What could have caused an EMI Filter Bead shorted by itself?

[Ken Leong]

Hi, I am a Reliability Engineer. Currently, I am facing difficulties in finding the root cause to a defective EMI filter bead. Resistance measurements across the four leads (vertically, horizontally, diagonally) show that they are all shorted. However, a good ferrite bead should only be shorted horizontally as tested. From the x-ray picture, I couldn't find any abnormality. It looks perfect with two wires running in parallel. I am hoping to find some explanation in the world of Physics? Thanks!

 

[sophiecentaur]

HI
How does it perform at the RF frequencies it's specified for? Do you know it is faulty?
Was the device subjected to high volts or current before it exhibited this behaviour? Could you just have fried it?

 

[berkeman]

Hi, I am a Reliability Engineer. Currently, I am facing difficulties in finding the root cause to a defective EMI filter bead. Resistance measurements across the four leads (vertically, horizontally, diagonally) show that they are all shorted. However, a good ferrite bead should only be shorted horizontally as tested. From the x-ray picture, I couldn't find any abnormality. It looks perfect with two wires running in parallel. I am hoping to find some explanation in the world of Physics? Thanks!

Welcome to the PF.

Can you provide the part number and a link to the datasheet? Can you upload a picture?

 

[CWatters]

Do you have a new one to compare your faulty one with?

 

[wirenut]

Was the unit installed on a board?
Did it initially pass QC? or was found failed during QC?
Did it fail after a time?
Was it removed for testing/ x-raying?
Can you upload the x-ray?

 

[Ken Leong]

The datasheet can be found here. Part number is CM3312R111R-10 from Laird Technology.

 

[berkeman]

Well that's super helpful, Ken. Except for the part where you explain what those pictures represent...

 

[wirenut]

Could you explain what each x-ray is?

 

[wirenut]

Darn it, Berkeman.
You beat me to it.

 

[berkeman]

View attachment 97487 View attachment 97487 View attachment 97488 View attachment 97489 The datasheet can be found here. Part number is CM3312R111R-10 from Laird Technology.

The dimensions of your parts in the x-rays do not seem to match those in the datasheet you linked...

 

[Ken Leong]

Sorry, not sure how to upload the files. The first 2 photos are of the same faulty ferrite bead which I had mentioned with no abnormality found. The 3rd photo is taken from another failed FB (2nd case) with cracked found inside. Both components are from different boards. The 4th is a sample from a good component.
Some background:
- Two cases of EMI filter failure. Both cases were found when the -12V input on the board was found shorted.
- Suspect a high current/voltage surge occurred.
- Components were removed from the boards for failure analysis.
- Comparison on the resistance measurement of both failed and good components were performed. Resistance measurements of failed components showed all leads are shorted.
- X-ray was taken on both failed components. Only one showed elements cracked within, and the other showed no symptom.
- OEM mentioned that there was a shelf life issue on this component. When it is removed from the packaging, the shelf life will be reduced to 3 months from the original 12 months. This part was later removed by OEM in year 2012.

 

[Ken Leong]

Hi Berkeman, how can you tell there is a different in dimension?

 

[jim hardy]

Like any transformer, inter winding short ?i never handled one of these tiny surface mount inductors. How in the world does one make them ?

What I've seen in transformers hundreds of time bigger that that one :

overcurrent heated it point of insulation failure ,,, it's a 5 amp device per datasheet you linked
overly enthusiastic soldering at assembly heated it to point of insulation failure
overvoltage between windings pierced insulation and spot welded the wires
manufacturing error applied too much tension and varnish insulation crept away after time
varnish insulation softened by solvent(flux remover?) or time and insulation crept away
some insulating material not correctly placed inside
solder bridge shorts terminals together - there's only 0.030 inch between the terminals and tolerance stackup could reduce that to 0.010 .

You may have to break it open.
or, the folks who made it can probably analyze it for you

Our safety department had a binocular microscope for analyzing asbestos. I found it handy for examining small electronic devices.

from datasheet you linked

Good luck, let us know what you find ?
You may have to break it open.

Surely your trade journal has advertisers for failure anaysis ?
http://www.asminternational.org/materials-resources/journals/journal-of-failure-analysis-prevention

old jim

 

[Ken Leong]

Hi Jim, thanks for your advice. How would you advice me to open up the component? Horizontally or vertically?

 

[Ken Leong]

Wirenut and Berkeman, is my explanation on the x-ray clear? :P

 

[Ken Leong]

Hi Sohpiecentaur and CWatters,
Thanks for your questions. Do you need further information?

 

[jim hardy]

Wow a lot of posts appeared while i way typing.. i hadn't seen the pictures.

- Components were removed from the boards for failure analysis.
- Comparison on the resistance measurement of both failed and good components were performed. Resistance measurements of failed components showed all leads are shorted.

In that sequence? The one in first two photos is shorted as we speak , after removal from the board ?

