TLV3012: internal referance decrease at cold temperature

Hi Doron,

are you sure that your temperature measurement is correct? Temperature measurements at these low temperatures can be difficult. Sometimes the sensor tip is heated by the sensor cable running out of the temperature chamber and seeing the much higher room temperature. The tip might not be totally cooled down to the actual temperature.

Kai

Dear Kai,

Let me share with you a bit of background.

This investigation has started after 7 out of 70 systems has came back from the field reporting an error at -40degC.

After a short investigation Doron has found out that the internal Vref of the device drifts from 1.25V (@ 20degC to -30degC) to 0.6V (@-40degC).

The phenomena occurs only on some of the devices and only at temperatures between -30 and -40degC.

After investigating this a bit we've found another thread on the e2e from last year reporting exactly the same phenomena.

http://e2e.ti.com/support/amplifiers/etc_amplifiers__other_linear/f/18/t/550590?tisearch=e2e-quicksearch&keymatch=tlv3012

This has led us to question whether there is any known issue with the device and if you can advise how to handle this.

Best regards,

Nir.

I have reached out to the failure analysis team to see if they have any information related to this earlier post. The team that addressed this back in 2016 are out on holiday break. I will post my findings as soon as I obtain some information.

In the meantime, would it be possible to do some troubleshooting? It would be nice to see if the problem follows the part. If we could swap a working part with the part that is showing poor performance at cold temp and vice versa, this would rule out any other explanation for the response.

If possible, this experiment would be very helpful prior to submitting parts for failure analysis.

Chuck

Doron and Nir
I looked into previously reported issues with our failure analysis team and we only found instances of delayed startup when the device was powered up at cold temperatures such as -40C. When you ran your tests from 55C to -40C, was the power supply always on or are you cycling the power occasionally? Likewise, previous reports show that the delayed startup at cold only impacts the reference voltage for a few seconds and then it recovers to normal voltage level. From your report, it sounds like the reference voltage never recovers which sounds like a different failure mechanism. At this point, I need to suggest that a couple of the devices be submitted for failure analysis.

In the meantime, if you do not mind doing a little troubleshooting, it would be interesting to know if the problem follows the device and if a known functioning device eliminates the problem. If you could swap a non-functioning device for one that is functioning correctly and vice versa swap the functional device with the non-functioning device, the functioning device should work in the new socket and the non-functioning device could continue to not work properly in the previously working position. This would rule out any board or external component error.
Chuck

Hi

I made a troubleshooting and here the results:
I change the fialure device with a new one and the internal reference voltage was stable at 1.24V at over temperature +55C to -55C.
I took another board and before replacing the device, I checked the reference voltage and it was stable (1.24V)at over temperature .
I change the device at the good board with the failure device and the internal referenc voltage was decrease to 720mV at -40C.
The exactly test is:
I turned ON the board at room temperature (~25C) - the internal reference was 1.24V.
Decreasing the temperature while the board is still powered ON, the internal reference at -30C was 1.24V.
Continue to decreasing the temperature, the internal reference at -35C was 940mV and at -40C was 720mV.
The board was powered all over the test.

This test proves that the problem is in the component itself and not in its environment.

BR
Doron