• Please review our updated Terms and Rules here

Physical size of electrolytic capacitors...

S

Stone

10k Member
Joined
Nov 3, 2009
Messages
12,814
Location
South Jersey, USA
I'm going to replace some caps on some motherboards that currently don't work. The first round is going to be a group of 6.3v 2200uf. These are 10x20 physically. I see that they are available in two other physical sizes as well; 8x20 and 10x25. I was wondering what difference, if any, the physical size makes in this application, i.e., motherboard.
 
None at all. The specs are the ultimate arbiter. In old equipment, one of the more interesting problems is that modern electrolytics are physically smaller. some people have resorted to scooping out the guts of the old unit and soldering the new one inside to preserve the looks.

I replaced the motor start/run electrolytics in our crica-1992 heat pump--the new ones were considerably smaller than the old. I had to fashion new brackets for them.

If at all possible, consider the solid electrolyte capacitors. They appear to exhibit a significantly longer life than the older wet electrolytics.
 
I don't see anything in that range (6.3v 2200uf) at all. Nothing even close.
Click to expand...

You probably don't actually need either 2200uf or 6.3V. A value like that is most often going to be found on the output side of the CPU VRM, in which case it's ESR that matters, not so much capacitance, and the voltage is going to be whatever the maximum voltage of the CPU is... which is <2V for anything reasonably modern.

The large caps were used because physically larger caps have inherently lower ESR than smaller ones (assuming everything else equal, like internal construction, particular electrolyte formula, etc.). Solid polymer caps have much lower ESR than electrolytics, so you can get away with smaller caps in those applications. On CPU VRMs, I usually replace the 2200uf 6.3V ones with 1000uf or 1200uf polymers, either 2.5V or 4V rated. If there are a lot of caps, sometimes I'll even go down to 820uf.

But don't take that as a recommendation for your particular boards... it's not something to just blindly go changing without knowing how the VRM is set up and exactly how the caps are used in the circuit, and since I don't know what boards you have, I can only speak in generalizations. When in doubt, just stick with electrolytics in the original values, making sure to get low ESR types. For use on motherboards, I like Panasonic FM and FR series, and Nichicon HM... whichever's cheapest at Mouser, it's been Panasonic FR as of late.
 
Solid-polymer electrolytics are definitely the high-priced spread and becoming the standard for motherboards--and it's a good thing. Unfortunately, and a sign of the times, the widest range of values and ratings is available primarily in SMT packages.

In general, the solid polymer caps have much lower ESR. My caps of choice are Nichicons. Here's the datasheet for the leaded type LG solid caps.

Another very good brand is Nippon Chemi-Con (NCC). Here's their datasheet.
 
Last edited:
Old Thrashbarg said:
But don't take that as a recommendation for your particular boards... it's not something to just blindly go changing without knowing how the VRM is set up and exactly how the caps are used in the circuit, and since I don't know what boards you have, I can only speak in generalizations.
Click to expand...
It's an ECS, K7SEM Rev. 1.0. I've got a couple of them and one is showing some swollen caps although neither of them works. So I figured I'd start with the obvious. While I'm at it I'll replace the same caps on both boards and maybe I'll get lucky.
 
Ah, I've done a few of those, and their close cousins. Before replacing any capacitors, I'd recommend first checking all the MOSFETs around the CPU... those old ECS boards had a nasty tendency to blow 'em out if they were run for very long with the bad caps.

There are a few different versions of the K7SEM... be sure to double-check all these values with what you actually have on the boards (since I'm just going by memory here), but on a Rev 1.0 you should probably have:

-Eight 2200uf 6.3V, in pairs next to the CPU socket. You can replace those with 820uf 2.5v polymers if you like... I did so one time since I had some polys laying around that I'd salvaged from another dead board, but it's probably not worth the additional cost over regular electrolytics if you actually have to buy the capacitors. I'd personally just stick with Nichicon HM in the original value.

-Four 3300uf 6.3V, in between the rear ports and the CPU MOSFETs. Stick with the original values on those. Panasonic FM seems to be the best deal for those.

-A few 1000uf, either 6.3V or 10v. Mostly around the memory slots, I think maybe a couple near the ATX power connector too? Just go with 1000uf 6.3v on those, Nichicon HM again.

-Two 1500uf 6.3V by the AGP slot. Nichicon HN for a better physical fit, or if the height won't interfere with any of the AGP cards you may want to use, the 20mm tall Panasonic FR are quite a bit cheaper.

