Intel D865GLC, higher uF

[DrZaius]

I've noticed that some Intel boards on the forum are being recapped with higher uF than the original caps. What are the advantages of doing this and how do I know that recapping with a higher uF will not destroy my board?

I will be recapping an Intel D865GLC and a FIC AM35 for the first time. The Intel has bad Nichicon 820uF 6.3v and the FIC has bad S.I. 1000uF 10v. I'm planning on replacing them with Panasonic FM equivalents, United Chemicon, or whatever I can get from digi-key.

Threads that mention higher uF for Nichicon 820uF 6.3v replacements:
Intel D865PESO
Just recapped Gateway 310 Series (Post 10)
Blown Caps Intel D865glc (Post 10)

I can try to post photos of either board if anyone is interested.

[DrZaius]

Re: Intel D865GLC, higher uF

Attached image of bad SI cap.

[willawake]

Re: Intel D865GLC, higher uF

well there is firstly the esr of the caps to take into account, the lower the esr the less capacitance you can get away with. so say if you have a typical low esr cap 1000uf, you could replace that with a lower esr cap like 820uf nichicon HN, or rubycon MCZ or with a very low esr polymer cap oscon SEPC 560uf.

these would probably work out ok

intel makes the same kind of substitutions for vrm out on their boards, can be a mix of 820uf nichicon hn or 820uf rubycon mcz or 560uf oscon sepc polymer or 560uf fujitsu polymer etc or just one of these types in a row instead

on the other hand of course if you have an 820uf nichicon HN then it would be ok to use a cap with higher esr like a panasonic fm but at 1000uf. It would be also quite reasonable to use 1200uf or 1500uf. These would be acceptable substitutions. ideally if you are using the next step up in capacitance its perfectly fine. generally the board manufacturers are using the minimal bulk capacitance, some are worse at that. so if you take that into account then you dont worry about increasing the total capacitance. it may increase overclocking performace but who knows.

i think its best to stick with similar or lower esr caps in the critical areas like vrm out and vdimm out though otherwise overclocking performance would probably be affected. I am not talking about capacitance here though, just esr.

still the board is not going to catch fire or something if you mess with the capacitance. it will affect functionality if you use general purpose caps ie not low esr caps though.

pity i cant give a scientific explanation hehe

[DrZaius]

Re: Intel D865GLC, higher uF

Thanks willawake! So the crap Nichicon 820uf are a lower ESR than the Panasonic FM? I looked at the digi-key capacitor catalog, but didn't see any ESR mentioned in the product specs. I was going to get Panasonic FM because I was going to combine them with a non-related digi-key order. What is your recommendation for Nichicon replacement?

[willawake]

Re: Intel D865GLC, higher uF

check impedance value
yanz made some tables which may be useful but still you will need to hunt in the various catalogues
https://www.badcaps.net/forum/showth...2&page=1&pp=20

[gonzo0815]

Re: Intel D865GLC, higher uF

which series was used on this board? If HM i think the Panasonic FM should be fine, if HZ then may be a ultra low esr series would be better.
For the FIC board the FM`s are pretty sufficient.

If you have problems sourcing the caps, you can buy them at badcaps.net or from BigPope if your from outside the US.

[willawake]

Re: Intel D865GLC, higher uF

could use chemicon psa 680uf 2.5v on vrm out like this intel 865 gbf board available from digikey

[DrZaius]

Re: Intel D865GLC, higher uF

The Nichicons on the Intel are labeled HN(M) and all but one are located in a line along the CPU. Along with the Nichicons are yellow 560uf 4v that don't look damaged. I didn't want to remove the heat sink to see what brand they are, but they look exactly the same as:

https://www.badcaps.net/forum/attach...achmentid=4169
https://www.badcaps.net/forum/attach...achmentid=2424

I think I'll forget the digi-key order and get some Samxons or Rubycons from Topcat to replace the bad Intel Nichicons and the SI on the FICkle board.

