Can I substitute these capacitors on a D865PERL motherboard?

[ribcage]

Hello, I'm repairing an old Intel motherboard D865PERL and I wonder if I can substitute capacitors in this way:

Nichicon VR(M) 470uF 10V 6.3x11mm B0341 85*C (red in picture)
and
Nichicon VR(M) 100uF 25V 6.3x11mm B0341 85*C (green in picture)
with Suncon "low impedance" 330uF 6.3V 6.3x11mm 105*C

Nichicon VR(M) 470uF 16V 8x11.5mm B0311 85*C (blue in picture)
with Suncon "low impedance" 560uF 6.3V 8x11.5mm 105*C



These are the most similar I can find of a reputable brand. I don't have much understanding but I have the impression that these small motherboard capacitors are never exposed to voltages higher than 5V? None of them are leaking or bulged but I did replace one Nichicon HM(M) that was bulged which solved numerous issues with the motherboard but I'm still having stability issues. That's why I'm considering replacing all the Nichicon capacitors on this board because they all have a similar production date (2003) and I've read about the faulty Nichicon capacitors back in the day.

[Per Hansson]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Nichicon VR is a general purpose capacitor (all 85°C caps are).
It is therefore better to stay closer to the originals spec and just go up to 105°C temp rating.
Panasonic NHG or Chemi-Con KMG would be better choices.
Also the caps may be used on 12v circuits so do not downrate them without checking with a DMM first.

[zunasthegreat]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Back in the day of P3 most used voltage was the 5V line, 12V line was barely used and at most I guess good 16V caps is the max rated Voltage you need. In any case use multimeter first like PH said to be 100%. Its also common from that ERA to find higher Voltage rated Caps on boards, if I remember it right it was a "duct-tape" fix to improve ESR and go a tad cheaper.

[kaboom]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I've used PW, HE, FC and FR as replacements for the GP caps on intel-oem boards.

Most scattered around the board are tolerable if GP; the ones around/for the PLL, NB & DIMM slots should be low(er) impedance. Follow the traces. That board uses a switching converter for the RAM and(?) part of the chipset. This is located below and to the right of the DIMM slots.

I've successfully substituted Panasonic FR for most caps on boards of this era, with exceptions for any switching converters. Looks like there are HMs for the VRM-in. You can replace with HZs, if you can get them. The yellow "functional polymers" on VRM out can be replaced with Nichicon LF/LE. VRM-out will have heavy power & GND planes, you need a good iron, preferably a solder station.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

That bulged Nichicon HM 2200uF 10V I replaced solved booting issues and RAM stability issues however I replaced it with a KMG 2200uF 25V that I had at hand bu it's a general purpose capacitor and it's even older than this motherboard. Posting here made me realize I did a stupid thing. So I decided not to recap those small ones for now and I just replaced the KMG with a new Panasonic FM 2200uF 6.3V. I will know in a few weeks if my problems are gone now.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

It didn't help. Could someone give me some hints? I will try to explain the situation. Basically the computer used to never crash except it needed many reboots until it finally booted. Then replacing that one bulged Nichicon HM solved booting issues and also allowed me to use RAM sticks which previously wouldn't work. I ruled out RAM and PSU, i tried different ones. Then I had a total PSU failure because of overheating but now it has no problems with overheating. Since then I'm getting random freezes anywhere in between hours and days, sometimes I can reproduce it with high disk or network activity. The capacitors currently are small Nichicon VR and small Rubycon, two big Rubycon near RAM area, one big Rubycon near CPU alongside 9 yellow Fujitsu polymer, four big Chemicon KZE near the rear I/O panel. There is some discoloration on both sides of the board near two big mosfets/transistors in the center of the board with something that looks like brown flux on their legs. The PCB is a bit warped and domed under on the other side of the CPU. None of the capacitors show visual signs of failure except for some not being perfectly vertical. There aren't any burned chips or other small components. One time after doing a memtest for like 5 hours, which went fine, I booted into OS and minutes later I heard a zapping/chirping sound in the case and it rebooted but never seen it again, no idea what it was. Should I just try replacing all capacitors? I'm having a hard time finding good replacements here. From what I have read on this forum, Nichicon VR and those yellow Fujitsu polymers and maybe KZE are candidates. Excuse me for being a bit confusing but I would greatly appreciate your input.

