Author Topic: Electrolytic cap with polarity marked wrong; now need to test many in-circuit!  (Read 8098 times)

0 Members and 1 Guest are viewing this topic.

Offline WorldPowerLabs

  • Regular Contributor
  • *
  • Posts: 62
Hi all,

We're in the process of building a run of boards.  This is a microcontroller-based timer.  One of the components on the board is a small, through-hole electrolytic capacitor (100uF, 16WV).

Well, after we'd built 100+ boards, I noticed that one of the caps in the bag had ITS NEGATIVE POLARITY BAND ON THE BODY POINTING AT THE LONG (+) LEAD!!  The shrinkwrap was 180 degrees out of place (and no, it's not loose -- it was installed wrong at the factory)!

These capacitors are from an authorized distributor, and are made by a very well-known and ostensibly reputable manufacturer.  I won't mention the maker, since the intent is not to bash a particular company.   I have contacted the manufacturer and they told me to stop using the lot of parts.

However, we now need a simple way to test the capacitors that have already been installed on the boards.  Obviously, we cannot check visually, since the "-" band may actually point toward the "+" lead.  The leads have been trimmed, so we cannot judge by lead length.  These caps have isolated metal cans; the can is NOT connected to either lead, so a continuity test from can to leads is out.  There are NO vent lines on the top of the can, nor are there any other visual marks besides the suspect band on the can's shrinkwrap.

Reworking all of the boards is NOT an option, due to a tight deadline.  We can replace SOME of the caps if they are definitely reversed.

I have seen some studies that indicate that reverse polarity, if the cap is operated well under its max. WV, is tolerable.  However, if the capacitance of the part decreases significantly because of damaged electrolyte, it is likely that our circuit will malfunction (due to power dips caused by the relay engaging).

So, in summary, this post is:
1.) to vent (no pun intended) about crappy QC and the time it costs us engineers
and
2.) to ask for a test method to check in-circuit.  Simpler is better, since it may not be an EE conducting the test.

 

Online tom66

  • Super Contributor
  • ***
  • Posts: 3646
  • Country: gb
  • Electron Fiddler, FPGA Hacker, Embedded Systems EE
Surprising that (Major Japanese Capacitor Manufacturer) would screw up like that.
 

Offline AndyC_772

  • Super Contributor
  • ***
  • Posts: 3592
  • Country: gb
  • Professional design engineer
    • Cawte Engineering | Reliable Electronics
Can you post the relevant portion of the schematic so we can see how the cap is used?

My first suggestion would be to apply a voltage across the cap and measure the current. I've just bench tested a 100uF cap similar to the one you describe, and within seconds of a voltage being applied the wrong way round, my PSU hit the 100mA current limit that I set. I can't believe that behaviour would be hard to detect in circuit.

Offline WorldPowerLabs

  • Regular Contributor
  • *
  • Posts: 62
Aurora, your post certainly makes sense.  Quality and reliability are important, so we may just have to find a way to rework the completed boards.  Although I am an EE, I'm not an expert in electrolytic caps specifically.  However, if the electrolytic experts tell me that there is no definitive way to check the polarity (and I'm beginning to suspect that there isn't) then we may just have to do as you suggest.  The customer will be building the final product over several days (may be as long as a week or two -- they are not certain), so there may be time to deliver half of the boards and rework the other half.

I cannot easily post only part of the schematic without posting the entire thing (would not really help to see only part).  I have a NDA with the client, so I'd have to check and see if the client is okay with posting the circuit.

Here's a little more information:  as I had mentioned, the cap is rated 16WVDC.  However, the most it would normally see in operation is 4.8V and there is a PTC in the positive supply rail, so it seems unlikely to me that the cap would fail catastrophically if the board is just connected to the outside world in the intended way.  However, I don't want to send these out with a possible latent failure waiting to happen.

I've certainly seen the aftermath of electrolytic explosions; the problem is that I just don't think this one is going to "pop" right away in normal use, if in fact we have any on the boards that are reversed.

