Recently I went to order a set of these 100uF electrolytic capacitors with 10% tolerance, needed for impedance balance for an audio output. When I noticed they were out of stock, I decided to order a number of the 20% version, thinking I could just measure them and sort out the "beyond 10%" tolerance caps. To my surprise, when measured with my Fluke 87V (DC), all caps measured just about the same at around 103.4uF (I expected them to vary as resistors would). This was more-or-less the same value I measured on most of the 10% caps. The datasheet specs the capacitance tolerance at 120Hz. I've also read that DC measured capacitance is often around 10% higher than the AC measured capacitance.

So my questions: Are actual electrolytic capacitance values generally relatively consistent across a wide frequency range? For example, can I pretty much assume that if the capacitance is 102uF at 20Hz, that it will be around the same at 20kHz? (Related to this, should the datasheet tolerance spec be read as the deviation of capacitance from the specified capacitance, independent of frequency, or does it also account for frequency dependence)?

Secondly, does the consistency of the measured DC capacitance across a number of different capacitors give me any indication of the consistency of actual capacitance values in AC? In other words, is it likely/probable that while all the caps measure the same at DC, that they could deviate significantly from each other at, say, 1kHz, with one cap measuring 85uF and another 105uF? Or can I assume that if the DC readings are all pretty much the same, the AC readings should be also?

  • \$\begingroup\$ The tolerances also relate to how much the value is expected/allowed to change after x hours at y temps and remain "within spec". So being matched within a few % now doesnt mean much in the long run. (Although yes, its likely that caps from the same batch will age similarly) \$\endgroup\$ – Wesley Lee May 21 '18 at 6:41

Electrolytic capacitors do have a mechanism whereby some DC measurements show different values. Once you are in the audio band, it's likely to be constant from 20Hz to 20kHz.

The question is, what method does your Fluke use to measure their capacitance? Does it apply 1kHz AC, in which case the measurement will be true for the audio band as well. Or does it apply a DC current, and measure how quickly the voltage changes over a period of seconds? This latter method is prone to errors from leakage current (big anyway with electrolytics, especially horrible with new ones), and from charge absorption (have you ever watched the voltage reading on a big electrolytic that's been charged and quickly discharged grow again, over the course of minutes?)

As the DC leakage depends on the charge history of the capacitor, there's no reason that similar DC measurements should correlate to similar AC measurements. So it's important to find out what measurement your meter is making.

Why 10% and 20% tolerance caps? The purpose in life for an electrolytic is storing energy in power supplies, not much else. They don't need a good tolerance, so they're not built to a good tolerance. The form of construction doesn't lend itself to good reproducible dimensions anyway. But with automated assembly on high quality machines, it's not surprising to find several capacitors made in the same batch to match to within percent.

  • \$\begingroup\$ Thanks for that info. I read through the Fluke manual before posting this and couldn't find any info on how it measures capacitance, and I haven't been able to find any results (yet) doing a web search. Might have to put a scope to it to find out. \$\endgroup\$ – User7251 May 21 '18 at 8:21

My take on this is that due to the extremely tight tolerances of so many type of electronic parts, and the precision in which they are made, and the ultra-clean rooms they must be made in, electrolytic capacitors are no longer a 'in the ballpark' category.

There will always be different readings with AC vs. DC testing, and changes between 120 HZ and a 1 KHZ sine-wave test, regardless of true accuracy. The AC and DC readings do not have to match, as long as the readings are within the posted tolerance.

Their better accuracy is a side effect, and a trend in general, to go for perfection as best as possible. Matched transistor pairs used to be a special order. Now they match very close right out of the box. Resistors with 0.1% tolerance are common and you can by 200 for $5 USD. They cost much more but you can buy 0.025% tolerance resistors from Caddock.

In their effort to make better super-capacitors they have found ways to make more accurate electrolytic capacitors. Repeated better accuracy is not an anomaly, it is a way of life for todays manufacturing plants.

It is marketing that is reluctant to add tighter specs on the capacitors label, but in time they will learn to trust and offer 2% tolerance, which can be had by special order. In 10 years they may be off-the-shelf common.

  • \$\begingroup\$ Fair enough, but there must be SOME criteria that Nichicon and other manufacturers use to rate some caps at 10% and others at 20%, especially if they can make more money on the higher tolerance. I was under the impression (possibly naively) that this was done by simply measuring them by some automated system, then sorting them based on the measurement results, and not necessarily by any design criteria, but I'm sure this varies by product. \$\endgroup\$ – User7251 May 21 '18 at 8:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.