EDIT: Sometimes the fans turn on for a second upon connecting the PSU to mains.
Once I initiated a shutdown from the OS and it froze up at the very end, then I had to hold the power button to make it shutdown.

[kaboom]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Make/model/pix of power supply? "Ruled out" doesn't help, esp when/after the overheating issue. Hope it's not some Deer or Powmax thing...


HM from 03-05 are suspect. Anything on +12 (VRM IN) gets stressed when questionable power supplies are used.

SMT S478 sockets are also becoming an issue, further worsened with intel's stupid pushpins flexing the board. Connections to the socket can break.

Lost several P4P800s this way. You'd remove the CPU/heatsink & reinstall; it'd POST & run great for hours. After cooling off w/o power or overnight, reboots and flakiness returned. Was able to consistently reproduce the fault by pushing on the HSF.

But a gradual loss of stability over time suggests cap trouble. Start with VRM input caps, and change them to FRs or HZs of the same diameter, but highest capacitance.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I didn't explain it well, the PSU which overheated is dead so I changed it. Right now I'm using a crappy Lite-on 250W because the computer has only a single SSD for storage, no graphics card (headless), so it's not using much power. But I tried using a Corsair HX1050 (1050W) which is very good. Both PSUs have same results, random freezing. There was only one HM on this board, which I replaced with a Panasonic FM like I mentioned.

Also like I said, I kind of can reproduce it with high network or disk activity. If I let it idle it seems to last much longer.

Which ones are the VRM input? The four big ones (1200uF 16V KZE in my case)? Or the 10 smaller ones near the CPU (yellow Fujitsu polymers + 1 Rubycon MCZ)?

[eccerr0r]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Any comments specifically about the D865PERL boards longevity past bad caps and socket failure? I think I have one of these and the cpu has been "overheating" or at least crashing stating that it overheated in the logs. I don't recall bloated caps on the board so not sure what the problem is yet.

I think my board has a plastic carrier and the fan clips onto the plastic carrier, so I don't need to deal with push pins directly to deal with the fan.

I think this cpu (32 bit only)/board is just a little better than my other pres-hott p4 (32/64 bit) board, but it too eats quite a bit of power... both are HT...

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I have this board since about when it was made and it has seen quite a bit of abuse, it would still be running perfectly fine if I weren't so stupid and took better care of cooling. Overheating was entirely my fault for reasons too dumb. The only bad thing with it was one bulging Nichicon HM 2200uF 10V. Other than that it was running perfectly for months at a time. I wouldn't say it's very hot, at least with Northwood P4 2.8 HT. Works with modern SSDs just fine. Modern Debian linux works great too.

Maybe you need to reapply thermal paste?

[momaka]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I've used PW, HE, FC and FR as replacements for the GP caps on intel-oem boards.

Most scattered around the board are tolerable if GP; the ones around/for the PLL, NB & DIMM slots should be low(er) impedance. Follow the traces. That board uses a switching converter for the RAM and(?) part of the chipset. This is located below and to the right of the DIMM slots.

I second this ^

Replace all small caps around the RAM, Northbridge, and Southbridge areas with at least entry-level low-ESR caps, like Kaboom suggested.

If you can't find those, here is a more extensive list of caps and series that you could use.
https://www.badcaps.net/forum/showpo...7&postcount=22

Looks like there are HMs for the VRM-in.

Might be Nichicon HD, too, at least based on that picture. O/p mentioned KZE caps near I/O area, so I think we can safely assume those are the VRM-in / VRM high-side caps. In fact, I've seen a lot of Intel motherboards use KZE and HD caps (typically 16V and 1200 or 1500 uF) for the CPU VRM high-side, so that's probably accurate.

The yellow "functional polymers" on VRM out can be replaced with Nichicon LF/LE.

