Replacing bad caps on motherboard

Hello,

i have an asrock x58 extreme motherboard. Problem is, when i benchmark graphics, it reboots randomly. Now i found some capacitors on the mainboard which look dirty and i want to try to replace them. They're close to the first x16 pci express slot.
Question is, whether i just can take a capacitor with same uF, high temperature withstand and replace them? Or, does the ripple current also apply and do i have to take notice of that.

Added a photo of the capacitors.

Grtz
34 answers Last reply
More about replacing bad caps motherboard
  1. They should be replaced by capacitors having the same specs. You tried a different GPU and PSU and it still reboots?
  2. Different gpu, yes an gt 710, it only takes about 30 watts, so it doesn't really take full power. It doesn't reboot, but it's so slow. But thing is, if i use this radeon 5970 in another pc with amd board, it works, benchmarks without problems. Same specs: these are i believe Nichicons caps: LF 100C 5K 9X2 and LF560J 5K 9X2. They are 10uF and 56uF. Probably the replacements are Panasonic 25SEP10M and 20SEP56M.
    The caps are a bit coloured, so are not clean anymore. Read somewhere on the internet that that means they're bad caps. And they are right next to the pci express slot i use for the radeon gpu. The psu is like only 2 or 3 years old, but the mainboard shows at 3.3 volts only 3.0 volt(could also be due to bad caps). Don't have another psu with 8 pin mainboard connection, so can't test that it's the psu.
    Yes, i lend the radeon out to someone, he also had reboots. It doesn't really make sense, one thing says it's the mainboard, another it's the radeon.
  3. Those solid state caps as opposed to electrolytic ones are difficult to troubleshoot by looks only, only very high overheating may discolor them.
  4. would be handy to have the code scheme they used at nichicons back in 2009 to letter their caps.

    This is the way they letter their caps now:

    Image
  5. If the Radeon 5970 caused the same reboot issue in another PC, then look no further.
  6. Did another test with the radeon in an amd pc. No reboots there. 4k furmark with 29 fps avg.
  7. CountMike said:
    Those solid state caps as opposed to electrolytic ones are difficult to troubleshoot by looks only, only very high overheating may discolor them.


    Hey Mike, the board is 8 years old, might have gotten some users damage. The only card that gets real hot it the radeon 5970. The complete house of the card gets hot too. Then the caps are right beneath the card, so might have taken some heat too.
  8. Here the specs of the lf series: Image.

    LF 100C 5K 9X2 =cap1
    100: 100 uF
    C: 16 volt

    and
    LF560J 5K 9X2 =cap2
    560: 560uF
    J: 6.3 volt

    Now i've found two cap's matching for the most specs. Just ESR for one should be 24mOhm and the cap i looked at at the shop was 12 mOhm and the leakcurrent of the new cap is way lower.
  9. Fixed it with thisi: Image

    Must add, overclocking your cpu with the radeon 5970 doesn't make it very stable. But non overclocked cpu, it's very impressing to see the 2 gpu's working at full capacity.
  10. Update: it's been running good for almost a day and this morning it was soup again. Right now, it's running quite stable with only one core. That makes it a 5870 or 5850 gpu.An Mstech 850 watt psu, according to the internet it's a bad psu with 350/400 real watts. Then this seems quite logic to me. The psu can barely handle the gpu and i should get a better one.

    *edit: reducing gpu clock frequency from 750mhz to 610 mhz seems to help for now running it in crossfire mode. The voltages the motherboard reads from the psu are somewhat low:3,3 volt=3.05volt, 12 volt = 11.6 volt on the mainboard. Checked the caps of the psu, but they seem all right, no bulges or leaks.
  11. Now this morning it was soup again. What i tried this time, was monitoring the voltages during a benchmark. Now the 12 volt drops considerably and might be the cause of the reboots. Image
  12. Same issue with the 3.3V rail. Using a DVM would be more accurate than software.
  13. GhislainG said:
    Same issue with the 3.3V rail. Using a DVM would be more accurate than software.


    Hey Ghislain,

    been testing the psu on another mainboard( much lighter one, dual core, 4 pin motherboardpowerconnection) and there the voltages in the bios seem all right. After reading some of my psu mstech 850 watt on the internet, it seems some say it's 400 real watts. Trying another psu, a zalman 400 watt with molex to 8 pin EPS socket converter.
  14. This is the hwmonitor screen after a successfull run of 720p in furmark: Image

    My psu has 2 +12v rails, and only one of them dropped a lot. Reconnected the powercables.

