Author Topic: How can this part fail...  (Read 3730 times)

0 Members and 1 Guest are viewing this topic.

Offline Rick LawTopic starter

  • Super Contributor
  • ***
  • Posts: 3423
  • Country: us
How can this part fail...
« on: April 12, 2015, 05:32:41 am »
This is not about repair, but I like to understand why things happen a certain way.  This one puzzles me.

Background info:

I have a (purchased) circuit with a buck convertor connected to a 100uH SMD inductor (cd54 size).  I shorted the output by mistake.  The high current caused some smoke (from then unknown parts).  After repairing, I now know the buck convertor (XL1509 with auto-cut-out at 2A) was undamaged and the inductor was the dead guy.  I have the stuff back to working now but I am trying to understand the problem.

I made the mistake on two different units so the failure was repeatable and the repair was repeatable.

With the repair done, I tried to understand how it failed...

What I don't understand is:

The enamel covered wire that made the 100uH inductor appears ok.  I removed the entire length  and inspected it.  I ran the DMM probe down the wire and the insulation seemed fine.  The continuity at the solder point seemed fine...  The wire is around 0.18mm diameter and about 1/3 meter in length.

Prior to disassembling the inductor, I connected the inductor to a scope - the method I used to measure inductance which is having it parallel to a capacitor and feed it a 1KHz square wave to measure the "ringing frequency".  Rather than seeing the familiar "ringing", I saw a spike and that was that.  It acted like a plan resistor.  The same with the second inductor on the second unit.

I can understand if the heat managed to melt poor quality enamel or perhaps even just some insulator that is not enamel.  That could have shorted the wire windings.  But having ran the DMM probe down the wire and the insulation seemed ok, that doesn't appear to be the issue.  May be the melted area is too small for the DMM to probe...

I cannot think of why if the wires are in OK condition that the inductance it should create is missing...  There should be no other elements in the inductor to fail.  Just a wire around a core.  So what could be wrong?

I am probably too close to the tree and missed the forest.  Any suggestions?


I have the second dead inductor intact for further investigation.  This is not to repair the inductor.  Solving puzzles and understanding it is what I seek.

Thanks
Rick
« Last Edit: April 12, 2015, 05:35:09 am by Rick Law »
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 12806
Re: How can this part fail...
« Reply #1 on: April 12, 2015, 05:51:11 am »
One shorted turn would be enough to make the inductor so lossy, it would be little better than a resistor.  You overloaded it, it saturated and the current increased significantly causing it to overheat, the enamel softened and a small defect probably developed between two adjacent turns where they touched.

Its very difficult to find a small hole in the enamel once you have disturbed it, especially as it would have ridges of displaced enamel either side of the defect.

Moral of the story is: If you are relying on the switcher chip's current limit to protect the circuit, make sure that the inductor can actually handle that current!



 

Offline Rick LawTopic starter

  • Super Contributor
  • ***
  • Posts: 3423
  • Country: us
Re: How can this part fail...
« Reply #2 on: April 12, 2015, 06:34:45 am »
One shorted turn would be enough to make the inductor so lossy, it would be little better than a resistor.  You overloaded it, it saturated and the current increased significantly causing it to overheat, the enamel softened and a small defect probably developed between two adjacent turns where they touched.

Its very difficult to find a small hole in the enamel once you have disturbed it, especially as it would have ridges of displaced enamel either side of the defect.
...

That's what I thought that the enamel was damaged.  I counted 62 turns.  I didn't think shorting a turn or two would have made a big difference so I am in error here.

The point about either side is a great one.  I can see that the two sides form kind of a ridge and prevented the DMM probe from making contact with the "hole" so the DMM doesn't see the exposed part.

You are right, when I pull the wire, it was a bit sticky which I attributed to the glue that held it.  It was no longer straight making contact with the DMM probe more problematic.

Thinking that the wire is OK might have been over-optimistic on my part.

...
Moral of the story is: If you are relying on the switcher chip's current limit to protect the circuit, make sure that the inductor can actually handle that current!

That was a purchased cheapo equipment.  It was not expected to be used the way I was playing with it.  The XL1509 buck part of the circuit with the failed inductor powers only the few chips on that board drawing merely about 50mA.  The load there that should not change to any significant extend.  The XL1509 buck was rated for 2A and similar inductor was rated for 500-700mA so there was plenty of headroom there.  I suppose their not putting current limit in is understandable.  Even a dumb SMD fuse would have been nice and would have prevented secondary damage on the board which I think likely occurred at least with one of the two units.  But given what that circuit was designed for, I can see why it was omitted.

Find and fixing the fault has been educational and rewarding.
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 12806
Re: How can this part fail...
« Reply #3 on: April 12, 2015, 06:44:01 am »
If you take a good unshielded bobbin core inductor and put it in your inductor test jig, you can easily see the effect of a shorted turn simply by slipping a close fitting soldered loop of copper wire over it to its middle.  It usually pretty much kills the Q dead.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: How can this part fail...
« Reply #4 on: April 12, 2015, 07:20:40 am »
To see the holes in the insulation simply take the wire and place in a glass filled with water, to which you have added a teaspoon of salt. Place a strip of foil in the cup as well, and attach to a 12V battery negative, and the wire to positive. Then you will see the holes by the bubbles that form there in the salt solution.

http://en.wikipedia.org/wiki/Electrolytic_cell

This is used during manufacture to test the wire, passing it through a water bath held at a high voltage, and checking if there is a current flow as the imperfections go through the bath.
 

Offline Rick LawTopic starter

  • Super Contributor
  • ***
  • Posts: 3423
  • Country: us
Re: How can this part fail...
« Reply #5 on: April 12, 2015, 04:54:10 pm »
Hey - water test is a great one! Just like soap water test for tire pressure.  Good one.

In hind sight - this is a case and point on faulty decision making and faulty analysis.

My error became so clear after a good night sleep:

1. I was prejudiced that the easy failure of the part was because some crazy cost cutting scheme in how this part was made.   I was convinced that the "core" was probably some magnetic/ferrite-power painted on layer that easily melted...  I was seeking confirmation for something I already subconsciously decided: must be some cheap core, cheap coating... whatever.   Such bias impeded me from really carefully thinking about the problem.

2. My test of the wire was faulty.  I held down the DMM probe lightly.  I was concerned that the probe would just scrape off the insulation coating.  Pre and Post unwind resistance may be more telling.  Continuous measurement of the resistance as I unwind the wire might have been even a better way.  At the time any "short" is unwind, the shorted points are no longer touching, there would have been a jump in resistance.  It might have been easier to notice on the DMM when the reading jumps.
This may not have worked - the wire may need the current (heat) to expand to force the small "holes" to touch.  A better way then DMM to test burn-through insulation was needed.

3. I was 2 dimensional instead of 3.  I got locked in thinking if there is a short, I have 61 out of 62 loops.  The short doesn't have to be adjacent loops.  It could be from one layer to the next.  The size of the core allows 3 loops side-by-side then the loops must go on top to the next layer.

For example: a short between layer-1-bottom-loop doesn't have to be only with layer-1-middle-loop.  The short could be from layer-1-bottom to layer-2-bottom loop thereby taking out 6 loops (wire could go layer 1 bottom, middle, top then layer 2 top, middle, bottom.)  So one short doesn't take 1/62 out, it could take many more.

So, my thought about 1 or 2 shorted turns should not matter is silly.  It was 2D thinking and the coil is 3D.

Reaffirming the need for better decision making is always a good thing...
« Last Edit: April 12, 2015, 05:06:14 pm by Rick Law »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf