FSP-350GLN(80) not booting mobo properly

[tartuf]

Greetings

My old 350W FSP-350GLN(80) PSU was working fine with my current computer for few years(and previously with few other computers as well). Nothing has changed in setup but since last few months I am getting problems while trying to boot PC.

What happens, is that PSU starts and I can hear fans, both PSU, CPU, MOBO and GPU. I can also hear HDD scratching and normal starting sounds. But DVI monitor claims that there is "no signal". Moreover, neither USB mouse nor PS2 keyboard does flash leds like it should do upon boot.

I have to turn PSU switch on and off multiple times hoping that it will start at some point. Sometimes it takes just few clicks, sometimes I have to wait for few minutes before it can start. Once things go off, everything is perfect, computer is very stable under highest loads and fans are quite quiet. Restarting using OS restart option or restart button never cause any issues and works instantly. But (cold)powering down makes the issue come back.

It's quite old PSU, probably at least 15 years old. But it never caused any issues and it was quite quality product by then. I didn't use it all the time and it was never stressed. I don't have multimeter but according to software, voltages are pretty perfect.

PSU internals look fine, fan spinning normally, almost no dust, capacitors not bent, no oxidization, no smell.

I own a second FSP-350GLN(80) that was never used, and was stored in mint conditions and it's unable to power my mobo, there's just no reaction at all(while passing paper-clip test and working fine with +5V/+12V from molex). Both were purchased same time, and as both are not working properly, I think that it's just age-related problem.

But essentially, what could be causing such behavior? I know that +5VSB is used for USB/PS2 and I know that computer somehow behaves like it would be in ErP state and wouldn't be able to fully recover(even when I never used ErP). Oddly, I do own some blacklisted-noname PSU's that are even older than my Fortrons and they have no problems with powering my mobo.



added:
Note, I probably don't plan on fixing my Fortrons so question is out of pure curiosity. I can get used Seasonic G360 for around $15, would that be good replacement, what are opinions on those Seasonics(I've read here that G360 is a bit different than other Seasonics G)? I can even pick from different Seasonic G360 revisions, I guess they are alphabetical: BGP360FSJBW, BGP360FSJCW, BGP360FSJDW. Sadly, probably newest D-version is also in worst visual condition.

[sam_sam_sam]

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

You probably need to recap the switching power especially the capacitors under 100uf but it is over 5 years old and are brand name capacitors I would recommend replacing all of them

[CapLeaker]

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

Heat up the power supply with wifeys hairdryer on medium for 30 seconds and start the computer. If that works, it’s time for a full recap after 15 years of use.

[tartuf]

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

After swapping PSU I have noticed that issue is still there, indicating that it's not *only* PSU. With old PSU I had to use power switch like 3-10 times before PC properly boots, with new PSU I have to use it 1-2 times.

Anyway I wonder what else could cause such issue. Dying MOBO(MSI B250M Pro-VDH, like 3 year old), GPU(Zotac GTX 1060 3GB, 3-4 year old)?

[CapLeaker]

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

flash the bios to the latest version and clear the CMOS. See if that helps any. I know my bro has an AMD with an Asus mobo. He only can run certain bios versions as otherwise he doesn't have any sound. Also check that coin cell battery. If it is sitting at 3V flat, its almost dead. Also try the video card in a different PCI slot. Good luck!

[tartuf]

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

I got latest BIOS(UEFI) and for few years it worked fine. I will flash it next time it happens but for now more oddities came. As for battery, it's brand new - purchased like week ago. Old battery is also fine, just tested.

As I said, I've been using this mobo with FSP-350GLN(80) PSU without issues for like 2 years. Then out of nowhere(no hardware/software changes) "no boot" issue started to happen. Everything boots and I can hear fans and disks running except for PS2/USB/display.

To test FSP PSU, I took old Modecom 350W FEEL-III PSU(that is blacklisted) just to see does it boot my consistently without issues. It did, so my guess was that it's PSU problem. But given that resoldering old FSP PSU takes a lot of time and effort, not to mention that my soldering gun is almost dead, and I suspected that FSP PSU will die to age, I've decided to buy Seasonic G360W(5yr old, $16 with shipping). After swapping PSUs, upon first launch it was perfect. I did some tests and reboots and it worked fine. But yesterday "no-boot" problem happened again. So I flipped 0/1 switch few times and it started normally. Next, today my PC wouldn't boot at all. So I have unplugged Seasonic G360W and plugged Modecom 350W FEEL-III again. And it works, I'm using it right now. It's blacklisted PSU, and older than my Fortrons.