Would you recommend doing that, though? Or have you seen any failed ones, that is? I'm curious, because I've seen a lot of board with these, but never saw one going bad. Would be good to know if they do, though, because I usually tend to trust those yellow Fujitsus and often ignore them.

The capacitors currently are small Nichicon VR and small Rubycon, two big Rubycon near RAM area, one big Rubycon near CPU alongside 9 yellow Fujitsu polymer, four big Chemicon KZE near the rear I/O panel.

The Chemicon KZE's near the I/O area are your CPU VRM input / VRM high-side caps. 16V, 1200 uF, if I had to guess. You can leave them, as KZE is known GOOD series. Might want to replace them only if you see the PCB area around them really darkened from heat.

The single Rubycon (MBZ series???) along with the 9 yellow Fujitsu FP caps are your CPU VRM output / VRM low-side (i.e. connected to CPU V_core.) Probably everything is OK there, unless kaboom can confirm that those Fujitsu FP caps fail.

The two big Rubycon caps near the RAM slots... MBZ series? If yes, you can probably leave them too.

It's the other small Nichicon VZ caps that kaboom and I mentioned above that you should replace first, as they are the most likely culprits.

There is some discoloration on both sides of the board near two big mosfets/transistors in the center of the board with something that looks like brown flux on their legs.

That's typical for a lot of Intel P4 -era motherboards.

I don't remember off top of my head if those TO-263 parts were indeed MOSFETs or linear regulators (1086 or 1084??)... but yea, just open up any Dell with i865 chipset and Intel -made motherboard, and you will see that discoloration / heat darkening there. In fact, the PC I'm typing this from right now (Dell Optiplex 170L) has those two regulators discolor the PCB around them too.

Basically, the i865 chipset is fairly power-hungry. Why Intel chose to power it and the AGP slot with a linear regulator design is beyond my comprehension (apart from the fact that it's an easy and cheap way to do it with less parts, though not great for efficiency.) Those regulators will dissipate a lot of heat and hence why the board gets darkened / discolored.

That said, those regulators are probably OK. However, any small caps you see around them you should definitely replace, as they are likely dry and developed high-ESR.

It's also a good idea to install a small fan on the NB heatsink. I usually do it with a 40x40x10 mm fan and power the fan with 5-7V so that it's quiet but provides just enough cooling to keep the NB heatsink from getting scorching hot.

The PCB is a bit warped and domed under on the other side of the CPU.

The warping is common on a lot of socket 478 boards. You can thank Intel's silly heatsink design for that. It basically relies on flexing the PCB to keep tension on the CPU heatsink. Depending on how warped your board is, that may or may not be part of the problem with the system crashing.

Some s478 heatsinks are better than others, though - in terms of retention design. The better ones typically have their own spring-loaded (usually leaf-type) mechanism, so they don't flex the board as much. The worst heatsink designs are those with the dual plastic levers with plastic lobes. A lot of these warp the living crap out of the motherboard (literally!) and can even crack BGA connections on the CPU socket. If that's the type of heatsink you have, I think kaboom may be onto something then to suggest CPU socket BGA issues, in which case you may or may not be able to reproduce the problem by flexing the motherboard / inserting pressure on the CPU heatsink (with the CPU heatsink unclamped from the motherboard and whole motherboard laying on a flat testing surface.)

EDIT: Sometimes the fans turn on for a second upon connecting the PSU to mains.

That may be another Intel-specific "oddity".
A lot of Dells with Intel-made boards from the P4 era do that.

SMT S478 sockets are also becoming an issue, further worsened with intel's stupid pushpins flexing the board. Connections to the socket can break.

Lost several P4P800s this way. You'd remove the CPU/heatsink & reinstall; it'd POST & run great for hours. After cooling off w/o power or overnight, reboots and flakiness returned. Was able to consistently reproduce the fault by pushing on the HSF.

Yup.

I have a Shuttle PC with an LGA775 P4 like that. Except on that one, I believe I caused the failure myself. Long story short, I can get the board to boot normally if I put downward pressure on the CPU heatsink. No pressure = no boot or POST.

