New Rubycon capacitor almost -20% under nominal, use anyway (in a 16702B PSU)?

[hp3310a]

Not a repair question but I don't know where else to put this:

I'm was planning to exchange the main caps in the PSU of my Agilent 16702B logic analyzer. I ordered new Rubycon HXK (220µF/450V)which are 5mm smaller in diameter than the originals (20x50mm instead of 25x50mm). That should improve air flow in the pretty cramped switch mode PSU considerably (attaching a picture with the top cover removed and one of the main PCB).

The new caps arrived and to my surprise, the new caps have a capacitance of -18% nominal. That's almost out of spec and also 8% lower than the existing ones.

Question is, should I use them anyway? Or order new ones somewhere else (these are from Digikey) and hope they will be better?

[nonius_]

Just had a look at the datasheet of those capacitors and the capacity tolerance is +/- 20%, so -18% is within tolerance. (and that's also assuming that your measurement device tested at 120Hz....)

For the primary buffer-electrolytics in a SMPS the slightly lower capacitance  is not an issue at all.

Most circuits aren't very critical or demanding w.r.t. tight tolerances of capacity. Capacity is very rarely a problem with electrolytics, you can often get away with much less or more capacitance in most applications (time-sensitive applications excepted).

should I use them anyway?

Myself, I'd have absolutely no qualms whatsover using them.

Were the original capacitors bad?  Only problem I can think of is ESR, but for THOSE capacitors (primary/high voltage section buffer electrolytics) ESR is rarely a problem; it's usually the secondary side, where currents are much larger, that capacitors go bad because of overheating/high ESR.

I'm was planning to exchange the main caps in the PSU

You mention it's not a repair; then why are you replacing those capacitors?

[hp3310a]

Thanks for the response!

The original fans are intolerably loud (of the 5 fans in the unit, these are the loudest because smallest), and they need to be replaced by some with about 20% less air-volume. I'm exchanging the capacitors because they improve air-flow.

I measured at 120Hz and the ESR of the new capacitors is about 1/3 lower than the originals.

[floobydust]

Don't forget to replace the Rifa caps!

I order primary electrolytic caps +20% in order to get the original value 
I think they know exactly how to manufacture electrolytic caps to -19% nowadays. The +80%/-20% tolerance spec. dates back to the 1940's I think.

[Gyro]

Don't forget to replace the Rifa caps!
...

Yes, that Rifa MP Y cap is starting to show cracking. The only visible X cap looks like an MKP but they have a habit of hiding.

[nonius_]

I should've mentioned that RIFA-capacitor! Glad others caught it....

Keep in mind that there may also be a paper capacitor in the power-entry/fuseholder/netfilter combination, if present in your machine. In my HP54504a it recently spilled its guts, leaving a nice mark on the wall behind it as it ejaculated its juices through the tiny gap of the powerswitch. Took a while for the stench to go away....

[ME]

It may cut the noise down if you declagged the fan blades , They look filthy!.

[hp3310a]

Don't forget to replace the Rifa caps!

Those were the first ones to go and the primary reason for taking the PSU apart. There are 3 RIFA Y2 caps, one with 4.7nF and two with 15nF, they were all nicely cracked. The X2 is a WIMA I think (the big blue one to the left), can't read the markings.

I think they know exactly how to manufacture electrolytic caps to -19% nowadays.

Yes, I find that highly suspicious, too. That can't be a coincidence.

Keep in mind that there may also be a paper capacitor in the power-entry/fuseholder/netfilter combination

I was planning to, it is not a Schaeffer though, I think.

It may cut the noise down if you declagged the fan blades , They look filthy!.

They are! Those will be replaced by new Sanyo 9S silent fans. The noise level is intolerable and the bearings are shot.

[mariush]

Hopefully when you say ordered , you mean you ordered it from a proper distributor of electronic components like Digikey, Newark/Farnell, Mouser etc and not Amazon or eBay ... those can be fakes (in the sense that they could use lower capacitance models and put a new sleeve on them).

Some multimeters have problems measuring high voltage capacitors, so a value lower by 18% could be fine. You also need to account for the meter's accuracy itself, which could be up to +/- 3-5%

[ondraN]

Try connecting the capacitors to their rated voltage for a while, then discharge them and measure them. If electrolytic capacitors are stored for a long time, they usually show less capacitance. That is my own observation.