Anyway what I've tried was to unplug everything from MOBO, every cable one by one, GPU and RAM sticks, fans while having Seasonic G360W connected - and nothing worked. Then I used Modecom PSU and it easily booted my PC with just 24-pin. I've tried FSP and Seasonic PSUs on different mobo and they work.

[tartuf]

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

Cleared CMOS, no change. Disconnected discrete GPU and switched to internal but it didn't change anything. Did few more tests and got some findings:

I found some old Dell PC that had original BIOS(not UEFI). Noticed that when it boots, it first spins all fans, then beeps("all fine") and then powers up USB in that order. Decided to test FPS PSU's against Seasonic. Old 200W noname PSU that came with that old PC worked fine: fans->beep->USB power; both FSP PSU's worked fine: fans->beep->USB power; Seasonic didn't work: fans->no beep->no USB power(endless fans). I don't know, maybe because it's 2006 PC(Intel LGA775, don't know know which one) and Seasonic doesn't have native +3.3V/+5V rails(buck converter) or maybe it's malfunctioning?

Anyway, my MSI motherboard is UEFI and never gave any beep(can't even find any setting related to beeping in UEFI, MSI probably cheaped out). And neither FSP nor Seasonic works correctly with it(fans and disks spin up but PS2/USB/display is unpowered).

[stj]

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

your msi probably has no beeper,
a lot of motherboard companies dont bother soldering one into the board

[tartuf]

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

There's another thing, when testing PSUs with that old Dell PC I've decided to test my second FSP-350GLN(80), one that never worked with MSI board and only passed paperclip test. To my surprise it works with Dell(Foxconn G33M02 mobo). But that's another hint at MSI mobo malfunctioning and not PSUs...

[CapLeaker]

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

did you try to swap the PCI express slot for the video card? You get into no POST hell too if the MOBO doesnt see an video output. I am guessing your AMD hasn't got onboard video. It can be as simple as not all power rails going to the video card causing an intermittent (broken solder joint(s) on the PCI express slot). With them video cards getting larger and larger, it seems the slots get flimsier by the minute.

[tartuf]

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

Actually my CPU is Intel Pentium G4600 with HD630 iGPU. I have unplugged my dGPU, Nvidia GTX 1060 3GB completely and connected iGPU but result was same as with dGPU. Tested on every PSU.

Albeit I will do test this more extensively when I get more time.

[CapLeaker]

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

If you do a CMOS clear, does the computer boot afterwards o.k.? It may go through memory training first.

[tartuf]

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

Nope, no booting. It's all same(actually it boots except for PS2/USB/monitor).

[CapLeaker]

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

Hmm. If the mono doesn’t have a piezo speaker on it, I’d put one on a plug and plug it in. See what beep codes you get. Take the battery out, so the computer has to load the default each time you turn it on. Remove both ram sticks and note how many beeps, then just work with one ram stick, or better a known good ram stick and see what happens as you go through the slots. Take the cpu of and check the pins.

[tartuf]

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

I did more extensive tests.

First of all, MSI B250M Pro-VD, the one I'm trying to get to work, does have some beep option in BIOS but it doesn't work. I've connected old speaker to mobo audio output and there is no effect. But it also does have "MSI EZ Debug" which are four LEDs(CPU, DRAM, VGA, BOOT). I kinda think it totally sucks and is horrible btw. How it works? Well first it lights CPU LED and when everthing is okay, it lights DRAM LED, when everything is okay it lights VGA LED and finally BOOT LED. Then when everything is fine it turns LEDs off.

Second, I have managed to pinpoint my USB/PS2/Display problem to 4/8-pin ATX slot/connector. Right now I am using it as my reference point.

Third, as for tests - I used four PSUs and two motherboards. Let me recap the problem first once again:

I've been using old FSP-350GLN(80) PSU to power my system that is total around 200W in fullest load. That PSU is at least 15 years old and remembers Barton XP +5V days, but it never failed me before. Note, it has three/four rails(+3.3V/+5V/+12V1/+12V2) and single CPU ATX 4-pin connector. My motherboard has 8-pin CPU connector but I could just use 4-pin and it would work fine. Alternatively I've also rewired one molex extender just to connect to CPU ATX 4-pin to ensure that it's not PSU cable failing, and in this config I could run 4-pin+rewired molex for 8-pin or just any of these in 4-pin slot. I've been using this setup for few years without issues, until now. It seems that after cold boot, when pressing power button my PC won't deliver power to PS2/USB/Display. Here is what I wrote before, PC acts like it wouldn't be receiving anything on CPU 4/8-pin connector while motherboard power is working fine. This is also indicated by "EZ Debug"(CPU LED is on and doesn't want to switch). I have tested with just motherboard, PSU, single RAM stick, CPU and USB mouse connected, the goal was to either pass "EZ Debug" CPU test or make USB mouse flash which would indicate it receives power. Now to results:


