LMR14020: Chip destroyed from the field

PS: LMR14020 on 5V and 12V never damaged

Hi, Daniele,

Sorry that you facing this kind of damage case. I didn't see much issue on you SCH， the input cap your mention is 100nf+22uF, 

Could you have some waveform to show the input voltage, and the 3 set of LMR14020 is connect in parallel , right , the other 2 set works well, did you try to disconnect that 2  set and only apply 3.3V rail. Not sure if there is a current from output back to Vin ,or other trace .

Elena

Hi Elena Gao, thanks for your relpy.

From the field as input supply there are:

 - SMCJ33A

 - diode for inversion protection

 - 220uF (electrolitic) + 5x 1uF X7R + 1x 100nF X7R

 - 1x WE 74402800015

 - 220uF (electrolitic) + 5x 1uF X7R + 1x 100nF X7R

 - 100nF X7R + 22uF mounted directly on the pin 2-7 of LMR14020

The 3,3V is the first that start, when the 3,3V is on, then 5V and 12V are enabled.

Please specify the point where you like to have a measure, I can do it for you.

Thake in mind that actually I don't know how to reproduce the damage.

Hi, Daniele,

Usually the damage we met is EOS, like input voltage spike or some pollution on board may short the pin. Confirm the input voltage is no overshoot, I saw it already have large Caps and TVS in power stage. Any ABA test to change good IC or board.

Did this chip coming from TI store or agency. if this is really quality issue we can support FA .

Elena

Hi Elena Gao,

Each board was repaired by replacing LMR14020, and all the logic under 3,3V (usually the microprocessor die in short circuit probably for an overvoltage, so we need at least to remove it before to have the power stage working again).

Usually we buy TI components from distributors (italian market).

My dubit is about the fact that in broken cases, only 3,3V output die while 5V and 12V output are allways good. Same batch. All in parallel at the same power input.
Is possible that 3,3V output with 32V (max) input is too large voltage span that LMR suffer?
Can the internal cycle-by-cycle current limit fail in some conditions?

Thanks

Hi, Daniele,

To help you solve this issue, could you provide your tracking of this purchase, like some label or better a top mark of IC. we can try to search the production information .

Others like ,please check the waveform of input side, the slew rate and max voltage,

Do you think PCB need to review, if yes, you can share your layout here for us to check.

Elena

I have still some pcs of the purchasing of 2017 (when production start). (the same component on the board is uneadable because we mount caps over it)

Here some measure on pin 2 for DC input, and pin 8 for switching stage (following measure are without additional caps over the LMR to be accessible with the scope):
@24V (nominal)

@32V (max)

@16V (min)

Actually we had just remastered the board to introduce definitively the X7R and 22uF near the 3,3V generator as input filter.
The first thing is to determine what can open the plastic case of LMR.

Thanks

Hi, Daniele,

One more waveform need ,Vin power up slew rate, that is Vin from 0V to 24V. 

This deice have may enter to uncertain condition if power up slow @ Vin UVLO, like 2V to 4V range, but we fix this issue in the next generation .

I'm asking your top mark is try to find the production year ,and as you mention is 2017, so I wonder you have on hand is not the latest version.

Elena

Powerup: In this moment I can only record conditions on a test bench. In real life (on the veicle) the powerup is a clean contact that insert the 24V battery. So I suppose that there is not a problem on this side. Anyway I can investigate on a veicle about the poweron condition: which event I have to trigger? Low voltage slow growing (2-4V) and some other events?

There is any dangerous situation in the opposite side: too high slew rate on poweron?

Versions: Interesting. How can I identify the versions? How to be sure to order the last one? Can I have a changelog of them?

regards

Hi, Daniele,

I can only think out some condition may damage the IC, we have some experience abnormal Vout occur when Vin ramp up very slow, even hold some time near UVLO, for example some point between2V to 4V, then the Vout have chance to be abnormal, like not a stable 3.3Voutput. Maybe big input cap on it.

Others I didn't see much  possibility of over voltage with a constant 24V supply. 

Also, the top plastic package is  fall off big hole, I am not familiar with this kind of material failure. so I want  check the device  top mark, it take some time for TE to verify, and need our PE to check ,I will update to you if they have some founding.

One suggestion is ,do you consider raise up a FA as customer quality return?

Elena

I actually had a lot of faith in the reliability of Texas and didn't think about production problems, so I didn't open any complaints. Since 2017 maybe it's too late..

Elena Gao I await your news, if the lot or version of my components (which I still have in stock) could be the cause, I will certainly move to get new and more recent ones.

Meanwhile I'll look for a measure campaign on a veicle to record the real conditions of poweron phase.

Thanks for now

Sure, glad to help you .I also want to figure out what is the real cause. And we are planning new technology on this old product, or new product with pin compatible,  Maybe 1 years later. :)  

If you have some founding or more information want to share, you can also directly to me elena-gao@ti.com 

Thanks for your support.
See you soon

Leave me some time and I will reply here. Thank you.

Hi, Daniele,

Thank you for waiting .I have get the feedback from production team, we can trakc your part information is system ,which show the date is Y2016. This is not our least version die version. But we could not 100% confirm it's route cause, internal circuit, or package material issue after a long time .My suggestion is  change to new production device, and notice to use the new product we plan to release with new technology. 

Wish this information could help you. 

Elena

Thanks Elena Gao,

    can you tell me how to recognize the old and new versions by the marking code?

Can I have access to some samples to test? Actually if I go on the sample request page seem that my profile is not enabled

My information is from internal production record, the old version should not selling on ti.com now, One easy way is looking the production date like 2020, it should be the new version.

Do you compony have a FAE , or sales or marketing person to support , Usually TI FAE can help their custom to request free sample, but I don't know your region and contact,  this kind of flow is out of E2E technical support, sorry I can't help it.

Sorry, but I'm unable to read the production date on the marking code, can you explain me?