• Please review our updated Terms and Rules here

Argghhhhhhh! IBM 5322 went Snap Crackle and Pop!

When I researched for a replacement for one of those Rifa paper dielectric caps that blew in my TRS-80 Model 12, I decided to replace all of the paper dielectric caps I encounter with polyethylene or polypropylene film safety-rated caps. I read that the paper dielectric is prone to absorbing moisture if the case is not perfectly sealed, so I hypothesize that the failure mode of these Rifa caps is for the case plastic to crack due to aging, allowing moisture to seep in, which then results in dielectric breakdown when they're energized. Followed by lots of smoke and the smell of burning aluminized paper. I haven't replaced those caps on my newly-acquired 5322 yet, but that's on my short list.

If Wikipedia can be trusted, then poly film caps are much less prone to moisture absorption, even if not hermetically sealed. Thus, I decided to search for poly film caps with similar ratings as replacements.

If I remember, I'll try later today to dig up the part numbers of the caps I chose to replace the Rifa caps that I found in my TRS-80 power supplies. Identical new (?) Rifa caps are still available, but I don't trust them. The part numbers I picked out might not be perfect matches for what you need, but they should at least help you narrow down your search through the tens of thousands of caps that Digi-Key carries!

My replacements had different lead spacings than the originals, but the Astec supplies in my TRS-80 computers had multiple holes for those caps to accept caps with different lead spacings. They apparently planned ahead to be able to source different caps in production.

BTW, I just now found this thread, and the link you posted with information about the power-on diagnostics is helpful. Thanks!
 
If there is a ~ (tilde) symbol to the right of the 'V', then that means AC.

The capacitor pictured [here] has two AC voltage ratings. My reading of the specifications under the photo is that each voltage rating is applicable to different test standards, i.e. the designer of a piece of equipment knows what standards they need to adhere to (for equipment certification), and from that, the designer then knows which of the voltage figures is applicable to them.

Maybe that's the same reason that your capacitors have dual voltage ratings.


Modem7,

I had of coursed referenced your site and info on the caps. My first post had a typo. The caps are all X type. What I should have said was that I believe they are all class X1 caps. To be honest it is a bit confusing as I can not find an exact data sheet for the caps. Also thanks for the clarification on the voltages. I think that is exactly what is going on. because one voltage is listed as VAC and the other as ~V (I will post pictures later today). Of course to confuse things the data sheets I have found seem to indicate that the caps are 630V rated. For example the 0.022uF has a part number of PME 625 MB. According to the info I have found the M is the voltage rating which corresponds to 630V. So I am still a bit confused....
 
If I remember, I'll try later today to dig up the part numbers of the caps I chose to replace the Rifa caps that I found in my TRS-80 power supplies. Identical new (?) Rifa caps are still available, but I don't trust them. The part numbers I picked out might not be perfect matches for what you need, but they should at least help you narrow down your search through the tens of thousands of caps that Digi-Key carries!

NF6X,

Thanks for that. It would be helpful to have a starting point. I have been trying to search Mouser myself but again I a not sure how this all works out. For example after I specify the voltage and capacitance what else do I need to worry about? Also what numbers should I upsize to improve projected life span?

Edit: Here is one I found on Mouser. What do you think?
 
Last edited:
This is a topic that comes up every couple of years here. X-rated caps are designed to fail in a specific way. That failure mode depends on the application.

For example, a line-to-line bypass application can use an "X" rated capacitor, because such a capacitor is allowed to fail as a short, meaning that it will, worst case, blow the mandated line fuse.

However, the same capacitor is not suitable for a line-to-chassis bypass application, as a failure as short would effectively connect one side of the AC line to the chassis and present an electrical shock hazard. Y-labeled capacitors are the only suitable device for this application.

The number after the X or Y determines what impulse voltage the capacitor will withstand before failing. X1=4kv, X2=2.5kv, Y1=8kv and Y2=4kv. Note that, in this case, a larger number does not mean better!

You can use a Y-rated capacitor in place of an X, but not the converse.

For general use, X2 and Y2 capacitors should be fine and, in fact, are the type most often encountered in consumer devices.

You may run across old paper capacitors used in line filtering that are rated at 630v DC. It's okay to replace them with film or ceramic safety caps that are rated for 250v AC. In general, always go by the AC rating when replacing a safety capacitor.

If I were to have to make a choice, I'd select an AC-rated safety ceramic disk capacitor. They're just about indestructible. The problem is that they max out at about 0.01 uF. But they don't fail very often and they don't catch fire.

Otherwise, polyester or polypropylene capacitors in this application are about equal. The "box" style comes about because the extra encasing material is supposed to suppress fire--it has nothing to do with the capacitor itself.
 
I'm not positive that I chose the correct safety ratings (X vs Y) for all of the caps that I replaced. The ones I posted may be a good starting point, but please do sanity-check them for your application.
 