Also like I said, I kind of can reproduce it with high network or disk activity. If I let it idle it seems to last much longer.

That may be a clue.

Network and HDD activity go through the Southbridge. So I think that concurs with the above advice to change most caps around the RAM, NB, and SB areas, in addition to the caps around the hot-running TO-263 regulators.

Also, try this last test: use only one stick of RAM and see if issue persists. Make sure to try it with different RAM slots as well. Let us know what you get for the results on this test.

[kaboom]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I didn't explain it well, the PSU which overheated is dead so I changed it. Right now I'm using a crappy Lite-on 250W because the computer has only a single SSD for storage, no graphics card (headless), so it's not using much power. But I tried using a Corsair HX1050 (1050W) which is very good. Both PSUs have same results, random freezing. There was only one HM on this board, which I replaced with a Panasonic FM like I mentioned.

Replacing HM with FM is OK. Re: power supplies- I'll elaborate below, but your Lite-on may not be as crappy as it seems. Generally solid, but sometimes poor cap choices, as dictated by the OEMs. Assuming, say, it's a "LiteOn for Dell/HP/etc." The HX1050 is not only very good, as you say, it's likely overkill by 5 times!

Also like I said, I kind of can reproduce it with high network or disk activity. If I let it idle it seems to last much longer.

NB power, some voltages of which are shared w/ that AGP slot. I think there are several in the vicinity of ~1.5v; maybe a ~1.2v also. I've seen silkscreens for NB caps (whose traces are rather heavy, implying a major NB rail) with larger diameters than the caps actually populated. It's been several years since I've had any 845/865 boards, but I'd usually put a 1200-1800u 6.3v HM-HZ-FR-FM there. Stock was usually no more than 220u 6.3v, implying the upstream supply was reasonably clean for the NB to work OK. When the output cap on that (usually) buck regulator fails, the now-too-small cap at the NB is insufficient for reliable operation. "Strident," "lockstep," "iterative" things (memtest, CPU load tester), paradoxically, sometimes passed. But fire up "Duke Nukem" or many browser tabs and total failure.

Which ones are the VRM input? The four big ones (1200uF 16V KZE in my case)? Or the 10 smaller ones near the CPU (yellow Fujitsu polymers + 1 Rubycon MCZ)?

Bolded ones are.

If you had power supply problems, the one that burned up/overheated, the input caps take a beating. Not only would they have to supply the pulses drawn by the high side MOSFETs, they'd also be dealing with severe ripple coming in on +12. Can you pull the board and solder equivalents across the 1200/16s? You'd have to test the board out of the case for this.

What brand is the failed PSU? Some are easily capable of 3v+ ripple on +12 when "having a bad day."

[kaboom]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Lots of ground to cover, so doing multiple replies.

Would you recommend doing that, though? Or have you seen any failed ones, that is? I'm curious, because I've seen a lot of board with these, but never saw one going bad. Would be good to know if they do, though, because I usually tend to trust those yellow Fujitsus and often ignore them.

I'd leave them if OK, as in "not domed." Not that they always do when bad.

The Chemicon KZE's near the I/O area are your CPU VRM input / VRM high-side caps. 16V, 1200 uF, if I had to guess. You can leave them, as KZE is known GOOD series. Might want to replace them only if you see the PCB area around them really darkened from heat.

I err on the side of caution, esp since this system is the same vintage as many other P4s that were ruined with bad PSUs; we don't know what was powering it, that PSU that failed. Not unreasonable to suspect the original PSU didn't fail "cleanly" if you know what I mean!

The single Rubycon (MBZ series???) along with the 9 yellow Fujitsu FP caps are your CPU VRM output / VRM low-side (i.e. connected to CPU V_core.) Probably everything is OK there, unless kaboom can confirm that those Fujitsu FP caps fail.

OP, if you have a good solder station, you can parallel several Nichicon LF or LE across VCore at the same time as you parallel those across +12 on VRM-in.

Leave the FPs for now.

