Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Ceramic capacitor failure

Status
Not open for further replies.

electric1

Junior Member level 2
Joined
Apr 23, 2011
Messages
20
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,507
Hi all,
I'm hoping someone can comment on my problem. I have a simple buck regulator based on LM5008 which steps down 60V to 5V to power an MCU circuit.
Load at 5V side is low, around 15mA. Input comes from battery pack, which doesn't exceed 60V max.
I'm seeing failures of ceramic cap C7, which is C3216X7S2A225K160AB made by TDK, surface mount 1206 size X7S type.
Failure rate is below 1%, but several boards already failed with same exact results, which means a systemic problem.
Failed caps exhibit low resistance, which causes R1 to fuse, shutting down the circuit.
Failure happens after several weeks in the field, never during initial QC.
Circuit is permanently attached to a large lithium battery, so there should be no transients.
70V TVS is present to quench any transients, just in case.

I have a suspicion that a batch of caps of lower voltage rating were mistakenly used in manufacturing, but I don't know how this can be proven or even checked for.
SMD caps have no markings and look alike. Is there any way to test voltage rating?
For example, if a 50V cap was installed instead of 100V and system is running around 53V nominal, would a small number of caps eventually fail due to slightly exceeded voltage rating? I imagine actual voltage breakage has a range and variability to it, so maybe some caps would fail, while others don't?
Also, the nature of the failure where capacitor develops DC continuity over time, does it point to a voltage based breakdown or some other failure mode?

Any thoughts and feedback is appreciated.

Regulator schematic.png
 

The best way to estimate the capacitor voltage rating is to determine capacitance voltage dependency.

Did you consider mechanical failure?
 

    electric1

    Points: 2
    Helpful Answer Positive Rating
Hi,

is it possible that the capacitor gets hurt whet it gets connected to the battery. The series resistance is low, thus I expect a high current peak. 60V combined with 2.2uF is a lot of energy.
Maybe the current hurts, but maybe combined with the wiring inductance it may cause a resonance with high voltage peaks.
Don't know.
Do some measurements.

***
But ceramics sometimes are mechanically more sensitive than they look.
Are they mounted close to the PCB edge? Are they mechanically or thermally stressed?
A mounting hole, or a screw nearby?
Even a wrongly calibrated assembling machine may cause damage.
Storage box, maybe even during assembling, test or packing.

I rather think about mechanical stress, than electrical stress.

Klaus
 

    electric1

    Points: 2
    Helpful Answer Positive Rating
Thanks both of you for your quick and insightful response.
Yes, I also considered mechanical stress as possible root cause. Cap is at the edge of PCB and some pressure could have been applied during installation. We also addressed this issue by designing a better PCB mount in latest product revision.
However, the failure never happens during QC, only weeks after field deployment.
Is it possible that stress during installation or inrush of initial DC connection cause it to fail at a much later time?

I just found this interesting paper https://www.niccomp.com/resource/files/ceramic/VoltageCoefficientofCapacitors-032012-R1.pdf
which indicates that ceramic cap is a poor design choice for my circuit (unstable over VDC), while LM5008 application notes specifically recommends 100V ceramic cap at the Vin pin. Which is right? Now I'm terrified that many more boards could fail due to capacitor aging.
 

Hi,

Yes, late fail is possible. Micro cracks may pass humuidity to the inside of the capacitor.
It may cause metal creepage, maybe supported by applied voltage.

Klaus
 

    electric1

    Points: 2
    Helpful Answer Positive Rating
how hot does it get under normal full power operation ... ? if it does not get warm - then mechanical stress is most likely culprit.
--- Updated ---

also for 100V part, the C @ 60V will be quite a bit less, closer to 1.2uF - so the ripple voltage and current will be higher ...
--- Updated ---

also - what is the operating freq ... ?
 
Last edited:

    electric1

    Points: 2
    Helpful Answer Positive Rating
how hot does it get under normal full power operation ... ? if it does not get warm - then mechanical stress is most likely culprit.
--- Updated ---

also for 100V part, the C @ 60V will be quite a bit less, closer to 1.2uF - so the ripple voltage and current will be higher ...
--- Updated ---

also - what is the operating freq ... ?
It doesn't even get warm during normal operation.
Frequency is approx. 200kHz
We also suspect that some mechanical stress could have been induced during depanelization by the manufacturer.
This capacitor is too close to the PCB edge, which was a design weakness and will be addressed in the next revision.
I will also use automotive rated Flexiterm part instead of general purpose rated one.

The biggest fear is how many more failures will happen over long time. If there is no failure within first 1-2 months, does it mean probability of failure goes down with time, or goes up or stays the same?
 

You don't get significant capacitor losses with 15 mA switcher load current.

We also suspect that some mechanical stress could have been induced during depanelization by the manufacturer.

That's actually the most likely way to cause MLCC failure. No matter if it happens immediately or some weeks after initial operation, I would still consider it early failure. Long term reliability of MLCC is generally good.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top