Inductor failure 1

[zappedagain]

We introduced a redesigned PCB a while back with a TI TPS54061 buck regulator to generate 5.1V from 24V. Out of about 30 PCBAs I've had two failures so far, one on the test bench and another infant failure while the system was being tested in manufacturing. The symptom is that the 5.1V regulator never completes the start up process and only gets to 1.3V. The TPS54061 starts at a lower frequency (f/4) and I can see it never moving on to the faster states.

I found the issue is that the 180 uH inductor is at 85 uH when tested in the circuit and that is too low for the buck to start properly; other PCBAs that are working properly show 180 uH in circuit. Now here is where it gets interesting...

I removed one inductor and had to use a lot of heat to do it, mainly with a hot air gun before I used a two-iron trick to remove it. After removing that part I tested it again and now it reads 180 uH! Is this some Neel temperature effect(ferromagnetic to paramagnetic) or something? This is a Coilcraft LPS4018-184MRB ferrite inductor.

So after I saw that I put the inductor back in the circuit; it reads 180 uH in circuit. I powered up and the buck started properly. Once. After that it failed multiple (every) times. I measured the inductance again and see 85 uH. Can this be some type of saturation effect? I need to check with the circuit designer, The inductor has a 240 mA peak current and the TPS54061 has a 350 mA typical current limit (500 mA max), so that might be a start up issue while charging the 10 uF output capacitor.

I pulled the inductor back off the PCB to repeat the heat test, and after about 60 seconds of heat the inductor is 100 uH, 120 seconds and it is at 110 uH but appears to level off there. After about five minutes of heat it went open circuit so that isn't a very good test.

So I've replaced the inductor on the two failed PCBAs and they both start fine 10 out of 10 times. But for how long? Did I just get bad parts or is there a root cause here that I'm missing? Is this a typical inductor failure mechanism in a buck regulator?

Z

 

[MacGyverS2000]

Can you post a pic of your layout?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[zappedagain]

Here is a zip file with the schematic and layout.

 

[itsmoked]

Danm Zapped! What have you stepped in now?

I would immediately pick-up the phone and call the best inductor supplier on earth: CoilCraft. They have excellent people. They will likely have already seen this and will know exactly what's happening. They're generally good enough to tell you which inductor you'll need if you tell them the controller chip. No one else can do that.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

I'm assuming your enable line is attached to a source lower than the pin's max of 8V (fixed a problem just like that earlier this week)... and you're not driving it high when not pulled low.

How did you come up with 180uH for the inductor? At that frequency and Vi-Vo difference, I would expect something less than 50uH.



Dan - Owner

http://www.Hi-TecDesigns.com

 

[zappedagain]

I'm right with you Keith. I talked with Coilcraft this morning and they suspected two things - a cracked case from using the wrong size P-n-P nozzle during part placement, or contamination from the cleaning process. Water clean flux can leave weak organic acids and if they collect in the inductor that eventually damage it. My manufacturer does use water soluble flux, and from their notes on this build it looks like the LPS4018 is installed and then washed; that's a no-no. A previous email indicates they will wash the PCBA and then install parts that are sensitive like this; it looks like somebody missed a step.

I disassembled a failed LPS4018 under the microscope and I can see variations in the insulation layer, and a lot of goop near the bottom edge. I'm pretty certain the flux is the issue (of course this is the unit that I put a lot of heat too, so I'll need to check the other failed unit too).

Coilcraft also said the variation with heat is probably due to spacing variations while things expand and contract. I did break a case open on one and it now reads about 85 uH, so that is in the same range.

Dan - I think the designer took this design right off TI's WeBench tool. By my quick calculations the min inductance is around 100 uH.

Z