Would you trust these caps?

[John_BS]

OK, so we now have some interesing results from Waveydipole's caps.



He sent me two samples of 470pF and 4n7 respectively. Both withstood 2kV with no issues.

• The 470pF samples seem to be temperature-stable around room temperature, and I've not tested them further.
• The 4n7 were all over the place: if you handled them in a cold workshop they would move from c. 4nF to 5.6nF

I remembered reading somewhere that hi-K dielectrics also exhibited a strong voltage coefficient, so a test jig was rigged up comprising the DC bias source ( 0 to 2kV) fed to "dodgy cap" via a couple of 10M resistors. The capacitance of "dodgy" was measured via a couple of series 1kV 4n7 polypropylene caps in order to protect my precious bridge! So, with no volts applied you expect (and get) about 1n6 from the series string.

The capacitance of our hi-K cap is then calculated using the standard forumula. The attached graph says it all.


John

[MrBungle]

Interesting results. Thanks for sharing.

[WaveyDipole]

John, thanks for taking the time and trouble to perform these tests and posting the results. I am glad your bridge survived the ordeal! At only 250v the capacitance is down by some 70%, and by 1kv down over 90%! Quite shocking.

I came across this article, which, while talking about SMD components, shows how the characteristics of parts with the same dielectric can vary significantly as component size shrinks. His observations were recorded at much lower voltages:
https://www.maximintegrated.com/en/a...ex.mvp/id/5527

Regarding these eBay parts that I purchased, they look pretty much identical, like they could have come from the same production line and were made with the same materials, yet your temperature observations alone (which I have now also carried out on my DER LCR meter and can confirm) would seem to indicate that they behave quite differently. Whether this suggests that they are made with different materials, or whether as per the article its down to the smaller size relative to capacitance, I'm not sure. It doesn't seem to make sense that something having 10x the capacitance, with the same voltage rating and using the same materials could be manufactured in a smaller package without something being compromised. Still, at least they did seem to withstand 2kv across them, even if, by that point, the capacitance was a small fraction of the nominal rating.

The conclusion of the article makes a good point about checking datasheets, which, of course is not usually possible with components purchased from eBay as sellers do not usually provide links to them and rarely, if ever, post anything but the most basic information. If the component is branded, it may be possible to find a datasheet elsewhere, but if it is unbranded, there is simply no way to be sure of anything.

Looking at your graph, if I was expecting something like the nominal capacitance I probably wouldn't use these for anything much above 100v, or at least if I did, then I would need to de-rate the capacitance accordingly.

I guess this goes to show that just because things come in identical packages, it does not mean that they are made the same, and for anything critical, check the datasheet and use bona-fide sources.

[John_BS]

For completeness, here's the plot of one of the 470p capacitors: clearly a very different dielectric.



John

[Al (astral highway)]

Originally Posted by John_BS

OK, so we now have some interesing results from Waveydipole's caps.

It’s the graph of the natural log (e) function!

Pops up in many places, didn’t expect it here though! I love it!