The two big Rubycon caps near the RAM slots... MBZ series? If yes, you can probably leave them too.

I'd like to see the case/airflow, but it passed memtest, so also leave for now.

It's the other small Nichicon VZ caps that kaboom and I mentioned above that you should replace first, as they are the most likely culprits.

That's typical for a lot of Intel P4 -era motherboards.

I don't remember off top of my head if those TO-263 parts were indeed MOSFETs or linear regulators (1086 or 1084??)... but yea, just open up any Dell with i865 chipset and Intel -made motherboard, and you will see that discoloration / heat darkening there. In fact, the PC I'm typing this from right now (Dell Optiplex 170L) has those two regulators discolor the PCB around them too.

Extremely typical, unfortunately. The likes of ASUS sometimes had buck converters instead, but their choice of caps "undid" any advantage gained; much less heat, but unreliable with age since the craps were unfit.

Basically, the i865 chipset is fairly power-hungry. Why Intel chose to power it and the AGP slot with a linear regulator design is beyond my comprehension (apart from the fact that it's an easy and cheap way to do it with less parts, though not great for efficiency.) Those regulators will dissipate a lot of heat and hence why the board gets darkened / discolored.

And because LT108x was a turnkey solution, though a little shortsighted within the bigger picture. And a bit ridiculous once OC'ing, AGP demands and poor case/cooling are considered.

That said, those regulators are probably OK. However, any small caps you see around them you should definitely replace, as they are likely dry and developed high-ESR.

If they're really bad, severe capacitance loss too. Like out-of-spec voltages for step-load transients- like "peaky" network or storage activity.

It's also a good idea to install a small fan on the NB heatsink. I usually do it with a 40x40x10 mm fan and power the fan with 5-7V so that it's quiet but provides just enough cooling to keep the NB heatsink from getting scorching hot.

Wasn't it something like 10-15W, depending on FSB, on those things?
In addition to the NB fan, this is flow-thru case territory for sure. And another BGA problem area, mainly with RoHS stuff. i865 is a little before then, so should still have "good" solder...

Lotta fans & lotta noise =/= cooling, esp with no makeup air!

The warping is common on a lot of socket 478 boards. You can thank Intel's silly heatsink design for that. It basically relies on flexing the PCB to keep tension on the CPU heatsink. Depending on how warped your board is, that may or may not be part of the problem with the system crashing.

Intel cheepness... again.

Rather than provide a means to take up tension, like captured springs (dell) or spring clips (gateway) in the HSF assembly, not-so-intel leveraged (punny) the PCB itself for this. It was often so bad I'd grind down the cams to where when the levers were fully seated, there was still adequate clamping w/o mutilating the board. It wasn't the pushpins per se, even though the HSF base could be swapped with a lever-cam one. It was the several mm's between the HSF seated height and the plane of the PCB.
Which sums up the "P4" era of Intel- use brute force.

Some s478 heatsinks are better than others, though - in terms of retention design. The better ones typically have their own spring-loaded (usually leaf-type) mechanism, so they don't flex the board as much.

"Brand name" PCs or non-intel HSFs, mainly...

The worst heatsink designs are those with the dual plastic levers with plastic lobes.

Those. Are. JUNK!
They were almost right with the levers straight up!
I was finding the lobes to be nearly 2mm too wide.

A lot of these warp the living crap out of the motherboard (literally!) and can even crack BGA connections on the CPU socket.If that's the type of heatsink you have, I think kaboom may be onto something then to suggest CPU socket BGA issues, in which case you may or may not be able to reproduce the problem by flexing the motherboard / inserting pressure on the CPU heatsink (with the CPU heatsink unclamped from the motherboard and whole motherboard laying on a flat testing surface.)

The worst thing about those is even after "rescuing" a board that had that kind, the BGAs can be damaged and still eventually fail with thermal cycling, even after upgraded to a spring loaded HSF running in a well-ventilated case.

I have a Shuttle PC with an LGA775 P4 like that. Except on that one, I believe I caused the failure myself. Long story short, I can get the board to boot normally if I put downward pressure on the CPU heatsink. No pressure = no boot or POST.

