain capacitors. As I said, on this Seasonic, visually everything looks super clean and 100% like new. Literally sterile with zero dust and gunk, even smell is good.

But to test new multimeter I have just tried to launch this Seasonic again, without any load. And voila - it booted perfectly fine without any load(not even own fan). PWR_OK is 5.04V. Tested multiple times - everything works fine. Right now I have decided to keep it running on paper-clip with few 12V/5V accessories attached just to check if it's going to work after it's loaded for some time.

That is because I somehow suspect that perhaps there's something happening in one of capacitors(don't know which one and it's pure guess), maybe it's working fine when it's completely empty but once it accumulates charge, it blocks PSU from running(PWR_OK). And second guess, perhaps it's not even PSU fault but motherboard itself that messed both(Fortron and Seasonic) PSUs this way.

Speaking of thing above, that 8-pin HY-510N PSU Supervisor chip(it's on second smaller PCB attached to main PCB) that controls pin PWR_OK(on 24-pin motherboard connector), I haven't still been able to find out should pin PDON_N be connected to other pins on HY-510N but I found out that pin PDON_N is directly connected to PS_ON(green wire) pin on 24-pin mobo connector. I have tried opening other PSUs but I couldn't find any diagram for them and it seems that there's is nothing similar to HY-510N on them. FSP350-GLN does have some other 8-pin chip but it's of different function, hence I suspect that PSU Supervisor chip on that PSU is part of some other, bigger chip, as there are few of those. And pin PDON_N on that HY-510N PSU Supervisor chip should be somehow connected to one of capacitors(hold-up capacitor, don't know proper name) as it should be measuring needed energy reserve inside that capacitor. But I am yet to find out which exactly capacitor it is on my PCB.

Now back to blacklisted Modecom PSU that always worked with my motherboard, I do believe that this one doesn't even have any PSU Supervisor chip. To remind how PSU Supervisor works, it has to give (around)5V on the PGO pin(output), if it does then PSU should pass PWR_OK test and boot PC correctly, if it doesn't then parts of mobo including fans should work but PC shouldn't turn CPU on. So this Modecom is probably one of those cheap PSU's that is bypassing the need for +5V from PSU Supervisor with normal +5VDC(red) rail so motherboard "thinks" that everything is fine all the time and always boots. This could explain why my mobo always works with this PSU while other PSU's are sometimes failing.

If I notice Seasonic PSU failing next time I will be checking PGO pin on HY-510N. During test runs I have noticed something odd, while Voltage on PGO pin should be same as on PWR_OK pin, and it usually is(5.04V), it sometimes falls down to 4.98V or even 3.x for less than second. But PWR_OK remains stable at 5.04V, which could mean that there's something else to stabilize Voltage between PWR_OK and PGO(and HY-510N chip).

[tartuf]

Re: FSP-350GLN(80) not booting mobo properly

I still didn't do any comprehensive test, but as it turned out, perhaps I might not need to. This is fifth day I'm using Seasonic G360 PSU on this PC. Not a single problem. Fortron 350GLN is also working fine. PWR_OK on both PSU's is at around 5V, stable. Odd, as both PSUs seemed to be malfunctioning? Perhaps not.

My conclusion is, that there was something shorting PWR_OK pin on motherboard. I've tried cleaning case and motherboard multiple times and analyzed PCB traces but found no oddities. Maybe it was some sort of gunk or wet animal hair or something just electrified - or maybe it was more serious malfunction, and it's still there. But for now, it seems it stopped.

By sending back-signal to PWR_OK which itself is supposed to give 5V signal, it overloaded one of the PSU capacitors on PWR_OK trace. My first quess is, it's capacitor that is between PWR_OK pin and PGO pin(on PSU Supervisor, as on image above) or maybe it's whole PSU Supervisor mechanism(I've read somewhere that some of those PSU Supervisor chips also do have their own micro capacitors built-in - I have no way of verifying this info).

Anyway, as capacitors discharge with time, things returned to normal. I'm quite happy that they didn't blow which could indicate their quality. Of course, now I should test them extensively but I can't be bothered to, unless PSU "dies" again, those capacitors are hard to access and would require me to either remove glue or desolder some wires and PCB parts.

Oddly, same time I hoped that it would die again so I could verify my theory by checking Voltage output on PGO when PWR_OK is less than 5V(they should be almost same) and then checking the trace between PGO and PWR_OK for connected elements and of course checking capacitor itself. And if it would be capacitor then I would just manually discharge it with multimeter or screwdriver to "resurrect" PSU.


So to summarize this whole topic, discharge your capacitors before assuming that they are dead or malfunction.