1. FSP-350GLN(80) that was working all the time for last 15 years:
MSI motherboard(one I'm trying to fix):
24-pin connected = CPU LED on, fans spinning
4-pin connected = LED tests passed, mouse flashing
molex to 4-pin = LED tests passed, mouse flashing
4-pin connected and molex to other 4-pin(8-pin total) = LED tests passed, mouse flashing

Please note, it worked flawlessly today, I am using it right now. Yesterday it didn't deliver anything to 4/8-pin. It's really random with this old PSU.

Foxconn motherboard, has only single 4-pin CPU ATX(old one just for testing):
24-pin connected = fans spinning
4-pin connected = LED tests passed, mouse flashing
molex to 4-pin = LED tests passed, mouse flashing

Additional note, yesterday it didn't manage to boot MSI motherboard properly but it booted Foxconn very well. Today it booted both.

2. FSP-350GLN(80) that was idle all the time for last 15 years:
MSI motherboard(one I'm trying to fix):
24-pin connected = nothing
4-pin connected = nothing
molex to 4-pin = nothing
4-pin connected and molex to other 4-pin(8-pin total) = nothing

Now something very weird because I did test and it seemed that this PSU won't ever work with MSI motherboard. At some point I accidentally dropped it from table, then after reconnecting and just for once it worked with 24-pin(CPU LED on, fans spinning), didn't work with 4-pin. Sadly 15min later I've tried to retry test and it was dead again.

Foxconn motherboard, has only single 4-pin CPU ATX(old one just for testing):
24-pin connected = fans spinning
4-pin connected = LED tests passed, mouse flashing
molex to 4-pin = LED tests passed, mouse flashing

3. Seasonic G360W PSU that have one 12V rail and buck converter to +3.3/+5V. Five year old PSU, warranty expired three months ago, bought used from reliable source(was working with low power CPU and integrated GPU):
MSI motherboard(one I'm trying to fix):
24-pin connected = CPU LED on, fans spinning
4-pin connected = nothing
molex to 4-pin = nothing
4-pin connected and molex to other 4-pin(8-pin total) = nothing
8-pin connected(has native 2x4-pin connector) = nothing

This is standard behavior when the problem is on and my Fortron PSU is also revealing such behavior. Please note that few days ago Seasonic PSU did manage to start and work properly without any issues. Next day I had to restart it once or twice before it managed to boot properly. But on third day and few days after it didn't power 4/8-pin.

Foxconn motherboard, has only single 4-pin CPU ATX(old one just for testing):
24-pin connected = fans spinning
4-pin connected = nothing
molex to 4-pin = nothing

Now this is even weirder, why wouldn't it be able to deliver power to Foxconn motherboard?

4. Modecom 350W FEEL-III(blacklisted), older and more used than my Fortrons:
MSI motherboard(one I'm trying to fix):
24-pin connected = CPU LED on, fans spinning
4-pin connected = LED tests passed, mouse flashing
molex to 4-pin = LED tests passed, mouse flashing
4-pin connected and molex to other 4-pin(8-pin total) = LED tests passed, mouse flashing

Foxconn motherboard, has only single 4-pin CPU ATX(old one just for testing):
24-pin connected = fans spinning
4-pin connected = LED tests passed, mouse flashing
molex to 4-pin = LED tests passed, mouse flashing

This PSU is the only one that so far managed to launch my MSI motherboard every time. But given that I almost never use it, I won't be surprised if at some point it will stop delivering power to ATX CPU 4-pin like my other PSUs did.

I did two series of tests, first with CMOS batter inserted and second without CMOS battery, no difference. I have to add, I have tried different molex rails for each PSU(to power ATX CPU 4-pin), and I did use both first 4-pin and second 4-pin slot and it all worked when it worked. In case of Seasonic I did both molex and 4/8-pin in pretty much every possible configuration and it never passed "EZ Boot" CPU LED test. Additionally, I have did extra test, where I have heated all PSUs with hairdryer before launching, but results were exactly same. Also cleaned and reconnected everything and applied new thermal paste. I've even poured some alcohol on motherboard pins as a lousy attempt at cleaning them as I am unable to rule out oxidization of ATX CPU 4/8-pin slot. Finally, tried different PSU cables and different wall plugs.