    This is the hwmonitor after a successfull run of 1080p in furmark: Image
  15. The voltage on the +12V rails should stay within the specs: +11.4 to +12.6 volts.
  16. GhislainG said:
    The voltage on the +12V rails should stay within the specs: +11.4 to +12.6 volts.


    It's closer to 11 volt. But still running. Used a converter from molex to pci express power( 6 pins) and the regular pci express power (8pins) from the psu.
    I do have lying around a Zalman 400 watt psu. It is quite stable running an amd quad core@4ghz, 3gb ddr3 memory and the radeon 5970.
    Could try that psu on my machine: xeon cpu, radeon 5970, 8gb ddr3 memory.
  17. You lose nothing trying it, but check the 12V rail(s) output. 400W is meaningless if it's an old P4 PSU.
  18. GhislainG said:
    You lose nothing trying it, but check the 12V rail(s) output. 400W is meaningless if it's an old P4 PSU.


    Jow, replaced the psu with the 400 watt zalman. Ran half furmark's 720p benchmark and these are the results: Image.

    The low 3.3 volt is solved by replacing the psu. The 12 volt is still low, like 11.1 volt.
  19. GhislainG said:
    Same issue with the 3.3V rail. Using a DVM would be more accurate than software.


    Then probably your usb ( front and back) is also not working as it should. With the zalman psu and 3,3 volt, the back/mainboard usb connection works better than with the 3.1 v psu.
  20. Could you borrow a known good and powerful enough PSU to test? You can obviously replace the caps, but I'm not yet convinced that would resolve all issues.
  21. GhislainG said:
    Could you borrow a known good and powerful enough PSU to test? You can obviously replace the caps, but I'm not yet convinced that would resolve all issues.


    Hey, i took the mstech 850 psu, and put it in the amd mainboard and it gives me right voltages, even with the ati 5970 inserted. So it isn't a psu problem, but a mainboard.

    Thought, maybe it's that simple to make photo's of the caps on a mainboard( discolouring is more easy visible) and replace all the caps that show discolour.

    One thing, on the x58 mainboard the 3,3 volt with the mstech psu is not good, with the zalman it is 3,3 volt. But if only used 4 pins to the mainboard (amd mainboard) it shows 3,3 volt.
  22. Ultimately it's your decision. Mine would be to either replace the motherboard or sell the good parts and build a new and faster system.
  23. GhislainG said:
    Ultimately it's your decision. Mine would be to either replace the motherboard or sell the good parts and build a new and faster system.


    Still thinking of replacing the caps. The brown visible on top op the caps, it's leaked from the inside of the cap. That and it's like 8 years old mainboard. Don't think it's cheap to get another mainboard with cpu, but i could buy the same socket mainboard as new instead. The overclocked xeon cpu, it's very fast when all cores are active. Like a much newer i7 cpu.
    Had some trouble with the booting of the computer, but after cleaning the ram slots, it boots right.
    Btw i found another 6.3 volt cap which is brown, all near the pci express slot though. A bit strange, that the other psu gives better 3,3 volt readings and the same 11.1 volt. And that last sateruday, the complete day, the mstech psu and the xeon and radeon 5970 were working right and at full speed. While the day after it didn't work. Only thing i did was clean the mainboard with an air compressor.
  24. You can definitely replace the caps; it won't be expensive and it may resolve the issues.
  25. GhislainG said:
    You can definitely replace the caps; it won't be expensive and it may resolve the issues.


    Hey Ghislain,

    tried to replace the caps, but couldn't get the caps out of the mainboard. So i put the mainboard away, without cpu, cooler and memory. Now, after some days i took the mainboard back, took another psu, took my i7 920 cpu and 2 gb memory and put it together. The voltage says 12.2 volt, so now it's a good voltage.
    The caps , it seemed only like 450 degrees celsius to melten the tin. Tried that but i couldn't manage that. And it's really tiny. My desolder iron was like 2mm, and the caps pins were like 0.7 mm. Been searching for a better desolder iron, but i guiess it takes like 150 euro for such.
    Do you think, lying the mainboard in a fridge can solve issues with the 12 volt?

    grtz

    *edit: guess the radeon 5970 just lowers the 12 volt with 0.4 to 0.9 volts.
  26. The fridge isn't a solution to motherboard issues unless it overheats and you keep it in there while using it (that would in turn create potential issues like high humidity, etc.)