My P4P800 died like this. Fully recapped and BGA CPU socket. After an evening of gaming, got no POST the next day. R&R the HSF, it POSTed & booted. Shortly after, got to where it would lock solid when touching/nuding the HSF. Failed completely in a few days. Saved the caps, battery holder & NB HS. It Did this before/after recapping, but "preloading" with the HSF kept it going for almost 3 years. Was a cheap ebay find, retired for no POST, so worth the experience. I got it running for a little while...

Network and HDD activity go through the Southbridge. So I think that concurs with the above advice to change most caps around the RAM, NB, and SB areas, in addition to the caps around the hot-running TO-263 regulators.

I should find/pull a hires image of this board and mark some of these caps for you, OP.

Also, try this last test: use only one stick of RAM and see if issue persists. Make sure to try it with different RAM slots as well. Let us know what you get for the results on this test.

Yes. Despite passing memtest with both, using only one stick draws less current from the NB supply, potentially (punny, again!) improving performance in other areas.

My trick for marginal caps is to use a fan-forced space heater or hair dryer to "preheat" the system. Nothing hotter than what a good shower or bath feels like though! 9/10 times, marginal caps improve to where stability improves. Electrolyte activity increases w/ temperature, so almost-dry caps "work better." Try that, OP.

This is viable even before removing the MB from the case and paralleling test caps across +12in and VCore.

Have fun.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Currently halfway at replacing all the smaller 85*C caps to Panasonic FC and FR. It's not easy for me... If this won't help then I will replace VRM input to FR of very similar specs, except 1500uF, got them already.

Those two Rubycons near RAM are MCZ 560uF, 6.3V, 8x25mm. I can't even find them in the datasheet.

I'm quite sick of testing so I'll just recap for now. I tried three different sets of RAM, the simplest one being 2x256MB, single sided. Usually I run 4x1GB, all combinations yield same results.

The power supply that failed is KM KOREA GP-300ATX from 2003. It ran without a fan for almost a year so I think it's pretty well built. Did some dumb things trying to make the computer quieter, won't happen again. When it failed, I found the PC running but it was frozen. Wouldn't be surprised if those four KZE caps for VRM input were damaged.

The Lite-on PSU, when connected to an ungrounded wall socket, tingles when touched. Not painful but certainly uncomfy. Although I've often had tingles from PCs when I had ungrounded sockets. I measured ~100V from the PSU to ground, mains are 230V. I guess it's normal but still strange.

Those two big mosfet things say PHB11N06LTA. Datasheet says "N-channel enhancement mode, logic level, field-effect power transistor in a plastic envelope using 'trench' technology."

[momaka]

Re: Can I substitute these capacitors on a D865PERL motherboard?

Currently halfway at replacing all the smaller 85*C caps to Panasonic FC and FR. It's not easy for me... If this won't help then I will replace VRM input to FR of very similar specs, except 1500uF, got them already.

If the small caps are not as easy for you, then I suggest going with what kaboom said and just add a few of the 1500 uF caps in parallel (solder them on the bottom of the PCB most likely) to the VRM caps. That would be quicker and take less effort while still able to tell you if that's the issue or not.

Those two Rubycons near RAM are MCZ 560uF, 6.3V, 8x25mm. I can't even find them in the datasheet.

That doesn't look right.
In the MCZ datasheet, the closest cap I can find to that size for 6.3V is 8x20, and that corresponds to a 1800 uF cap. So there's no way your cap(s) there measure 8x25 and are only 560 uF. Either your size measurements are off, or you misread the capacitance / voltage.... OR someone replaced that cap with a fake MCZ before you??
Either way, double-check what you found.

The power supply that failed is KM KOREA GP-300ATX from 2003. It ran without a fan for almost a year so I think it's pretty well built.

That doesn't inspire any confidence.
Running just about any old ATX PSU from that era without a fan will usually result in its caps cooking and failing in short order. So I can only image when that thing failed what it looked like on the inside. And I don't even want to imagine the ripple it spit out on its "DC" rails.

