LT3650 Lion battery charger heat issue

[Moonwalker]

Thanks all for the suggestions.

@treez - Already have a Pic on the board so might add that temp sensor since its quite cheap and accurate enough for this application.
- However cant understand why the onboard temp sensor of the LTC should not be relied on?
- Good idea about the 200mA fuse - but an eeprom fuse is much simpler
- Faults possible could be points 1-3 only since by replacing LTC the units worked again.

The unit is a portable wireless Trailer lights. It has an internal Lion battery for operation with a DC socket for charging. It is sealed in a plastic housing so cant transfer the heat by conduction. Also cant add venting holes since it is splash proof. In the lab the unit enclosure does get a little hot on the outside during charging but not too much and did not have any failures after repeated tests. However test were carried out at 25deg ambient temp not in the sun.

Now the defective units had a shorted out battery which couldnt be recovered back with proper charger.
Can it be that the battery went bust with overtemp or was it shorted out due to the blown IC?

 

[treez]

It is sealed in a plastic housing so cant transfer the heat by conduction. Also cant add venting holes since it is splash proof.

..this gives me flashbacks to when I used to work in the automotive flashing (strobing) beacons industry. The heat developed in a sealed transparent enclosure out in the sunlight is immense. It resulted in loads of failures for us. And as you well know, in USA you have places like Death Valley and Arizona, the hottest places on earth.

When I first arrived at the company, there was a large storeroom, and its floor was completely covered with flashing beacons, all flashing away, for weeks on end…when I asked the quality guy what was going on, he said they kept failing soon after they were put on top of a vehicle but never failed whilst in the storeroom, and they couldn’t work out why. I think we both know why, it was entrapped heat due to the sunlight.

So yes, it does sound like a heat issue. Perhaps even the LTC3650 is melting its solder and sliding slightly and shorting its centre pad with the pin pads on the pcb….just look at the chip, there is barely any clearance at all between the centre pad and the side pads.

Also, the internal transistor may be blowing up from heat.

Also, that doesn’t sound good for the lithium battery to be inside that searingly hot enclosure…..you say the lithium batteries have shorted…well that sounds like heat damage. Because I think lithiums are not to be operated in ambient above about 45degC….So inside of your enclosure would be way too hot for a lithium.

Then driving into a shorted load would likely blow the LTC3650, as the buck inductor current would just staircase up to a very high level.

I would seriously do the test with a temp sensor in the unit and see what temperature it gets to in there.

- - - Updated - - -

The LTC3650 clearance from pad to pad is 0.22mm, and the clearance from centre pad to outer pads is 0.275mm……..i wouldn’t mind betting that your PCB manufacturer didn’t even manage to get solder resist layed between those pads….so there is possibility for a bit of moisture etc to short between them,,…..and as you know, electronic components often have moisture in them, and often need to be pre-baked to drive out the moisture……..and in your non ventilated enclosure, that moisture can sit nicely between the pads of your ltc3650 and short them together, especially since there is probably no solder resist there to stop that from happening.

I would say that those style of tiny chip packages are not for use where such contamination may occur. Use a decent SO8 package with a decent gapping between pads.

Also, the ltc3650 has overtemp for 125degc, but the part is only rated up to 85degc…what you could do is try and find a chip which is rated to 125degc.

Also, there is a thermistor to stop charging when the lithiums go above 45degc, but of course, in your enclosure in the sunlight, those lithiums may already have miles above 45degc and blown up.

So anyway, did you do the test of driving a ltc3650 cct into a short and see how long it lasts?

 

[mtwieg]

You need to sort out whether the failures were due to thermal stress or something else. Have you tried verifying that the thermal protection functions? Try operating the unit at a high ambient temperature and see if it fails.

Assuming that the thermal protection is working, I'd assume the failure is due to line transients surpassing the 32V limit of the IC.

 

[treez]

Also, a lithium battery caged up inside a transparent enclosure in the sunshine, with presumably a led load heating away inside it, sounds like a mega bad idea.....mega mega bad

 

[Moonwalker]

What are you saying treez is that we should not use Lion batteries in the sun now?
The heating issue is only during charging which only lasts 2-3hrs. This should normally be done during the night but ofcourse the client can do what he wants.
The enclosure is not fully transparent and it should be facing sideways during use and not upwards. The only concern is that the client might put it in the cabin and plug in charger and leave it exposed to the sun.

I will do some testing on the thermal protection and let you know.

update - One unit was confirmed to have been plugged in a 24V socket in a truck. So if the engine was ON it might explain the failure. But the other defective unit client said he only gave it 12V so that is still a mystery.

 

[treez]

have you put a scope on sample and hold and sudenly plugged your unit into a 28VDC power supply?.....because you might notice big overvoltage ringing in the input LC filter, which overvoltages the chip.

Yes, thats it, charging a battery with the thing in the cab in bright sunlight. That coudl blow the lithiums, did you put the NTC pin connected to the lithium, so you can sense batt temp whilst charging?

 

[mtwieg]

update - One unit was confirmed to have been plugged in a 24V socket in a truck. So if the engine was ON it might explain the failure. But the other defective unit client said he only gave it 12V so that is still a mystery.

Was it 12V from a car/truck? If so then transients up to 80V should still be expected.