UCC28701: Abnormal behaviour on NTC pin in UCC28701

Hi Ghiotto.

The expert of this device is on time bank and he will return back tomorrow. Thanks for your patience

Hi Manuel,

thank you for reply, some other news about this topic? This is urgent since we are producing....I look forward to heraing from you soon.

BR

Hi Roberto,

Sorry for the delay, I just reached to our expert and we will be back to you soon.

Thank you.

HI Roberto,

Sorry for the late reply, I was on leave last week.

The tolerance is not reported in the datasheet for both the pull up current and the NTC fault trigger threshold and I have checked this internally as well. The variation in both these parameters remains more/ less constant across temperature as well. Can you please remove the 4k7 in series with the NTC and test? We have successfully used the 100k NTC part (esp. NTCS0805E3104FXT) in our designs and it shuts down appropriately at the threshold value.

Please let us know if you have questions.

Thank you

Regards,

Harish

Hi Mr Harish,

thank you for suggestion, now we will try to bypass the 4k7 resistor, so connecting directly the 100K NTC to pin1 to UCC28701.

But we need the 4K7 resistor in series because otherwise there is an early thermal cut-off between 75°C and 80°C, while we need a cut-off between 100°C and 125°C.

Anyway, in your opinion, why sholud we have compenents that behave differently (i.e. different voltage thresholds) at the same load and ambient conditions? Could be due the total impedance connected to this pin? Is it possible to lower the total impedance, for example, connecting a 9K0 + 10K NTC?

Thank you for replying,

BR 

Roberto G.

Hi Roberto,

Thank you for the feedback.

sure, let us know after bypassing the resistor.

Have you tried swapping the controller/ NTC and related components where the board behaves as expected vs an early shutdown case? In that case we can try to see if the batch/ lot number of the parts and ask our quality/ product team for their feedback.

Regards,

Harish

Hi Harish,

thank you for reply, my collegue has done the load test without the 4K7 resistor: in this case the Voltage Threshold increases up to 1.4V, see plot below (from previous 1.28V).! So I think there is no interference from the resistor 4K7 in series with NTC.

As regard the swapping: yes, it was the first step after finding the problem, we put the controller in another card where the controller behaviour was correct and we found that with the new "defective" controller the threshold was higher, so it do not depends on the control board but it depends on the controller and that's because I requested your support. 

Please find attached the two data codes from our two different suppliers (in both reels we have found controller with different NTC thresholds).

Please let me know what do you think or suggestion for this topic.

Thanks, BR

Roberto G.

Hi Roberto,

Thank you for the feedback. 

I will check with the internal team on the cause for variation in threshold. We will take a couple of days to review and revert back. I will keep the post open till then.

Regards,

Harish

Hi Roberto,

I have checked with the team internally and they have checked Lot #3195871 and Lot #2359879, they were tested under the same program, same revision, same test limits. So having shut down under 1.28V is not normal. Even the threshold distribution numbers do not show more than 3% difference. Can you clarify on how many units (quantity) show this difference?

Thank you

Regards,

Harish

Hi Harish,

we found the problem when we tested some samples coming from a new supplier. Here we found a defective component. Putting this controller to another electronic card taken out from production (assembled by a different supplier), again it displays the problem, while the controller taken out from the production cards and soldered in the previous electronic card (coming from the new supplier) went well (==no problem, the NTC threshold was ok). So it is quite clear that the problem is on the controller, i.e., the problem is not depending from supplier since the "defective controller" behaves in the same manner in the two different electornic cards.

After, we tested some electronic cards from the production ( nr. 3 electronic cards from nr. 3 different  production lots, the supplier is not the new one, but it is an consolidated one) and, above nr. 9 electronic cards, we found another defective controller, with the threshold of 1.28V. For this reason we are worry and that's the reason why we're investigating in. I gave you the two production number of the reels related to defective controllers (coming from the new and old supplier).

The NTC threshold is tested on 100% of production? Is it possible to have a test report? Could be the assembling process critical? Why should we have variation of 30% on this thersold? Do you have suggestion of other tests to do in order to understand and solve this problem?

Thank you

Roberto G.

Hi Roberto,

This parameter is tested in production.

I will check with the team to see if there is any report.

Thank you

Regards,

Harish

Hi Roberto,

I asked internally and it turns out that they are not aware of any issues like this in the past. Their assumption is 30% variation on this threshold, it’s most likely the assembly process causing the threshold to shift. 

To debug the root cause, you please reach out to our field team or submit a customer return case for further investigation.

https://www.ti.com/support-quality/additional-information/customer-returns.html

Thank you

Regards,

harish

Hi Harish,

then, our supplier buyed the bobbins from Moser (UCC28701DBVT and UCC28701DBVR): have we to call Moser for this CPR? Is there a standard procedure? If possible, could we deliver directly to TI?

And what we prepare? Single defective controllers or the whole elecronic card?

Thank you,

BR 

Roberto G.

Hi Roberto,

I think you need to find the CPR form from the link which I sent before. This will be going to the quality / product engineering team who will investigate potential causes.

Please reach out to your field team of TI and they should be able to help with this. I think it should be controllers (good and bad) as they are not going to power the entire board.

Thank you

Regards,

Harish