-There are also a few smaller ones... I think it's 470uf 16V for the USB ports and such, I wouldn't worry too much about those, but if you're already going to be ordering parts it might not hurt to get 'em anyway. It also wouldn't be a bad idea to grab a few 100uf 16v caps... sometimes the little squatty ones on those boards go bad, so it wouldn't be a bad idea to have some spares on hand in case you still run into any weird gremlins after replacing all the larger caps. Chemicon KMG is fine for those, and quite cheap.
 
How do I check a MOSFET? It's a transistor, right?

I've also got a couple of K6XV3+/66 boards and a couple of K7S5A boards, too. All of these boards seemed to die a slow death, over a period of many, many months. At first they failed to boot occasionally. Then it became more often. Then it became a chore to get them to boot at all, meaning many, many attempts before a successful boot. And finally... nothing ever again. :) Does this scenerio sound anything like cap failure to you. If so I might have stumbled into a lot of work I hadn't planned on this winter. :) I saved these particular boards 'cause I liked them and if they're fixable by recapping I might just give it a go. What do you think?
 
How do I check a MOSFET? It's a transistor, right?
Click to expand...

Well, on those ECS boards, you can often tell visually when they're bad, as they tend to explode. But to be sure, you should probably test 'em anyway. For best results, use a multimeter with a diode test function, and follow this procedure. Note that not all MOSFETs have the same lead arragement... look up the datasheet for your particular parts to verify which lead is which. Alternately, when just looking for shorted ones, like what usually happens on those ECS boards, you can also generally get away with just checking resistances in both directions between the legs, making sure they're all high/infinite.

I'm not familiar with the K6XV3+, but they K7S5A is basically the full ATX version of the K7SEM... a bit different in the particular number and values of caps, but the general design is pretty much the same. And yes, those symptoms are consistent with bad caps, but again, check to make sure the MOSFETs didn't get fried before investing much effort into 'em.
 
Funny thing about the MOSFETs on these boards... none has the Drain leg connected. In fact they've all (the Drain legs) been removed, except for one MOSFET on the K7S5A. And they all look normal with no visible signs of anything.
 
Yeah, the drain is connected to the mounting tab on most MOSFETs, even the through-hole ones with three legs. SMD MOSFETs rarely have the middle leg, since it's kinda redundant to put it on there when the mounting tab is soldered to the board anyway.
 
Nice price, but unfortunately those are just general purpose caps... not low ESR types. That makes them useless for modern motherboards and switching power supplies.
 
Old Thrashbarg said:
Nice price, but unfortunately those are just general purpose caps... not low ESR types. That makes them useless for modern motherboards and switching power supplies.
Click to expand...

You're doing better than I am if you can determine the Nichicon part number from that photo. From what I can tell, most late 90s motherboards didn't use anything better--to the contrary, much worse, "Rulycon" caps and all. For patching up some not-quite-yet-vintage boards, they'd be fine. They'd probably be fine for vintage stuff also.

It's curious that of all of the LCD displays that I've re-capped with Nichicon 105C caps, none has ever failed--even those that died less than 6 months after purchase. I tend to think that by and large, modern stuff is stuffed with garbage components.
 
You're doing better than I am if you can determine the Nichicon part number from that photo.
Click to expand...

I do a lot of electronics repair, which mostly consists of capacitor replacements these days, so after a few years of repetition I've gradually built up a pretty close familiarity with all the major companies' current and past product lineups.

Assuming the ones they're selling match the ones in the picture, the caps on Goldmine are the VX series. Not the current VX axial series, but the old one that was discontinued years ago... late '80s or early '90s I believe. It was replaced by the VR series, which are slightly better specced and a bit smaller, but are still pretty low in the product hierarchy.

And no, they are nowhere close to what was used in late '90s motherboards. Starting with PII/PIII-class (including some late Socket 7) boards, they moved to using VRM designs that needed low ESR caps. Of course the parts weren't up to the specs of the stuff used on newer motherboards, but they would still be several grades above something like the VX series. Even the cheap brand-x and counterfeit caps generally had good specs when they were new, it's just that they didn't last very long. Same goes for PC PSUs, LCD monitors, and whatnot... you don't need super low ESR (and in some cases it can actually cause problems, depending on the design of the filtering circuit), but you still need something far better than the VX series. Hell, even setting aside ESR, those VX aren't even 105C rated.

Those things would be fine for vintage equipment (assuming they're still good after sitting around for 20+ years), and linear power supplies and such as long as there's adequate ventilation to keep 'em cool, but they have no place in most modern equipment.
 
Last edited:
Old Thrashbarg said:
Those things would be fine for vintage equipment (assuming they're still good after sitting around for 20+ years), and linear power supplies and such as long as there's adequate ventilation to keep 'em cool, but they have no place in most modern equipment.
Click to expand...

Well, that would certainly explain the price... Thanks for the info.
 
You must log in or register to reply here.
Back
Top