6.3 volt 1800uf caps

[hydafl]

what would be safe upgrade for these caps in a motherboard, or simply use these ratings? 6.3 volts seems a bare minimum

[budm]

it is not just about Voltage rating, you have to consider the ESR rating, ripple rating, etc... too.

[ChaosLegionnaire]

yes, u also have to see where on the motherboard it is. if it is near the cpu vrm, near the ram slots, chipsets, pci, agp, usb or fan headers. depending on where it is, certain specs like esr matter more or capacitance matters more or a combination of several factors.

as for the voltage rating being a bare minimum, there are only three voltages used by computer motherboards, 3.3v, 5v and 12v. so 6.3v can be used for the 3.3v and 5v rail. no issues. even on the 5v rail, there is still a 1.3v buffer which is sufficient. never an issue. for the 12v rail, 16v caps have to be used.

what motherboard is it anyway?

[hydafl]

It's an old Dell with Athlon processor. very old I suppose. basically back up and practice board. came out of an optiplex refurb.
just hate putting anything like this in the trash.

[silentbogo]

Just replace with any low-ESR equivalent. You can find cheap Nichicon or Chemi-con caps that fit the bill for pennies.
Also, those old sAM2 and LGA775 machines are less sensitive to ripple on VRM, which for all intents and purposes translates to "use whatever you have with a nearest equivalent spec". For a >10y.o. machine a "slightly off-spec" cheap chinese capacitor is better than dried-out or blown old capacitor. Since it's on the CPU VRM, it's probably a good idea to replace all caps at once, and try to avoid mismatching caps.

[momaka]

Originally Posted by ChaosLegionnaire

as for the voltage rating being a bare minimum, there are only three voltages used by computer motherboards, 3.3v, 5v and 12v. so 6.3v can be used for the 3.3v and 5v rail. no issues.

Yes.

Not only that, but the motherboard will usually step down these voltage further for use by the CPU, NB, RAM, and etc. All of these are usually 3.3V or less. So for these, you actually CAN lower the voltage rating of the caps, depending on what they connect to. For example, old Pentium II CPUs typically require 2 to 2.8V for their V_core. Thus, the lowest cap voltage you can use there is 4V (although 2.5V may work in a system with a PII that has a 2V core, it is possible that someone could downgrade the CPU to one that uses 2.8V core, and that would yield unsafe operation.) On newer CPUs like Pentium 3 and Athlon, V_Core is typically 1.8V or less. So on those motherboards, you easily can get away with 2.5V caps - at least for the ones that filter power to the CPU.

RAM is different story. SDRAM takes 3.3V, so 6.3V caps are needed. DDR takes 2.5-2.7V, so minimum 4V-rated caps are required. And DDR2 and newer take 2.1V or less, so 2.5V caps should be enough.

Of course, if you are not sure what each cap is filtering, then stick to 6.3V rated caps and you will be fine. You could also use 10V and 16V rated caps, but they will be physically larger and may not possibly fit because of that.

Originally Posted by silentbogo

For a >10y.o. machine a "slightly off-spec" cheap chinese capacitor is better than dried-out or blown old capacitor.

Very true.
However, I still don't encourage anyone to use cheap Chinese caps, because even if they work fine, there's no telling for how long. After all, it would be a waste of time and money if the recapped motherboard failed again with bad caps due to the cheap caps not lasting. So saving a few pennies is usually not worth it in the long run, at least to me.

On the other hand, I do encourage re-using caps from other old/non-working hardware, so long as the caps are OK (both in terms of specs, as well as quality brand... or at least a less problematic brand/series.)

Originally Posted by silentbogo

Since it's on the CPU VRM, it's probably a good idea to replace all caps at once, and try to avoid mismatching caps.

No, you can mismatch caps any way you like without adverse effects, as long as you are doing that to caps that are connected together in parallel. That is, let's say you have two 1800 uF caps in series, for example. The total capacitance of the circuit they filter is 1800 + 1800 = 3600 uF. But if you don't have exactly two 1800 uF caps, you could use something like, say a 2200 uF cap and a 1500 uF cap for a total capacitance of 3700 uF. Or if the circuit expects low ESR caps, but you really are in a pinch, you could use a 2200 uF cap with "worse" specs and parallel that with a 820 uF polymer with very "good" specs. The total capacitance will then be about 3000 uF - a little low, but most of the time that is not too far out of spec. Main point is that with the polymer in there, the two caps will still show low equivalent ESR, despite one being much higher than the other.

As such, I often do "hybrid" recaps on my motherboards, where I mix wet-electorlytics with polymers for the CPU VRM (usually for older boards that have fewer cap spots, so that I can put a few higher-capacity caps with not-the-best ESR and a few very low ESR polymers to take on the majority of the filtering.) This actually give the CPU VRM the best of both worlds: higher capacity for more stable voltage and lower voltage ripple, along with low enough ESR to suppress all transients.