Anyone who works with older electronic equipment will before long learn to spot Rifa capacitors, a distinctive yellow-translucent component often used in mains filters, that is notorious for failures. It’s commonly thought to be due to their absorbing water, but based upon [Jerry Walker]’s long experience, he’s not so sure about that. Thus he’s taken a large stock of the parts and subjected them to tests in order to get to the bottom of the Rifa question once and for all.
What he was able to gather both from the parts he removed from older equipment and by applying AC and DC voltages to test capacitors, was that those which had been used in DC applications had a much lower likelihood of exhibiting precursors to failure, and also a much longer time before failure when connected to AC mains.
Indeed, it’s only at the end of the video that he reveals one of the parts in front of him is an ex-DC part that’s been hooked up to the mains all the time without blowing up. It’s likely then that these capacitors didn’t perform tot heir spec only when used in AC applications. He still recommends replacing them wherever they are found and we’d completely agree with him, but it’s fascinating to have some light shed on these notorious parts.
Thanks [Ian Somers] for the tip.
Header: Cjp24, CC BY-SA 4.0
25 thoughts on “Why Do Rifa Capacitors Fail?”
Just a note: this failure is not limited to Rifa. Wima do exactly the same thing – a bang, brown goo and a nasty stink that hangs around for a week.
While rupture is not limited to RIFA capacitors, the brown goo you mention is a different failure mode in a different type of capacitor.
Ian, those old Wima capacitors with the brown easily melting cover I remember very well from the sixties, the time of electron tubes working on 400V or so. Repair was easy, if exchanging tubes did not work, exchanging Wima capacitors always did the trick.
“tot heir” :)
Swedish quality wasn’t what it used to be…
I think that there is a classification error here, in that Rifa make (or made: https://en.wikipedia.org/wiki/RIFA_AB ) many different styles of capacitors and translucent prismatic capacitors are made by many other manufacturers.
The problem, it seems, from literally seconds of Googling, is with the RIFA X2 type capacitors.
In the retro computing world they’re just referred to as RIFA caps for ease, basically anyone will know what is meant by them because they’re absolutely notorious.
Don’t fear the rifa
Sorry, couldn’t resist!
Name check right!
“Hey, I don’t wanna mess with no Rifa addicts ok”
“Run back home to yo momma boy!”
What did he actually prove? That moisture apparently isn’t the problem. His tests didn’t prove anything more about WHY they fail and his assumption that anyone would make even without those tests is that dielectric breakdown causes a short. His conclusion:
“It’s not just a case of being exposed to moisture that causes this. My personal opinion is that it’s breakdown of the plasticizers in the the dielectric separator that causes it but it doesn’t really matter the point was change them. They will fail and the fact that they fail in other applications is irrelevant. They will fail if they’re across the main.”
It kind of makes sense, though. The capacitor should (probably) fail all the same with DC or AC if the problem was strictly moisture. I’m not sure if it’s possible to take apart one that’s been used in AC and hasn’t failed yet, to inspect the dielectric, but that would be pretty conclusive IMO.
The Rifa’s caps are oddball- paper dielectric with clear epoxy housing. Because they are usually doing mains EMI filtering, changing to another make meant new EMC tests so German manufacturers (i.e. Schaffner) stuck with them.
They are also from an era 80’s before mains transients were categorized and not really up to the task.
I had new old stock in bins and the housing cracks and they seem to bulge out. I say it’s not moisture, but deforming plastics and corona etching of the metallization, like what happens with polypropylene X-caps.
They are still for sale today (Vishay) PME271 and no reply to an email asking WTF are you still selling these for lol. They do burn up and make a mess.
“…changing to another make meant new EMC tests…”
There are no harmonized standards in the LVD or EMC or RED directives that specifically require this; the same for 47CFR. It can be a requirement by the NRTL or Notified Body or whatever – mostly because of a poorly written report by the compliance engineer.