[PCBONEZ]

Re: Intel D865GLC, higher uF

If it helps the the original rating on the SI 1000uf 10v is Ripple 710 and ESR .090

SI only has one series of 105c caps.

[bgavin]

Re: Intel D865GLC, higher uF

I keep an ever-growing spread sheet of capacitor stuff on one of my web sites.

www bgavinsound com reference

You can fill in the dots and slashes as required. You may have access blocked, depending upon the location of your subnet. I get tired of the dicks in Eastern Europe and Asia using my site for DOS testing, so I block many of those subnets.

[PCBONEZ]

Re: Intel D865GLC, higher uF

Except Least to Best goes FC, FK, FM, FJ

Taking 1000uf-1200uf at 10v in 8x20mm can size as the comparison mark:
(Can size matters more than uF when comparing ESR/Ripple.)

Series: Ripple/ESR

FC: 995/.069
FK: 1220/.044
FM: 1560/.030
FJ: 1870/.016

For comparison
MBZ: 1870/.019 (Used 1500uf for the 8x20mm can)

FJ is rated better than MBZ.
MBZ is rated better than FM.

There is also a Panny FL series which is sort of rare to find.
From what I can tell FJ and FL have the same ratings.
I don't have a complete data sheet for FL. - Just specs on a couple sizes.
FL's are for some direct market I think and they don't show up for sale to the public very often.

.

[Tom41]

Re: Intel D865GLC, higher uF

I hear that most of the motherboard caps are in groups, doing things like VRM or smoothing signals from PCI slots. When they're like that (often in a line), it's most likely that all the caps are connected in parallel. So rather than 5 1000uF caps, you're effectively getting a single 5000uF cap.

I also hear it's possible, if you know which caps are in parallel, to replace bad caps with fewer capacitors - provided they're on the same parallel circuit and the voltage and capacitance ratings add up to the same as the originals.

Slightly off-topic I know, but am I right in thinking this way?

[PCBONEZ]

Re: Intel D865GLC, higher uF

Groups: Typically you have the:
-
CPU VRM caps.
-
An input filter to that which may be near it or near the PSU connector.
(Usually has the bar shaped wire wrapped core inductor rather than the doughnut shaped cores like in the VRM.)
-
Caps on smaller VR's for specific chips or slots. There will be a VR chip nearby.
(These may have input filters too.)
-
The spattering of caps with not associated with a VR are used to keep the rails clean as the traces move around the board. These keep 'cross-talk' noise and noise from add-in cards off the power rails. Basically they assist (and are in parallel with) the caps in the PSU but are close to the loads. [The number of these used will depend on how long the traces are on the board and how branched (tree like) power distribution layout is.] - These caps are in parallel with each other but it's not to raise ratings it's to shorten the path to a filter. If they weren't there any ripple present from a non-PSU source would have to pass through the motherboard back to the PSU (or another branch line) to get filtered out. - They are simply a LOCAL "bleed-off" for ripple.
-
Those small uF (usually higher volts) caps used to keep EMI from coming IN through external wires. (USB, Sound, PS/2)
-
Lytics without vents are usually just for signal coupling and don't have so much water in the electrolyte to be a problem.

As to what you are thinking about (using caps in parallel for something like that reason) is most common in CPU VR's but it's not to raise capacitance so much as to raise the circuits overall ripple rating, lower ESR, and spread the heat load. - Spreading the heat load is a huge consideration that few people think about much.
[Removing ripple heats up caps internally.].

.

[PCBONEZ]

Re: Intel D865GLC, higher uF

The heat load is why you wouldn't use (say) one 3300uf in place of two 1500uf.
The 3300uF would overheat.

[PCBONEZ]

Re: Intel D865GLC, higher uF

And it's the reason a designer would use two 1500uf instead on one 3300uf.

[Naturally I'm assuming the ripple and ESR of the two 1500's = the 3300.]