    My first reply on January 2 probably was correct; couldn't you use another GPU? It should be 6-7 years old by now.
  27. GhislainG said:
    The fridge isn't a solution to motherboard issues unless it overheats and you keep it in there while using it (that would in turn create potential issues like high humidity, etc.)

    My first reply on January 2 probably was correct; couldn't you use another GPU? It should be 6-7 years old by now.


    Hey Ghislain,

    well i got a brand new mainboard. It completely doesn't support the radeon. Must be some incompatibility between the x58 chipset and the amd chipset. Put the old mainboard back together with an i7 cpu i had lying around, and voltage of 12 volt is like 11.8 volt, but sometimes it's 11.9 volts. On the xeon cpu, i had installed before, there was a little black mark on the cpu, between the pins. Yesterday, i ran, with an updated version of furmark, a benchmark and it ran well. Just the gpu temp was almost 100 degrees celsius. So i installed msi afterburner and activated the fan profile. I don't completely trust the complete pc, cause in the past one day it was running fine benchmarking the gpu, and the other day it completely wasn't. But since then, there has come out a cpu patch. So they found a bug in the cpu, which now is resolved. I hadn't installed that patch yesterday, it was still running fine gpu benchmarking it. There is a small chance, there are hackers around the world which dive into that bug and destabilize the pc. Since it's a very old system. You can install, or it will automatically install, when you run iobit care system cleaner. But i must say, the graphics 4k or not, of a displayport to hdmi 2.0 converter, is very nice. Might be an old gpu, but the colours are wonderfull.
    So now i got the asrock x58 extreme, with a i7 920 cpu, running at 2.4ghz.
    One thing, when finding errors on a mainboard like this, i shouldn't look too much at voltages. It can still run fine with a little lower voltage. The xeon/motherboard power input for the cpu, well it could be some 12 volt links. And if overclocked a lot, maybe the internal resistance it getting more and the 12 volt drops. One psu can handle that better than others.
    So for yesterday, gpu and mobo were working right.

    Nice day!
  28. That's good news!
  29. GhislainG said:
    That's good news!


    Hey Ghislain,

    it worked, even overclocked cpu, for a week. This morning, after a night putting it off, it rebooted randomly again. But if i put the fsb clock to stocks, it worked again. Seems the radeon 5970 just doesn't like fsb overclocking. Could turn up the multipier though.
    At best, when overclocked fsb, the machine goes into windows, (or safe mode), and reboots. So it's not a driver issue. Must be some issue with the fsb pulling the pci express clock up with it.

    *edit: if i put the pci express clock to 90mhz and the fsb to 166mhz it does work.
    *edit2: think if i take of the power connectors to the mainboard for a minute and reconnect it, it would work better again. Then when turned off at night i could try to turn off the psu at the backside instead of just turning the pc off. For now i have to test it with Furmark.
  30. FSB overclocking works as long as the the PCIe bus is kept at 100 MHz. You sure like playing with that hardware more than I do.
  31. GhislainG said:
    FSB overclocking works as long as the the PCIe bus is kept at 100 MHz. You sure like playing with that hardware more than I do.


    Hey Ghislain,

    the pc is still working till date. 4k furmark benchmark reaches 22fps, so i can't game in 4k with it. Even overclocked the gpu from 725/1000 to 800/1100mhz.
    Got a chinese 1366 mainboard, but it doesn't support the radeon 5970 gpu, it doesn't give any image. And can't overclock the cpu even a tiny bit with it. But a lot of settings in the bios i don't have a clue what they are for.
    The cpu fan sinks in rpm when i do a benchmark, probably caused by the 12 volt that lowers to 11 volt.

    Grtz

    *edit: screen flashing probably resolved by upgrading the tv software/firmware.
  32. At least you seem to appreciate playing with legacy components.
  33. Why not u get urself a new x58 mobo
  34. vexicus said:
    Why not u get urself a new x58 mobo

    You missed this in the OP's last post: "Got a chinese 1366 mainboard".
Ask a new question

Read More

Benchmark Motherboards Graphics