Wouldn't be surprised if those four KZE caps for VRM input were damaged.

KZE's are troopers. I've seen them take a ton of abuse. Not saying that it's not possible for them to be bad... but I'd bet my money on a lot of other things as failed before these.

The Lite-on PSU, when connected to an ungrounded wall socket, tingles when touched. Not painful but certainly uncomfy. Although I've often had tingles from PCs when I had ungrounded sockets. I measured ~100V from the PSU to ground, mains are 230V. I guess it's normal but still strange.

That's because that Lite-On PSU probably has proper Y2-class caps for reducing EMI and RFI. Usually, the way these are connected between Live, Neutral, and Ground creates an AC high-impedance voltage divider path to ground. ~100V sounds about right for 220/230/240V AC line. If you were in North America, you'd see only ~50-55V on the case with the PSU ungrounded.

This IS normal. Running the PSU ungrounded is NOT, though. So that's why you're getting "the tingles".

Let me guess? You never got any shock with the KM Korea PSU, did you? If that was the case, then most likely that KM Korea PSU was probably a gutless wonder that didn't have any proper EMI/RFI filtering. And if that is so and you really ran it without a fan for a year.... ... then I can't blame that Intel mobo for not working right anymore.

Those two big mosfet things say PHB11N06LTA. Datasheet says "N-channel enhancement mode, logic level, field-effect power transistor in a plastic envelope using ’trench’ technology."

Ah OK. I guess I could have taken one of my Intel mobos and checked, but was too lazy to dig stuff out of my closet to get to them, lol.

In any case, even MOSFETs like that can still be used as linear regulators... and I'm pretty sure they are. If they were used in a buck circuit, you'd see some inductors around them (be it toroidal, PI, SMD/SMT, or whatever other type that exists.) Since there are none, this is a linear regulator circuit.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

After changing around 40 caps, the situation didn't change. One thing I noticed when I turned off all fans for a second is that there is a quiet screeching, dirty electrical noise coming from somewhere, couldn't pinpoint it exactly but seems like it's one of the transistors/mosfets. Today I replaced those KZE caps to FR, will see if that helps. Cleaning the holes after removing old caps is such a pain... I want a desoldering station bad.

I didn't misread those Rubycon caps. They are indeed strange, perhaps custom made for Intel or something? I can't take a picture right now without shutting it off but I found one on the net, zoom in on that tall cap near RAM slot:

[momaka]

Re: Can I substitute these capacitors on a D865PERL motherboard?

After changing around 40 caps, the situation didn't change.

That's not good.
Time to take the board out of the case and test it on your bench. While at it, see if putting slight pressure on any point on the board makes it freeze or crash (particularly on the CPU heatsink and nearby area.)
When you are testing the board flat on the bench, you can also try unclamping the CPU heatsink (but do NOT remove the CPU heatsink) to see if that changes behavior.

One thing I noticed when I turned off all fans for a second is that there is a quiet screeching, dirty electrical noise coming from somewhere, couldn't pinpoint it exactly but seems like it's one of the transistors/mosfets.

Probably coil whine from the CPU VRM. A lot of Intel boards from that era did it. Most of my Dells you can hear their coil whine over the fans if the case is open.

But that's a good observation there and we will keep it in mind. After all, it may also be a voltage regulator oscillating somewhere too.

Cleaning the holes after removing old caps is such a pain... I want a desoldering station bad.

I've used an expensive desoldering station before and all I can say is they are not that great for motherboards.

For cleaning out holes, the best method I find is just with a dental pick or sewing needle and melt solder on one side while pushing the solder out of the hole on the other with the pick/needle. (That said, if you do use a sewing needle, it might help to glue a "head" to it on the other side, so it's easier hold and push through.) And of course, good flux and fresh solder helps a lot too, along with a powerful enough iron. Sometimes I also struggle on a board simply because I forget to change the solder on my iron's tip when it becomes too oxidized.

Using the correct temperature helps too, so that the solder on the tip doesn't oxidize and then become terrible at transferring heat or not wanting to wick to the board.