- OEM mentioned that there was a shelf life issue on this component. When it is removed from the packaging, the shelf life will be reduced to 3 months from the original 12 months.

Ask OEM if he's having "Tin Whisker" issues... i curse lead free solder. We refused to use it in my nuclear plant.

How would you advice me to open up the component? Horizontally or vertically?

Ideally i'd want to split it horizontally at center plane.That way you'd see the wires laying side by side undisturbed..and anything else in there that dpesn't belong.
Where i worked we didn't have micro-machinery, any bolt under one inch was a small one. I'd have asked a tech with steady hand to see if he could remove the top with a Dremel tool.
If it's hard ceramic ,,, wow.. diamond wheel ?

Surely there's somebody here accustomed to surface mount ?

i'll watch for now, better qualified folks than me are aboard.

 

[jim hardy]

http://www.mouser.com/ds/2/223/laird technologies_cm3312r111r-10-608891.pdf

ruh-roh.

 

[Ken Leong]

Haha...Jim, thanks!
Yes, the first two photos are shorted component after removed from the board.
Tin whisker? Hmm...I will look into it.

 

[Baluncore]

I agree that tin whiskers are a probable cause of an external short between the conductors.

Metal whiskers are like fuses, they can be temporarily blown, until they regrow. If you connect the two separate conductors to a high current, low voltage cell, (say 1.2V), does the short disappear? If not, increase the voltage to 12VDC, has the short been removed? If not then you have a problem bigger than whiskers.

DC voltage in the presence of moisture can promote the growth of whiskers and is a real problem with tin solder. What is the potential difference between the two parallel conductors when operating normally on the PCB. Are the currents power supplies or signals?

 

[sophiecentaur]

Hi Sohpiecentaur and CWatters,
Thanks for your questions. Do you need further information?

The other guys seem to have dealt quite competently with it. I shall leave it to those able younger members to sort out your problem and return to my pipe and slippers.
. . . . . .
Actually, on second thoughts, I could point out that the shadow of the Xray makes the device look like a directional coupler. Is that a daft suggestion, chaps?

 

[CWatters]

Sorry but I'm confused...

The first 2 photos are of the same faulty ferrite bead which I had mentioned with no abnormality found.
The 3rd photo is taken from another failed FB (2nd case) with cracked found inside. Both components are from different boards. The 4th is a sample from a good component.

I can see why #3 might be shorted diagonally but are you saying that #1/2 is also shorted diagonally?

 

[jim hardy]

I can see why #3 might be shorted diagonally but are you saying that #1/2 is also shorted diagonally?

I share your consternation.

Haha...Jim, thanks!
Yes, the first two photos are shorted component after removed from the board.

That doesn't unambiguously answer the question "have you checked AFTER REMOVAL that it is still shorted?

I hadnt seen these when i typed my generalization earlier .

My first impression of OP's pictures was that the ends got torn and bent in process of removing the part
and we see the remains of either a solder bridge or a flaw in the board etch here - pads so close together they bridged in manufacture.

@Ken Leong : Put the board and the part under a 20X binocular microscope.

 

[berkeman]

- OEM mentioned that there was a shelf life issue on this component. When it is removed from the packaging, the shelf life will be reduced to 3 months from the original 12 months.

What? I've never heard of a "shelf life" in or out of packaging for inductive components. For ICs there certainly be a moisture uptake issue before being soldered onto a PCBA, but not for inductors. Are you sure the OEM said this about the inductor?

 

[berkeman]

Hi Berkeman, how can you tell there is a different in dimension?

I guess it's okay after all. The datasheet drawing looks shorter, but the dimensions listed have the ratio of 8.5/3, which is close to what your x-rays show. Sorry for the false alarm about that.

 

[CWatters]

What? I've never heard of a "shelf life" in or out of packaging for inductive components. For ICs there certainly be a moisture uptake issue before being soldered onto a PCBA, but not for inductors. Are you sure the OEM said this about the inductor?

Perhaps it's to do with the solder contacts? Perhaps they tarnish and become harder to solder out of the packaging?

 

[berkeman]

Perhaps it's to do with the solder contacts? Perhaps they tarnish and become harder to solder out of the packaging?

Hmm, that sounds like a possibility. I'm home now, but will ask our Ops folks on Monday whether they've heard of this. Thanks.

 

[Baluncore]

The condensation of water from air onto tin surfaces forms an oxide film that prevents solder wetting with normal fluxes. As part of the packing process, either a desiccant is added, or packing is done and sealed in a dry air or nitrogen environment. As quoted, storage in non-condensing air gives only a 3 month shelf life.

After components are removed from the manufacturers dry packaging, the use of a sealed plastic bag to store components will encourage condensation if the temperature falls below the dew point. That will rapidly destroy component usability.

Way back in the last century, corrosive plumbing fluxes were formulated to remove the oxide layer when soldering. The fluxes used now with tin solder are not as aggressive and so do not cut through or lift the oxide.