And now the fun part: PSU or Motherboard? Both or something else?

[CapLeaker]

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

I would start with borrowing a known good more recent PSU first. This swapping MOBO’s of different vintages using more old PSU’s isn’t very good, as one board could work fine and the other doesn’t at all.

[tartuf]

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

I might get access to more computer parts but it will take some time.

But even if new PSU would work upon first launch, what are the chances it will work all the time? As this issue is pretty much random and I am sometimes able to boot my PC properly using any of my PSUs(except for one FPS that doesn't work with MSI) just to find out that after few days they can't pass CPU LED test... and then few days after this such failure they work again.

[tartuf]

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

I found some multimeter(had to fix it first) and made some testing with it. I do believe that results are very promising. I've tested four PSUs and MSI motherboard(Ohm test). Didn't do short-circuit test and didn't test capacitors but can do it later.

Here are the results, first motherboard 24-pin and 4/8-pin Ohm tests. Note I had to use 2000 Ohm setting instead of 200 Ohm setting as my multimeter is still not working 100% properly:

24-pin atx:
[0400,0400,0000,0790,0000,0790,0000,0001,0840,1520,1520,0400]
[0400,0001,0000,0001,0000,0000,0000,0001,0790,0790,0790,0000]
.............................................[CLAMP]...........................................

8-pin atx:
[0000,0000,0000,0000]
[0001,0001,0001,0001](when plugged those four go from like 500 to over 1500 but then settle at 1)
............[CLAMP]..........

I don't know how to read these results, especially those with just "1". Pay attention to 5th-pin counted from bottom right(PWR_OK).


And now PSU tests, all PSUs launched with paperclip and tested extensively:

Seasonic G360(load 12V molex fan, ~0.5A // no load):
+3.30V: 3.36V // 3.36V
+5.00V: 5.04V // 5.04V
+12.0V: 12.24V // 12.23V
+05VSB: 5.10V // 5.10V
-5.00V:
-12.0V: -11.49V // -11.46V
PWR_OK: 5.04V // 0.06V (if I unplug fan in middle of load and leave it, it will stay at 5.04V)

Again, pay attention to PWR_OK. I don't know if it's normal but it certainly doesn't look normal to me. This PSU was purchased few days ago, worked fine with MSI motherboard for few days and then it stopped - no power to USB/PS2/display. The culprit? Of course PWR_OK. But I have managed to go around this issue, I have connnected big fan to 12V molex and when the fan is powered the Voltage in PWR_OK is in normal range! And this boots PC normally, while without fan connected there is no USB/PS2/display.


Fortron that was never used #2(load 12V molex fan, ~0.5A // no load):
+3.30V: 3.66V // 3.56V
+5.00V: 5.21V // 5.18V
+12.0V: 12.21V // 12.30V
+05VSB: 5.05V // 5.05V
-5.00V:
-12.0V: -12.31V // -12.33V
PWR_OK: 5.21V // 5.17V

This is obviously out-of-range PSU, it didn't even manage to power MSI motherboard at all so I am guessing that these Voltage ranges are just not tolerated by MSI motherboard. As I said before, it runs old Foxconn motherboard just fine.


Fortron that was used all the time #1(load 12V molex fan, ~0.5A // no load):
+3.30V: 3.56V // 3.44V
+5.00V: 5.18V // 5.15V
+12.0V: 12.15V // 12.23V
+05VSB: 5.10V // 5.10V
-5.00V:
-12.0V: -12.27V // -12.28V
PWR_OK: 5.17V // 5.14V

Surprisingly good results for something that was used non-stop for over 15 years. Seems that high +3.3V is normal for these Fortrons. This was the PSU that initially was running MSI motherboard until PS2/USB/display died. Oddly, since few last days I am running this PSU again and it's working with MSI motherboard perfectly. But when it didn't, I would suspect PWR_OK giving nearly 0V.


