High Ambient Temperature Effects on SMPS

[Maxwell077]

Hi All,

We are building for huge LED screen displays.
To power up LED systems, we are using enclosure boxes that placed Meanwell LRS-350-24 inside.

In one of our project, we have faced with an issue.

Issue: MLCC cap 0.1uF 50VDC X7R 0603 over the DC voltage line of LED module is short-circuited.
When it happens : Approximately after two months later.
Quantity : Our system has more than 48.000 MLCC caps as 0.1uF, max 75pcs is defected after given period and it is randomly happened.
Observation : Main focus point is after a long term working of SMPS's inside an enclosure.

Questions: May SMPS can cause any transient voltages or voltage spikes after long-term working in hot ambient temperature that may defect any electronic component.
This high temperature is not about ambient, it is about closed enclosure box.
May SMPS degrage as its performance and cause some abnormal operations that may burn capacitors on its powered LED modules.

Best Regards

 

[cupoftea]

chip caps often go short due to flexion of PCB, which can be caused by expansion/contraction in some cases.
Electrically, i woudlnt expect an 0603 50v to go short unless a serious overvoltage went on it....that may be due to some other thing failing in the circuit of course...do you have any electrolytic caps in the circuit.......do the meanwells still work when the product is returned?

 

[mtwieg]

Borrowing this from murata:

Short circuit failure mode in MLCCs has a few potential causes, one of which is overvoltage stress. I assume these caps are bypassing the 24V output of the meanwell supplies? I would never expect a 24V supply to produce transients >50V, unless the supply has been damaged already.

The cause I would look into first would be cracking due to mechanical stress. The mechanical stress could be due to external forces on the PCBA, but if this is for a LED display then I'm guessing they don't move or vibrate significantly.

Thermal behavior can also lead to mechanical stress, even if the capacitor itself is not dissipating heat. If the entire PCBA changes temperature, some strain is imposed on the capacitor due to it having a different coefficient of thermal expansion relative to the PCB it's attached to. The actual point of failure is hard to predict, and has to do with the extent of the temperature fluctuation, the number of thermal cycles, the soldering process, etc.

This is why so-called "soft termination" MLCCs exist, I believe.

 

[Maxwell077]

The cause I would look into first would be cracking due to mechanical stress. The mechanical stress could be due to external forces on the PCBA, but if this is for a LED display then I'm guessing they don't move or vibrate significantly.

In my obervations and tests, it is most probably not about mechanical stress.
LED modules are potted with epoxy-resin and mounted with a strict mechanical cases and one system placed on bridge - no issue but lots of vibration-, other system is mounted on strict wall -we have issues, less vibration than bridge-.

Short circuit failure mode in MLCCs has a few potential causes, one of which is overvoltage stress. I assume these caps are bypassing the 24V output of the meanwell supplies? I would never expect a 24V supply to produce transients >50V, unless the supply has been damaged already.

The issues seems more electrical and thermal basis.
I need to respecify that, capacitor failues are arised after long period such as more than a month.
LED modules are on the facade of building, but MW SMPS are inside of building just right behind of the modules. These SMPS were worked in same closed-enclosure more than a month.

I check for derating curves and lifetime of internal components of SMPS's.
Can SMPS cause for transients or spikes if output aluminium electrolytic caps or voltage regulator circuits or any other sensitive circuits are degraded?
As we knew, aluminium caps are sensitive to ambient temp and SMPS power performance also degraded with temperature.
For eg; after performance degration, load is still 300W but SMPS can output 150W.
LED modules are highly variable load type, because sometimes all modules can be all white, sometimes all black or all mixed colors.
This kind of load variability may cause issues that could damage MLCC caps.

 

[cupoftea]

..yes if the 'lytic caps dry out, you can get overvoltage output.....should be easy to check if dryed out that much or not. Also, cold temperature can cause high esr in lytic cap, so its voltage may go higher.

 

[Maxwell077]

Could this overvoltage exceed 50V's to damage MLCC caps that has 50V rating.

 

[cupoftea]

Yes...but it should be easy for you to find out if the meanwell caps are thus afflicted...just power them up and see.
Other thing is depanelisation damage to the chip caps....specially if they are right near board edge?

Also, can you be sure that in all stages of assembly/transportation, no-one has physically abused these PCBs , such that the chip caps got cracked?

 

[mtwieg]

In my obervations and tests, it is most probably not about mechanical stress.

What do you base this on? Have you inspected the failed boards and their power supplies?

Can SMPS cause for transients or spikes if output aluminium electrolytic caps or voltage regulator circuits or any other sensitive circuits are degraded?
As we knew, aluminium caps are sensitive to ambient temp and SMPS power performance also degraded with temperature.
For eg; after performance degration, load is still 300W but SMPS can output 150W.

A faulty electrolytic cap by itself wouldn't cause the output voltage to rise. It's possible that voltage transients might become more severe, but going over double the nominal output voltage for any amount of time is far beyond what I've seen from any COTS supplies. The only way I can see such overvoltage ocurring is if something else is actually sourcing significant power back into the supply. This can happen in some applications like motor drive, but I doubt it could be significant in LED drivers.

And to actually make an MLCC fail short you'd need a transient well beyond 50V, or it would have to be very long duration or very frequent. It should be easy to recreate those conditions in the lab.

LED modules are highly variable load type, because sometimes all modules can be all white, sometimes all black or all mixed colors.
This kind of load variability may cause issues that could damage MLCC caps.

Load transients can cause transient rise in the supply voltage, maybe ten or twenty percent. A factor of two is far beyond anything I've encountered.