 

[Jeff Rosenbury]

Ideally i'd want to split it horizontally at center plane.That way you'd see the wires laying side by side undisturbed..and anything else in there that dpesn't belong.
Where i worked we didn't have micro-machinery, any bolt under one inch was a small one. I'd have asked a tech with steady hand to see if he could remove the top with a Dremel tool.

Dremels wiggle (precess?) too much for surface mount work at this level. I usually try using a razor knife. Perhaps a jewelers saw might work with ceramics, but some parts are about the size of a sawblade's width.

If it's plastic, a box cutter gives a strong, clean, and disposable cut. (Frequent blade replacement is a must.) If you need something a little harder, there are scribe tools with harder points.

 

[Jeff Rosenbury]

There is an anomalous dot in the x-ray. Do you know what it is? It can't be good, but I doubt any problem it caused would show up at DC.

What material is the case?

You are aware this item is no longer supported.

It's hard to imagine a DC short without some metal between the two bars which should show on the X-ray. There are conductive plastics and salt bridges to consider, but these seem unlikely.

Here's a link to something on tin whiskers. They seem a stretch, but maybe...

Perhaps tin whisker growth can happen in the case packaging and that's why they were discontinued?

 

[Ken Leong]

Hi Guys, thanks for sharing and giving me all the valuable feedbacks and suggestions. Just to answer to some of your queries.
- The first two pics are from the same component (sorry for the confusion) and yes, it is shorted diagonally across the terminations when measured standalone after removal from the board.
- The board was conformal coated, not sure if tin whisker can be the possible cause of failure. Will further check on that.
- The OEM did mentioned about the storage issue and that's probably why they had discontinued this part.

Jim, the shadows you saw are the solder remaining after removal from the board.
Baluncore, thanks for your explanation on condensation impact on the component during storage.

P.S. How do you guys paste my comments when replying?

 

[Baluncore]

P.S. How do you guys paste my comments when replying?

We highlight your text, then a tab drops and we click on quote or reply. It pastes your quote in our new post box.

 

[davenn]

- The first two pics are from the same component (sorry for the confusion) and yes, it is shorted diagonally across the terminations when measured standalone after removal from the board.

so it should also be shorted between both terminations on each end as well then ?

Dave

 

[Ken Leong]

so it should also be shorted between both terminations on each end as well then ?

View attachment 97668


Dave

Yes, Dave. All four terminations are shorted.

 

[davenn]

Yes, Dave. All four terminations are shorted.

OK

pretty crazy when there is no obvious sign of a short
Is it really a short ... 0.0 Ohms or is it just low resistance a few 10's of Ohms or similar ?
I would warm up a couple of faulty ones ( that look OK in x-ray - like the one I posted) and see if it drives off any internal moisture
and test resistance again

I just find it so hard to see how you can have a short cct without anything obvious

 

[jim hardy]

Yes, Dave. All four terminations are shorted.

To Dave's question

Ahhhhh, terminology... One man's short is another man's leakage...

I always use a an analog meter for such tests. That's because on the RX1 ohms range it applies tens of milliamps to the device under test.

Wetness gives a part-scale reading determined by the electrochemistry at the fault.
A hard short circuit that we'd expect to see on those Xrays will read very near zero

A digital meter on a high ohms scale will report a low number for a few hundred ohms of something conductive like carbonized insulation.

I'll bet Dave was thinking the same thing.

Individual tin whiskers will usually get burnt away by the current from an analog multimeter on low ohms scale.

I have seen however a "tin beard" , thousands of whiskers in a clump, growing between two adjacent terminals on a barrier terminal block. It passed enough current to actuate a solenoid valve and trip the power plant .
Tin is not the only metal susceptible

http://nepp.nasa.gov/whisker/reference/tech_papers/2007-brusse-metal-whiskers.pdf

That's why i suggest somehow getting the top off one of the failed ones so you can see inside. A clean interior says "well, it wasn't that."

old jim

 

[Baluncore]

We need some idea of the circuit application / impedance. Does the bead isolate high impedance, low power signals or does it isolate low impedance, higher power supplies? The presence of intact single whiskers in a power supply situation is unlikely, but there may be a surface coating of metal due to vaporisation of many whiskers over time.

An accurate resistance measurement would be most informative.

At some point you need to use destructive testing to further the research. It is almost certain that any whiskers present will be lost if you attempt to physically expose the component's internals for visual examination. That is why in my post #20, I suggested fusing whiskers with a DC supply. By using sequentially larger charged capacitors as an energy source you might get a good estimate the mass of the whisker(s) or short involved.

 

[Clark Magnuson]

That is a 5 mOhm 5 amp single turn balun made of Copper and Ferrite. It it shorted out, that must be an isolation test with an Ohm meter. No debris could represent a short that would shut down that kind of power line. My guess would be contaminated part or board or the part was damaged when soldered, caused a crack. Pieces of Ferrite do not conduct much.