Asus P5B

[m5b4]

Hi,

I was doing the 771 mod on one of my computers (with Asus P5B) and when I took the heatsink off, I found the caps around the socket to be leaking. They are TMV 4V 680uF, which, I guess, are used on vrm output. I will need to replace them, but I haven't been able to find the datasheet for them.

I would like to replace them polymer, because I think are better and more durable. But when searching for polymers locally, I found them really hard to get in the Czech Republic. I managed to find a place that has X-CON ULR, which seem like quality caps (at least from I was able to find). But I don't know if they are good replacement for TMV.

I also don't know which capacity to choose. I found somewhere, that when replacing electrolytic with polymer, the capacity should be halved. I'm able to get 330uF 16V, 470uF 6.3V, 560uF 6.3V, 1000uF 6.3V, 1200uF 4V with size that fits the board.

So my questions: Is it a good idea to replace them with polymers? If so, are X-CON ULR a good replacement? If so, which capacity to choose?

[Per Hansson]

That's strange, I would expect such a board to have solid polymers already! (Got pics?)
Anyway the Samxon ULR will work well, but do stay close to the originals capacitance.
The part about halving only really applies to old boards where quite high capacitance ultra low ESR caps where used.

So I'd go for the 560uF 6.3v which are within 20% of the originals...

[m5b4]

Thank you for your answer. These are all the photos I took. I haven't taken a photo of the whole board, so I found one on google. The bad caps are the 9 caps around cpu. I can't take more photos right now, because I already assembled the computer back.

It's quite strange that the computer is running fine, I wouldn't have noticed the caps, if I hadn't disassembled it for the mod. I know that it's probably not a good idea to let it run like this much longer, so I would like to replace the caps as soon as possible.

So the 560uF caps should be fine?

I would also like to overclock it a bit (FSB about 400MHz). Could the lower capacitance affect overclocking?

[Per Hansson]

The 560µF caps should work fine.
I can't locate a datasheet for your original caps but I'm confident they will have lower ESR, being true solid polymer caps...
And thus they should allow you to overclock the CPU higher, but not the memory or chipset obviously.
On that note I noticed you have KZG caps on the board, I'd tell you to change them but on the other hand if the board is working fine maybe leave well enough working...

[Bungz]

Screw that replace all the chemicon stuff on there, nice board and if you are working it hard with a overclock I would spend the time and £ now.

Id also try and find decent polymers to replace the leaking caps, 400fsb is easily doable with a P5B but its still a decent OC.

[m5b4]

Originally Posted by Bungz

Screw that replace all the chemicon stuff on there, nice board and if you are working it hard with a overclock I would spend the time and £ now.

I don't know if I understand it correctly. Are you suggesting replacing all chemicon caps? I thought that they were a good brand.

Originally Posted by Bungz

Id also try and find decent polymers to replace the leaking caps, 400fsb is easily doable with a P5B but its still a decent OC.

Do you think that X-CON ULR aren't decent? What would you suggest?

[Bungz]

Replace all Chemicon from that era, especially KZG they all fail.

I am not as knowledgeable with polymer caps and the different ranges off the top of my head but I would replace polys with polys, especially as you have size constraints due to the cpu heatsink.

[m5b4]

Originally Posted by Bungz

Replace all Chemicon from that era, especially KZG they all fail.

OK, I will take the board out again to see what other caps are there.

Originally Posted by Bungz

I am not as knowledgeable with polymer caps and the different ranges off the top of my head but I would replace polys with polys, especially as you have size constraints due to the cpu heatsink.

The original caps aren't polymer, but electrolytic caps, but I would like to replace them with polymer. From what I was able to find, the X-CON caps are made by Samxon, which should be a good brand. But I couldn't find datasheet for the original TMV caps to compare the specs.

[m5b4]

So I took the board out to take more pictures and make a list of all caps.

4x KZG 1000uF 16V
9x TMV 680uF 4V
8x FJ 1500uF 6.3V
11x TK ATWY 820uF 6.3V
1x KY 470uF 6.3V
10x KMG 100uF 16V
2x HD(M) 330uF 6.3V
1x M 6363U 680uF 6.3V
1x FJS 470uF 16V

Only TVM are leaking, all of the others seems OK. Which of them should I replace?

[Per Hansson]

If there are only 4 KZG caps not much work, replace with something like Rubycon ZLH.

Might be interesting to check them with an ESR meter after.
I see allot of KZG caps running 24/7 for years and just some seem to fail.
I think it's down to production batch etc.

[Wester547]

No, it seems that all KZG, at least those produced from 2001 (their inception) to 2008 (I don't recall a report of a failed KZG with 2009 datecodes or later), are prone to failure because the electrolyte is simply unstable. Way too many reported failures, especially in storage, for it to be limited to a bad batch or two. They sometimes last 24/7 because it only takes time for the electrolyte to break down and outgas, and heat only accelerates all failures. I noticed that the OP's board is full of Toshin Kogyo ATWYs - those are also reputed for their unreliability, as are the ATWB series.

Chemi-con is considered a good brand. The series to avoid from them are KZG, KZJ, KZV, TMV, and TMJ. The TMV is probably about equal to Nichicon HZ. All those series are out of production anyway, superceded by polymers.

[Bungz]

I have the same board beside me that is also covered in TMV and KZG

Thinking about trying these

http://uk.rs-online.com/web/p/alumin...itors/8513848/

Think they should be fine for the VRM low side, someone can correct me if not.

