What cap mfr. to use?

"low ESR" is not good enough for a purchasing spec. By the standards of just a few decades ago, all of today's electrolytics are low ESR. But whether anything meets your requirement is driven by the requirement, not bullet points in the capacitor maker's glossies.

It's easy to find electrolytics of same voltage and capacitance which have ESR specs different by a factor of 10 or more. It shouldn't be too surprising that package and size have a lot to do with ESR. If your requirements are not met by any commercially available capacitors, you might think about making your design more realistic :-).

Digikey etc. let you sort by ESR.

It is also possible to choose capacitors which are too low in ESR and lead to instability with low-dropout regulators or other circuits.

Tim.

The only one of those I haven't heard of is United, but that says nothing about the quality, I'm just not familiar. The others are all familiar and reputable brands, I wouldn't hesitate to use any of them.

Just go to digikey and find the values you need, don't worry too much about the brand or about having them all the same brand. Pay attention to the physical size and package style as it varies and often matters with compact modern equipment. They will be listed by temperature and by lifespan at that temperature. I have always used standard 105c "low ESR" capacitors as replacements in switching power supplies and never had an issue.

And a lot of, maybe most, switchers are stabilized by the filter cap esr.

John

How critically are you trying to match the values?

Please remember that 'lytics are manufactured and sold with a rather wide tolerance range. For example, the Panasonic FM and FC series have a +/- 20% tolerance specification, and the nominal values they list (in the Digi-Key catalog at least) are around 20-30% apart... so there's some amount of overlap between the values you'd expect to get.

Older caps were often sold with -20% +80% tolerance, which meant that you could end up getting quite a lot more capacitance than you had "paid for".

In most cases, these days, I'd just look at the space I have available to fit the cap, and then pick the "equal or next largest capacitance, equal or higher working voltage" to what the BOM calls for. Unless there was a particular need for a specific value (e.g. for timing purpose) I wouldn't sweat about things like "the BOM calls for a 220 uF, and all they have in that size is a 330 uF."

And, if there were timing issues involved, I don't think I'd be using a 'lytic at all in that application!

I've heard good things about Nichicon, and have used them in some repair/retrofit projects. Don't have anywhere near enough information about long-term performance to be of help with your question, though.

For what it's worth, I've seen several of the PC-motherboard manufacturers touting their use of solid-electrolyte capacitors for the CPU VRM... and this is a high-current, high-ripple, low-ESR- is-very-important application. One manufacturer was citing a "50,000 hour" lifetime figure on the motherboard carton (although I think this assumes very good cooling of the board).

Speaking as someone who spent the last 10 years of his career as a senior engineer in a department repairing broadcast equipment, we found B C components (formerly Philips, formerly Mullard) to be among the most reliable and some we used to get from Farnell/Newark, with a sort of diamond design on them (I think possibly manufactured by multicomp), as the worst.

In article , Dave Platt writes

Are you sure there isn't a microdot in the small print with the words "electrolytic capacitors excepted" printed on it?

"lifetime" is a word of dubious interpretation. Does it mean that it won't break in that time, that it can be repaired in that time if it does break, or something else?

If they claimed an MTBF of 50,000 hours, that would be different.

Sylvia.

AFAIUI, MTBF typically applies only to the flat part at the bottom of teh bathtub curve. They conveniently hack off the infantile failures at the left and the increasing failures as the useful life expires on the right. IOW, a product can have a much higher MTFB than the time it takes to wear out.

Also, 50,000 hours 24/7 is only 5.7 years, which is more-or-less what you'd expect out of a motherboard.

And if you run a 105C cap at 40C you'll get ~ 85 times the datasheet lifetime. So a 2,000 hr 105C cap would last over 19 years. Watch the ripple current of course !

Interesting point there. One decoupling cap on a mobo of mine near the graphics card slot was visibly bulged whereas others weren't. I imagine it was hot air being blown onto it by the GPU fan.

Graham

In article , Spehro Pefhany writes

Not from electrolytic caps, which was kind of the point. Even Panasonic specify only 10k hours for their top of the range electros.

I guess that would be a rather serious problem if the temperature inside your computer case averaged 105°C = 221°F.

In article , Spehro Pefhany writes

Seen a modern mainboard recently? Voltage regulation MOSFETs with heatsinks on, some with their own extractor fans? Heat coupled to the capacitors by a few millimetres of thick copper PCB trace for excellent heat conduction?

But isn't that at "high" temperatures, or some other extreme operating condition?

Certainly.

Have you put a thermocouple on the capacitors?

AT RATED TEMP and RATED RIPPLE CURRENT. Rarely found in combination except by bad design, and it does happen sometimes.

You can get far better by de-rating. Rule of thumb is to reduce the temp from rated value by 10C and double the lifetime etc etc.

Graham

Yes.

Grham

Point 1: You can ventilate your case and CPU far better than 'normal' methods.

Point 2: leave the side panels off like I tend to do !

Graham

You need a themocouple and a suitable meter for that ! ;~)

Graham

Shit, I'd generally be happy with 2000 hours MTBF.