TPS61032: Intermittent Failures - Cannot determine why

Hi Jonathan,

May I know the failure rate and the maximum load current? and please share the layout. 

For these 10 failure part, have you use DMM to check the Vout, SW pin to GND resistance? If Vout, SW pins are not short to ground, solder the IC on a good board to see if the Vout is 5V. 

Zack,

Thank you so much for the reply. Regulator drives a transceiver.  It is intended to supply 50mA nominal with 1.1 Amp pulse that is 10mS wide.  There is 5 second minimum rest between high current loads. The failure rate is 16%  (6 out of 60 boards) so far but it is a new design so failures could occur going forward.  SW to GND is over 500 ohms and Vout to GND is over 1K. On all (6) failures changing the TPS61032 resolves the problem and the unit starts working again.

I observed waveform on SW pin. 5V Nominal with 17V ringing. Data sheet says 7V max. Are these peaks permitted? I changed R15 from 10 Ohm to 0 ohm. Added 10uF X7R Cap from pin 6 to pin 5 directly on the IC Pins. No change on SW waveform. I added 1000pF / 5 Ohm series snubber on SW pin. Peaks reduced to 8 Volts which is still above 7V max permitted voltage. Increasing snubber further starts degrading switch fall time when under load.

Hi Jonathan,

The failure rate is pretty high.

I want to check two things with you:

1. When the failures happened, have you checked the Vin, SW, Vout pin waveforms? What if you remove the load?

2. "On all (6) failures changing the TPS61032 resolves the problem and the unit starts working again." Do you mean you change with new IC then the boards work again? Or you change the IC on failed board to a new board then the Vout is good again.

And please share me the layout, not the fabrication hardware picture. I want to check whether the issue is related to PCB layout.

The layout and measuring method has a big impact on scope waveform. Extra noise could be mistakenly acquired in a measurement if the measurement is taken in the wrong location or with wrong technique. Simply remove the hat from the probe tip and wrap a wire around the probe barrel for the ground connection. This minimizes the inductance by reducing the length of the ground return loop.

1. After IC has failed SW signal looks good and rises to 5V linearly without ringing. Vout signal has nothing. 2. De-solder/Remove IC from board and replace/re-solder with brand new IC only in existing board. 3. I sent you “friend request” and in invitation I provided link to PADS .pcb file for full layout. 4. I will double check tomorrow. Do you mean wrap wire on probe barrel and then ground as close to signal as possible? There is large exposed ground available at the IC .

Hi Jonathan,

I have accepted the friend request. 

Can you re-solder the failed IC from existing board to a new board? 

Here is a link to show you how to measure SW waveform: http://www.ti.com/lit/an/slva494a/slva494a.pdf