Modecom(load 12V molex fan, ~0.5A // no load):
+3.30V: 3.47V // 3.33V
+5.00V: 5.11V // 5.09V
+12.0V: 12.10V // 12.23V
+05VSB: +5.00V // 5.00V
-5.00V:
-12.0V: -11.20V // -11.23V
PWR_OK: 5.10V // 5.09V

Blacklisted PSU that never failed to start MSI motherboard. Perhaps because it's so budget that it doesn't have safety features that would be blocking it from posting? From wikipedia about PWR_OK:

Cheaper and/or lower quality power supplies do not follow the ATX specification of a separate monitoring circuit; they instead wire the power good output to one of the 5 V lines. This means the processor will never reset given bad power unless the 5 V line drops low enough to turn off the trigger, which could be too low for proper operation.

Anyway I am currently 80% sure that PWR_OK is at fault, and it's because motherboard is somehow damaging PSUs. And those with more safety features are more prone to this. Fortron is very interesting case as it stops working just to start working after few days, but Seasonic simply died for good. Perhaps there's some odd charge accumulating somewhere. As I have multimeter, I can test more stuff on motherboard but I don't even know where as I have no idea which caps or resistors are related to PWR_OK... maybe some VRM?

But as I've managed to get Seasonic to run using 12V fan bypass, I also want to know what to do to get it working properly, it has nearly 0.0V on PWR_ON. Perhaps I should start separate topic on this issue?

And by the way, does ErP has any relation to PWR_OK?

[tartuf]

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

To add about odd PWR_OK behavior on my Seasonic(and probably Fortron too when it's playing dead):

When I connect PAPST 4112GX fan which is around 6-7W on 12V(and I can connect it to any 12V, not only molex) the PWR_OK almost always goes to 5.04V from 0.669V. I wrote always, because out of around 100 tries maybe two times it didn't work when PSU was already running for like 10min. It always worked when PSU was being cold-booted with fan already attached. Now, when it works I can disconnect fan anytime and leave PSU running with normal PWR_OK Voltage. In fact if I will turn PSU off and then turn it on like 5 minutes later, even without fan connected it will have normal PWR_OK Voltage. Such way, I can turn it off and on do it maybe 2-5 times more and it will work. Then PWR_OK goes down to 0.669V and stays there until I reconnect fan. Moreover if I flip 0/1 switch quickly few times even when PSU does not produce proper PWR_OK Voltage, from time to time it might kick in and launch properly this way.

When I connect fan to boot PSU properly with 5.04V on PWR_OK, then disconnect fan and leave PSU working for short time, and then turn it off to see PWR_OK readings, I find some oddities. Note that I am turning PSU off with 0/1 switch while leaving it connected to power plug and paper clip is always on(like power button would be held all the time). So after switching PSU off with 0/1 switch, the Voltage quickly drops down to around 0.033V and then goes down to like 0.004V but after that it doesn't go to 0, but instead rebounds back to 0.005V and up to like 0.033V(and then it goes down to 0.000V). And when it's bouncing up like that I can sometimes turn it on without fan connected and PWR_OK will work.

[jiroy]

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

To add about odd PWR_OK behavior on my Seasonic(and probably Fortron too when it's playing dead):

When I connect PAPST 4112GX fan which is around 6-7W on 12V(and I can connect it to any 12V, not only molex) the PWR_OK almost always goes to 5.04V from 0.669V. I wrote always, because out of around 100 tries maybe two times it didn't work when PSU was already running for like 10min. It always worked when PSU was being cold-booted with fan already attached. Now, when it works I can disconnect fan anytime and leave PSU running with normal PWR_OK Voltage. In fact if I will turn PSU off and then turn it on like 5 minutes later, even without fan connected it will have normal PWR_OK Voltage. Such way, I can turn it off and on do it maybe 2-5 times more and it will work. Then PWR_OK goes down to 0.669V and stays there until I reconnect fan. Moreover if I flip 0/1 switch quickly few times even when PSU does not produce proper PWR_OK Voltage, from time to time it might kick in and launch properly this way.

When I connect fan to boot PSU properly with 5.04V on PWR_OK, then disconnect fan and leave PSU working for short time, and then turn it off to see PWR_OK readings, I find some oddities. Note that I am turning PSU off with 0/1 switch while leaving it connected to power plug and paper clip is always on(like power button would be held all the time). So after switching PSU off with 0/1 switch, the Voltage quickly drops down to around 0.033V and then goes down to like 0.004V but after that it doesn't go to 0, but instead rebounds back to 0.005V and up to like 0.033V(and then it goes down to 0.000V). And when it's bouncing up like that I can sometimes turn it on without fan connected and PWR_OK will work.

Nice efforts ..Keep it up