[ChaosLegionnaire]

470µF is too low when the originals are 680µF. typically on these lga775 polymer cap boards, i've seen them use 560, 680 and 820µF caps for the cpu vrm output. either of these 3 will work. 470µF is too low and may cause inexplicable glitches when using a more power hungry cpu like a quad.

[Bungz]

Will step them up then.

Thanks.

Can the OP use RS? I would think they will ship to europe.

[kc8adu]

watch for clearance issues.
the sanyo [email protected] in the badcaps store is a perfect replacement.

[mockingbird]

Here's what you need to change:

4x KZG 1000uF 16V
9x TMV 680uF 4V
11x TK ATWY 820uF 6.3V

You also have two unpopulated positions on the VRM. The one on top is a 16V, the one on the bottom is VRM low ( 1-5V). You should fill those.

Don't leave a CPU in there while you're doing it. It's better to protect the pins with a plastic shield or perhaps a junk old 775 CPU you don't need.

Originally Posted by ChaosLegionnaire

470µF is too low when the originals are 680µF. typically on these lga775 polymer cap boards, i've seen them use 560, 680 and 820µF caps for the cpu vrm output. either of these 3 will work. 470µF is too low and may cause inexplicable glitches when using a more power hungry cpu like a quad.

470uF might be fine, even for a quad 771 xeon, because they only typically consume 80W (eventhough they're specified at 90-120W).

This is especially true if you're using polymers and if you're adding the missing 10th cap. This way, you're only losing 1420uF (6120 - 4700).

I've done a few of these boards so far. One was a polymod with 16V 270uF Nichicon LF on the high and 2.5V 820uF FPCAP on the low, and the computer this is being type on was Nichicon HZ 1800uF on the high and some 8x8 polymer on the low.

So far yours is the first I'm seeing with TMV. All the one's that have been worked on here so far have had those 8x8 Panasonic FL.

I'm attaching the modified BIOS for the Xeon microcodes.

BIOS

Don't expect too much in terms of FSB control with this board. Its FSB ratio divider isn't exactly something to write home about. You might get a bit frustrated with it, but once you learn how it works, you should be fine.

Another tip is to take your time when working with this board. Desoldering and clearing the holes on these boards isn't something that should be done flippantly. Take your time, and if you're finding that the bottom pads are being ripped off (which is not a complete loss because there are pads on the top side as well - albeit they are more difficult to get the lead to adhere to while soldering from the underside of the board), take a break, and pursue your strategy differently.

Finally, pay close attention to that "TK" cap near the southbridge heatsink. It's always almost the first one to bloat on this model. Definitely consider using a polymer for that particular one.

[Behemot]

I'll have a look what I can find. I've run out of 1000/16 ultra-low ESR caps (at least till I get some new Suncons) but hey you can use my KZHs 1000/16 as that's input from the PSU. Should find some used polymers, I would not be afraid of that, have already re-used dozens of them with no problem, they have really long life.

What's the shop with ULR's here? The availability of anything good here is very limited. Besides myself of course, there is one, maybe two shops stocking good caps. But the interest in polymers is still low so I don't stock them myself. I can always do with what I have, incl. those caps ditched from dead boards…

[m5b4]

I'd like to thank everyone for help.

I've decided to replace all the TMV and KZG caps, because they are on the cpu vrm. But I'm not sure about replacing TK ATWY, they all look OK. Do you think that it would increase stability when overclocking?

Should I also replace the KZG with polymers?

I've also remembered I have one broken board lying around. I've checked and found that there are some polymers there that I could reuse. There are ten caps around socket, which I think are 820uF 2.5V, but they are in 6.3x8mm form factor. Do you think that I should try to fit them there? There are also 4 other polymers. I don't know which models they are, but I'm uploading photos of all of them.

Originally Posted by mockingbird

You also have two unpopulated positions on the VRM. The one on top is a 16V, the one on the bottom is VRM low ( 1-5V). You should fill those.

I'm not sure what unpopulated positions you mean. I've marked it on the photo. Is that correct?

Originally Posted by mockingbird

I'm attaching the modified BIOS for the Xeon microcodes.

Thank you, but I've already modified and flashed the latest BIOS according the guide on delidded.com. Or does this BIOS include other improvements?

Originally Posted by Behemot

I'll have a look what I can find.

It would be great if you could find some replacements for me. I already bought some caps from you for another board I was repairing about a year ago.

Originally Posted by Behemot

What's the shop with ULR's here? The availability of anything good here is very limited. Besides myself of course, there is one, maybe two shops stocking good caps.

I found them on tme.eu and on gsmcentrum.cz. It seems like they are reselling the stuff from TME, but they have better prices (at least on the caps I checked). What are the other shops selling quality caps?

[Behemot]

Think I've run into that GSM stuff but they have almost no stock.

I found some used D8×8 560/6.3, 560/4 and maybe few others, OS-CONs, Chemi-Cons and Fujitsu polymers. The 16V caps can be anything from low-ESR to polymers, it's the input from PSU so it often handles lot of different values.

The red ones with strange logo are Apaq, I'd stay away from that. The FP ones are Nichicon FPCAP but there is no way to identify the exact series.

EDIT:// The blue ones are Chemi-Con PSE.

[m5b4]

Originally Posted by Behemot

The 16V caps can be anything from low-ESR to polymers, it's the input from PSU so it often handles lot of different values.

OK, but what would you recommend?

Would you also have a polymer replacement for the TK cap near the southbridge as mockingbird suggest? Do you think that I should replace all the TK caps?