When the EMC and safety reports are not done internally, it is a common scam by test labs to not list alternate components in the construction files because it increases repeat business that requires no engineering time, but is charged as such.
If your company is writing poor compliance reports that do not include alternate components in the CDF, perhaps they need to find a new safety/compliance engineer. Just sayin…
Young engineer asking – how do you recommend prescribing alternate components in test reports without needing to test those alternates?
Non-answer follows. This is not a guide to regulatory design requirements or product conformity submittals. Your mileage may vary. Not for use in an open container. Do not use as a sexual aid.
There is no prescribed test series or universally accepted methods for all alternate components that are built into other stuff. For X and Y caps, I have had an informal system, based on my professional standing and relationship with agency assessment engineers, to run bode plots on the mains power converter for EMC qualification; that and the basic RLC sweeps across the intended bandwidth. In practice, I have done pre-comp emissions and immunity testing for various combinations of EMC-critical components.
Internally, all companies that I’ve worked at, and most clients, have a formal ISO process for qualifying components. At my last full-time employer, I only allowed three manufacturers of X and Y caps. My internal safety and performance tests exceeded requirements for the stuff found in the IEC60384 series.
Theoretically, the test houses are supposed to accept a CB report from a company accredited per IEC QC 001005 or the other similar nonsense such as the ISO 9k stuff.
You don’t know all the alternate H/W components that could be used in a product at EMC lab testing time. Compliance testing is considered a one-time event by management due to the expense and panic to get it to market, I find.
Listing and testing a few alternates makes it very complicated and expensive (exponential) due to the parts’ interactions. A mains EMI filter the X-cap, CM choke, Y-caps all interact and resonate. There is no data or curves for these Rifa part’s impedance, no quality system is going to save you. There is a risk another make/model of cap/inductor can cause the product to fail EMC. I informally test in-house but what i.e. Schaffner does who knows.
Indeed, EMC also requires your firmware version – I had one product that passed EMC with one F/W version and then it failed with the newest – because the (new) constant LCD screen refreshes made a lot (too much) EMI.
So strictly speaking, product changes as far as H/W and F/W can affect EMC. This is why I mentioned the Rifa lemon capacitor line is still for sale?!… despite decades of smoke. One has to ask what’s the dedication to using these parts. I assumed it’s EMC and associated voodoo with the part’s lack of EMC data or antique metallized paper dielectric construction.
how old are these? it’s notorious that industry espionage is not always successful so that wrong information leads to wrong mixtures hence premature failure….
When used in a DC application, is the voltage regulated before doing so or is it just rectified mains AC?
These capacitors are used 1980s lasted power supplied the cases crack and they smoke they are paper so they fail like the transparent wima capacitors that look the same style fail as well
there isn’t fault with the capacitor; it’s mostly an application learning experience. Film caps are intended for delicate signal processing and filtering. They are not suitable for exposure the horrible line voltage environment. just have a look at the ‘recent’ safety standards of X and Y capacitors, as they’re known; and they happen to be ceramic!
I think that the capacitors in question _are_ rated as X2, though:
Is that the type that is under discussion here?
interesting that the datasheet says ‘not recommended for new designs!’ so, it ceased production for some reason- could be for safety, or a dozen other things. I’m just superstitious about not using film caps for safety or connection to undefined environment-some connection outside the box; since they don’t have a lot of Voltage safety margin and don’t age well. I was testing some ceramic 0805 chips a few years ago and those 50V MLCC parts would withstand 600VDC from a hipot leakage test-set!
I inherited a top-condition Revox A77 open reel deck from a late uncle, a few years ago. While giving the deck the once over (inspection, cleaning, lube, etc) and exercising the transport, there was a bang and some magic smoke escaped.
Turns out they used such capacitors as AC voltage dividers for the reel motors, and one had apparently shorted. Replaced all such with modern equivalents… no more problems and reel motor tension back to spec.
I’m frankly surprised (and grateful) that more components don’t fail with age like that.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)