ITS4142 failing in field, without any obvious reason

[CosteC]

Hello guys

I got number of ITS4142 failing in field.
Each is driving LED lamp (24V, 35 mA) on door of cabinet. There literally a meter of cable between lamps and unit, 24 V is stable, no other damage in this or other modules. No damage to cabling like cable cut by door, but ITS got holes burned in them (see photo)
Sometimes 2 out 3 used, sometimes all 3. Failures happen months after installation.

Any ideas?

Datasheet:
https://www.infineon.com/dgdl/Infineon-ITS4142N_20060309-DS-v01_01-en.pdf?fileId=db3a304412b407950112b4296b0840e1

[BennoG]

They could drive 1.5A so that would be no problem.
Things come to mind:
- Fake components.
- Short circuit on the led.
- Somehow a lot more current than the 35mA (it should be self protect)
- Ringing
- not enough decoupling to satisfy the internal charge pump.

 Benno

[CosteC]

They could drive 1.5A so that would be no problem.
Things come to mind:
- Fake components.
- Short circuit on the led.
- Somehow a lot more current than the 35mA (it should be self protect)
- Ringing
- not enough decoupling to satisfy the internal charge pump.

 Benno

Thank you Benno

- Fake components hypothesis I investigate now, waiting for response from Infineon.
- Shorted LED or more current seems fishy - LEDs work when module is replaced, cabling is not cut in visible way ( I cannot check myself, units are continent away)
- Not enough decoupling is interesting. Indeed there is no good decoupling at those chips.
- Ringing - I doubt it, cable to LEDs is maybe a meter long, ITS4142 is designed to drive relays and other inductive loads and it is rated for it. There shall be no PWM on those outputs too. I do not see any overheating marks either. It looks like sudden explosion. Very puzzling.

Do you have experience with poorly decoupled ITS4142 or similar?

[RoGeorge]

At a first look, they seem too close to each other.  In general, capsules with a radiator tab rely on the PCB copper for heat dissipation.  Use an infrared camera to measure the temperature of each chip during the worst working regime, inside its enclosure and at the highest expected room temperature.

Apart from that, I would also look for very high frequency oscillations, or if the absolute maximums request of the datasheet are obeyed at all time, including during transient regimes.  Even when the schematic is correct, a bad layout of the PCB traces might make a stable circuit unstable.

It's hard to identify the reason without schematic + PCB layout + mechanical enclosure.

[CosteC]

Normal load current is around 35 mA at 24 V - far from absolute nominal ratings.
I am sure decoupling is insufficient and there is insufficient reverse polarity protection - no resistance in GND terminal. On other hand reversing power require swapping two cables with screwdriver and polarity is correctly marked on enclosure.