Thanks to everyone for the informative replies specially Chuck who cleared up a number of the issues in a very concise manner. After doing a bit of reading and research the Polypropylene seem to be the best option for replacement. After doing a bit of research this is what I came up with:

0.1uF, 250V, X Measurements: 23mmx14mmx8mm Lead Spacing: 20mm EPCOS B32653A104K
0.01uF, 250V, X Measurements: 18mmx11mmx5mm Lead Spacing: 15mm EPCOS B32652A103K
0.022uF, 250V, X1 Measurements: 18mmx14mmx8mm Lead Spacing: 15mm EPCOS B32652A223K

Anyone, with far more knowledge then me, see nay problems with the replacement choices? I believe the EPCOS caps all exceed the DC voltage requirements of the original, while matching the AC voltage, are made of PP and have a higher temp tolerance at 110 (vs. 85 for the originals). My only concern is fitting in the 0.1uF replacement physically on the board. It sits right up against the 0.01uF cap in a tight area. BTW: I am not familiar with EPCOS as a company but from what I gather they are German company that used to be owned by Siemens and is now owned by TDK so they should be reliable.
 
Do the EPCOS caps have identical lead spacings to the Rifa caps that they'll be replacing, or at least identical lead spacing to holes available on the board? That could be an important detail because if the Mouser pictures are accurate, they appear to have very short PCB pins. The caps that I used came with leads around 2" long, allowing for considerable fudge factor.

Are all three of the originals X-rated? One of the caps visible in your pictures doesn't have its safety rating visible on top.

If all three original caps are X-rated, then those EPCOS caps should be fine electrically as long as they'll work out physically. I like the pretty blue color, too. :)
 
Even the not-yet-blown ones appear to have cracks in the plastic case. My hypothesis is that those cracks are what let moisture in, leading to their eventual demise.
 
Do the EPCOS caps have identical lead spacings to the Rifa caps that they'll be replacing, or at least identical lead spacing to holes available on the board? That could be an important detail because if the Mouser pictures are accurate, they appear to have very short PCB pins. The caps that I used came with leads around 2" long, allowing for considerable fudge factor.

They are pretty close. Actually the two smaller ones are exactly the same while the bigger one is about 2mm wider on the EPCOS. Also remember the pictures are "for reference only". ;) I'd be very surprised if they came with such short leads!

Are all three of the originals X-rated? One of the caps visible in your pictures doesn't have its safety rating visible on top.

If all three original caps are X-rated, then those EPCOS caps should be fine electrically as long as they'll work out physically. I like the pretty blue color, too. :)

As far as I know. None have a grounding pin so they can't be Y. One datasheet I did find indicated that one of the caps was class X1 and not X2 as expected but I have not been able to substantiate that any where else.
 
The PME 271 M series are X2 Class, Data sheet here and the data sheet for the PME 265 Series is here
 
My internet connection is on the fritz, I can't get the data sheets for the Epcos caps you linked to, but it looks like they are not X2 class, Not sure if they would be suitable to replace the 265 series, Have you checked the data sheets.
 
My internet connection is on the fritz, I can't get the data sheets for the Epcos caps you linked to, but it looks like they are not X2 class, Not sure if they would be suitable to replace the 265 series, Have you checked the data sheets.

it does not say specifically the class but it does indicate they are for switching mode PSUs which I believe computer PSUs fall under, no?
 
it does not say specifically the class but it does indicate they are for switching mode PSUs which I believe computer PSUs fall under, no?

Oh, wait.. I mistakenly assumed that those EPCOS caps were safety-rated. It does not appear that they are, and I do not recommend using them for this application. Pick out capacitors which are specifically rated with the same safety class (X or Y) as the originals, and the same or higher impulse voltage rating.

Caps which are intended for use in switchmode power supply circuits are designed and specified to withstand high ripple currents in the switching portion of the PSU circuit, typically switching somewhere between a kHz and a MHz, while safety-rated caps are designed to fail in a particular way when used for EMI suppression and connected directly to the AC line. Those are two different applications, and either or both types may be present in any particular PSU.
 
The PME 271 M series are X2 Class, Data sheet here and the data sheet for the PME 265 Series is here

Interesting, they must have upgrade the PME 271s. The part I have is a 250V part and 85C tolerance. However, the data sheet indicates it is a 275 part with a 110C tolerance.
 
Oh, wait.. I mistakenly assumed that those EPCOS caps were safety-rated. It does not appear that they are, and I do not recommend using them for this application. Pick out capacitors which are specifically rated with the same safety class (X or Y) as the originals, and the same or higher impulse voltage rating.

Thanks for catching that. Back to the drawing board.
 
it does not say specifically the class but it does indicate they are for switching mode PSUs which I believe computer PSUs fall under, no?

Yes but--the caps we're talking about have very specific failure modes when connected to the AC line--said modes are intended to keep you breathing regularly.

Really, use X or Y-rated caps. It's not a big deal.
 
Okay after much more searching and more guidance I found the following Kemet replacements:

0.01uF: PHE820MB5100MR17
0.022uF: PHE820MB5220MR17
0.10uF: PHE820MB6100MR17

The data sheet can be found here. The nice thing about the Kemet capacitors is that they are the replacement for RIFA so physical size is almost an exact match. They seem to have improved the AC V rating and the temperature tolerance. Only bad thing is that they are Polyester and not PP. Although to be honest most of the Safety Caps seem to be polyester.

The only other issue is that the 0.022uF is not stocked by Mouser. I would need to order this one separately from digi-key which adds to expense and time. So does the hive mind agrees these are a good choice for replacements? TIA!
 
Last edited:
Back
Top