I appreciate all of the wisdom you've all shared so far.

Ben
....
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 9305
  • Country: gb
OK I know I said don't fiddle around trying to test capacitors BUT, I can't help wanting to be helpful...

A quick google found a patent that may be of interest to you:

http://www.google.com/patents?id=1bsCAAAAEBAJ&printsec=abstract&zoom=4#v=onepage&q&f=false

I have attached it.

I used the search terms:  determining capacitor polarity by test

Also Agilent have a product so such can be done.....but it is costly and complex.

http://www.home.agilent.com/agilent/editorial.jspx?ckey=26530:atg:faq&id=26530:atg:faq&nid=-536900137.0.00&lc=eng&cc=GB

and another interesting Patent:

http://www.google.com/patents?id=qQo6AAAAEBAJ&printsec=abstract&zoom=4#v=onepage&q&f=false

Fraser
« Last Edit: March 21, 2013, 05:40:08 pm by Aurora »
 

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
Bite the bullet, replace them. Finding a measurement method, and performing the measurement on each board might anyhow take as long as as just replacing the caps.

You said 100+ boards, so lets say 150 boards. Assuming an average of 10 minutes to change a cap (very very generous estimate, slow workers, including workers having a coffee and wee break, etc.), you get changed six caps per hour, making 48 caps / working day. E.g. one poor guy working three days, or e.g. three guys working one day. I would go out looking for three guys now, if necessary contracting the work out. I would do it now, not messing around with any measurement activity.
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline AndyC_772

  • Super Contributor
  • ***
  • Posts: 3592
  • Country: gb
  • Professional design engineer
    • Cawte Engineering | Reliable Electronics
I have to agree - 100ish boards isn't that big a job, just get on with it.

I think the need for a 'weapons grade' response as described above depends largely on whether or not you've already shipped any faulty parts to your customer. If so then it's appropriate. If not, you have the option to simply fix the problem, and agree a revised delivery schedule with your customer.

It probably goes without saying that you should be testing the replacement caps before they're fitted. Safety precautions may be in order.

Offline mariush

  • Super Contributor
  • ***
  • Posts: 3938
  • Country: ro
  • .
Quote
I also note that you said the capacitors were in a 'bag' I can only buy capacitors in bulk on bandolier tape....I have to wonder if the capacitors are genuine releases, or are they 'back door' rejects from the factory ? Could they in fact be QC failures that someone managed to inject into the supply chain ?

AFAIK Panasonic sells them in bulk bags of 250pcs or something like that.

I agree with Aurora on everything he/she said, it would be too much of a risk if it was my company.

Go buy some Hakko 808 desoldering guns and do a mini production line and replace everything.

 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 1612
  • Country: us
I agree with everyone above, don't ship anything you think might have a part backwards.  But out of curiosity, have you intentionally installed a cap backwards in one of your boards and powered it up to see what happens?  Would the fault get caught in final testing?  If it causes the board to stop working, that might be your test.
 

Offline AndyC_772

  • Super Contributor
  • ***
  • Posts: 3592
  • Country: gb
  • Professional design engineer
    • Cawte Engineering | Reliable Electronics
...and if it doesn't, maybe it's time to update your test process ;)

Offline Neilm

  • Super Contributor
  • ***
  • Posts: 1453
  • Country: gb
Trouble with electrolytic caps is you can have them running in reverse polarity for a while. When I launched my last product, we found that the footprint was wrong for one of the caps and was installed wrong. We had gone through all our testing, qualification, alpha and beta trials (say 50 boards). None failed through any of these tests.

It was found when three boards of the first production run failed. We reworked over 50 instruments - recalling them from our US distributor having explained the issue.

Result - we get a big thumbs up from them on our responsiveness and pro-active stance from a problem we caused.