I didn't misread those Rubycon caps. They are indeed strange, perhaps custom made for Intel or something?

Well, I stand corrected then.
Those were indeed likely a custom order for Intel.

I was just surprised how big they are relative to their capacitance. Like I said, I know the latter Xbox 360 revisions came with Rubycon MFZ 6.3V, 820 uF caps in 8x25 size... which is a pretty big size for that capacity (and MFZ is very similar to MCZ). So I guess this being a slightly older board, Rubycon probably hadn't quite mastered the ultra-low ESR electrolyte used in MCZ caps, so perhaps that's why those 6.3V, 560uF caps are so large - to get ultra-low ESR.

[ribcage]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I guess it's hard to fix stupid. Turns out it was my SSD. At least someone might find it useful to know that Nichicon VR, Rubycon YK, Chemi-con KZE, Rubycon MCZ and yellow Fujitsu caps near CPU on these boards are most likely healthy capacitors if they look fine.

The SSD is Kingston A400 960GB. Basically I installed OS on a different drive and put the ssd in as secondary. I noticed the ssd was very slow to read, like 4MB/s but for some reason only old files were so, if I wrote new data it was fast to read. First I tried reading the entire disk from start to end, took a few days and didn't help. Then I ran a linux prog "badblocks -n", which does a non-destructive read-write test, writing random data to a block, then back the original data. This took a few days and now the SSD is at full speed, faster than it ever was. What could also work is dd if=/my/ssd of=/my/ssd. Which should also rewrite/refresh all the data, except the SSD might detect identical data present and might not do anything. On the badblocks program I also used -b 4194304 -c 1, to do it in 4MB chunks, 1 chunk at a time, because default is 64 kilobytes at a time and since SSDs can only erase like 1MB or more at a time, small chunks would cause too much writes and erases and wear out the ssd. The ssd started crashing the PC only after the PSU failure but it became slow even before the psu failure. This might be related to bad caps, because when I had that single bulged Nichicon HM, I would need to shut down and start the PC like 20 times at worst, to get it to boot and SSDs don't like power loss as far as I know.

[momaka]

Re: Can I substitute these capacitors on a D865PERL motherboard?

I guess it's hard to fix stupid. Turns out it was my SSD.

Well, at least you got to the bottom of the issue. And good thing it's nothing more serious like we thought it might be.

At least someone might find it useful to know that Nichicon VR, Rubycon YK, Chemi-con KZE, Rubycon MCZ and yellow Fujitsu caps near CPU on these boards are most likely healthy capacitors if they look fine.

The KZE, MCZ, and yellow Fujitsu, yes.
But I still think it's a good thing you replaced those old VRs and YKs. On a hot board, they are probably getting close to the end of their life now (if not past it.) An ESR meter would have answered that nicely for us, though.

The SSD is Kingston A400 960GB. Basically I installed OS on a different drive and put the ssd in as secondary. I noticed the ssd was very slow to read, like 4MB/s but for some reason only old files were so, if I wrote new data it was fast to read.

Yeah, that's sort of the issue I started running with on my (now former) work laptop, except mine didn't get to being that bad, where it would cause crashes. Just my boot times got a lot longer - and all of this in just a year of use. It was a cheapo Inland Pro series 120 GB SSD.

That said, there may be a way to refresh the HDD data, mentioned here, through software so that this doesn't happen.

I haven't tried it, though, simply because my work contract ended before I got to use my work laptop more. I do still have the HDD from that work laptop, since I bought it with my own money (and put back the clunker mechanical that was in there before returning the laptop, lol.)

The ssd started crashing the PC only after the PSU failure but it became slow even before the psu failure. This might be related to bad caps, because when I had that single bulged Nichicon HM, I would need to shut down and start the PC like 20 times at worst, to get it to boot and SSDs don't like power loss as far as I know.

I doubt the bad cap caused the SSD data failures you were seeing. Probably something that was going to happen anyways. And then the bad cap(s) was just another issue on top.

But again, glad to see this is all squared away now.