Neil
Two things are infinite: the universe and human stupidity; and I'm not sure about the the universe. - Albert Einstein
Tesla referral code https://ts.la/neil53539
 

Offline JuKu

  • Frequent Contributor
  • **
  • Posts: 566
  • Country: fi
    • LitePlacer - The Low Cost DIY Pick and Place Machine
...six caps per hour...
Boards lined up on the bench and proper tools, that should not take more than a minute. With breaks and lunch, doing 150 vaps should take less than a day.
http://www.liteplacer.com - The Low Cost DIY Pick and Place Machine
 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2442
  • Country: gb
    • my website holding page
If you have any doubt about it, get them changed. Any decent tech can change a cap in a couple of minutes so it should be a mornings work at most.

Offline Bored@Work

  • Super Contributor
  • ***
  • Posts: 3932
  • Country: 00
...six caps per hour...
Boards lined up on the bench and proper tools, that should not take more than a minute. With breaks and lunch, doing 150 vaps should take less than a day.

As I wrote
Quote
very very generous estimate
I delete PMs unread. If you have something to say, say it in public.
For all else: Profile->[Modify Profile]Buddies/Ignore List->Edit Ignore List
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
IMHO If you value your reputation you will change out ALL of the suspect capacitors instead of fiddling around trying to prove/disprove the wrapper polarity indication. I know you have already stated that the tight deadline precludes fitting new parts but consider the negative effect on your business of capacitor failures early on in service. If I was you customer, as I am to many electronics comnpanies, I would insist that you changed all teh capacitors to ensure product reliability. You could even ne risking your customers reputation if they are on-selling the boards in appliances. Please consider this carefully before relying upon a possibly unreliable test method with the components mounted.

My suggestion would be as follows:

1. Declare the problem to your customer...transparency is king where trust is involved.
2. Tell your customer what you intend to do about the problem and the time it will take.
3. Find out from your customer if you can correct the issue on a small batch of boards for them to use, and then send the remainder as soon as completed. It may buy you time.
4. Make a FORMAL complaint and compensation claim to your component supplier as such an error is the fault of the manufacturer and they will claim against them for any losses.
5. The manufacturers reputation is valuable so do not detail them here if you wish to have a good relationship with them on this case. You have already been very discreet.
6. Investigate hiring more temporary staff or overtime work to meet the deadline of your customer. The costs should be carefully recorded in order that they may be claimed. Do not consider trying to fleece the supplier, just claim what is owed...but then I believe you to be honourable so probably no need to mention this !
7. If you do not already have them, buy some high performance desoldering guns. I have used them for large capacitor change-outs and they are very fast indeed.
8. Set up a reworking production line to mass remove and mass fit the capacitors. Efficiency is king when on tight deadlines.
9. Record all progress and keep your custoner advised at all times of your predicted delivery numbers and dates.

You can see that this is no small matter, and you should make your supplier aware of the efforts that you have been forced to employ in order to meet your customers needs. If I were your customer, I would be sympathetic, provided you ensured that ALL suspect components had been removed and the boards re-tested with a suitable burn-in period.

I work in Government and can honestly say that if a supplier took risks with the reliability of a product and I later found out that they attempted to disguise a serious problem in production, I would recommend an investigation with potential loss of contract or the supplier being heavily monitored with a loss of trust occuring.

Do not even think about running electrolytic capacitors in the wrong polarisation....jeeeez that would rate you as a cowboy outfit for sure !

I also note that you said the capacitors were in a 'bag' I can only buy capacitors in bulk on bandolier tape....I have to wonder if the capacitors are genuine releases, or are they 'back door' rejects from the factory ? Could they in fact be QC failures that someone managed to inject into the supply chain ? If one is wrong then it is likely that many could be wrong as the mass and speed of manufacture could replicate the error on thousands of capacitors before being discovered. All should be considered highly suspect. Have you worked out a failure ratio on the remianing stock ? That should give you a very rough insight into the likely faulty capacitors on your boards. If no others can be found you may just have been unlucky with one capacitor but somehow I wouldn't count on that.

Just my 2 cents worth, but remember.....its very easy to lose your good reputation and it can be impossible to recover it in the eyes of your customer.

Fraser

This man speaks the truth... I also know people that have received fakes from quality vendors, through authorized vendors like Digikey. Could have been a third party between the manufacturer and the distributor. They try to keep this stuff quiet for obvious reasons. Just say you have a problem and they should work to get your replacmenets.
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 7378
  • Country: us
    • SiliconValleyGarage
who says the long wire indicated the + ? thought about that ?

i'd go with what is painted on the can. the polarity of the plates is formed during production. when they spray the ink on the sleeve they are still held in the conveyer where the forming was done... so the marking should be right.

i;d never buy caps in bags. ammopack 5mm pitch is what i want.

since we are talking electrolytics : the can is ALWAYS negative. ( due to the electrolyte used. ) simply check with an ohmmeter that there is no 0 hom between the can and the + rail ...
« Last Edit: March 21, 2013, 10:18:41 pm by free_electron »
Professional Electron Wrangler.
Any comments, or points of view expressed, are my own and not endorsed , induced or compensated by my employer(s).
 

Offline WorldPowerLabs

  • Regular Contributor
  • *
  • Posts: 62
Hi all,

I appreciate all of the input.  Although the schedule on this portion of the project is very tight, I was able to get 3 extra days added to the deadline.  I went ahead and removed all of the questionable caps myself, rather than risk problems in the future.  Neil is right -- I've seen electrolytic caps survive reverse polarity for a while and appear to work, but it's not a good proposition.

Fifteen years ago, before I became an EE, I worked in PCB assembly full-time so although replacing components in plated-through holes is not fun, I'm about as quick as anyone.  It took about 4 or 5 hours to remove the caps; I won't have replacements until tomorrow morning at the earliest.

As far as the long wire indicating "+," that's what the datasheet states and that's how the other 249 or so caps were arranged.  So, you may be right that the part was inked after it was formed and thus the marking on the body perhaps could be trusted and maybe the leads were cut opposite of how they should have been.  The problem is that I don't want to trust a part whose printing says one thing and whose other identifying characteristic is directly opposite in nature.  I speculate that the part may have rotated out of position between forming and printing because the printed band, although pointing to the longer lead, was somewhat askew with respect to the lead.

These caps were bulk in a bag.  A factory bag is 200 and we had ordered 250.  Thus, we had one sealed factory bag and one partial bag sealed by the distributor.  The questionable part was in the distributor-sealed bag.  The distributor is one of the big 4 or so companies in that business.

It did occur to me that the case is normally negative.  However, checking several samples with a DMM shows >10Mohm between either lead and the case.  It seems like these have an isolated can, although I didn't see specific mention either way in the datasheet.  This is true even if I apply a voltage across the cap to ensure that it is formed.

Thank you all for your input.  I have to say that the manufacturer was not especially helpful nor very easy to contact by phone in this matter.
 

Offline WorldPowerLabs

  • Regular Contributor
  • *
  • Posts: 62
Update:  I disassembled one of the caps, and indeed, there seems to be no connection between the metal can and anything else.  The leads pass through a rubber (silicone) insulator.  There is a plastic sleeve around the plate assembly.  So, the 10+ Mohm reading was just the leakage through the electrolyte itself, which was bridging the area between the cylindrical insulating sleeve and the metal can...
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 9305
  • Country: gb
The Agilent product indicates that it uses that negative plates proximity to the shall as a its means of detecting polarity. It also stated that not all capacitors are built the same. If all capacitirs had their negative plate actually in contact with the metal shell, polarity identification would be easy. I have many axial capacitors that do use the outer shell as the negative connection though.

Fraser
 

Offline JuKu

  • Frequent Contributor
  • **
  • Posts: 566
  • Country: fi
    • LitePlacer - The Low Cost DIY Pick and Place Machine
Yup, I tried to emphasize that and encourage to just get on with the rework. One capacitor, 100 something boards is not a big deal.
http://www.liteplacer.com - The Low Cost DIY Pick and Place Machine
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf