TK1 AVDDIO_PEX and DVDDIO_PEX short to GND

Autonomous Machines Jetson & Embedded Systems Jetson TK1
1

We found that some boards resistant from +1.05v_run to GND are abnormal.Normal boards should be about 1kohm. but we got 4 pcs of boards are 1-2 ohm, and 2 pcs of boards are about 50-60 ohm.

after checking, the abnormal pins are TK1 AVDDIO_PEX and DVDDIO_PEX pins. and those boards can boot up.
So my questions how make it happened? and can i still use those boards or replace TK1 IC?

2

Hi hujun, can you provide more details of your tests? Did you discharge the rail fully before test? Did you isolate the AVDDO_PEX and DVDDIO_PEX pins from other pins on net +1.05V_RUN?

3

Hi Trumany,

we have an automatic tester in factory after SMD. And we set some test pads on PCB for some of voltages which include +1.05v_RUN.At the test beginning,the tester tests the resistance between those test pads and GND (without power). And it swaps the test pad by electric relays, and it measures the resistance by a multi-meter. After resistance testing, the tester power the board, and test the voltages.

And in our schematics, +1.05v_RUN net are directly connected to following parts:
TK1 VDDIO_SATA (AK30)
TK1 AVDD_SATA_PLL (AH31)
TK1 AVDD_PEX_PLL (AC21)
TK1 AVDDIO_PEX
TK1 DVDDIO_PEX
Anyway, i try to isolate those nets. finally i found the short pins are AVDDIO_PEX and DVDDIO_PEX on TK1. and then i removed TK1 from PCB board, and check the pins (solder balls) on TK1. And confirmed abnormal TK1 resistance on AVDDIO_PEX and DVDDIO_PEX pins.

So I would like to know:

  1. dose it real fail when the resistance about 40-60 ohm?
  2. if it is real failure, how can i avoid this?
4

Seems we are talking about your custom board, right? And you checked the pin balls of TK1 chip (removed from board), the resistance is abnormal, right? If so, it is related to chip not board, the 40 - 60 ohm seems like the value of impedance matching of PCIe/USB. Not sure what cause it since it is once powered up before removed from board. Can you check the left new chip which is not mounted to board yet to check if any one has such issue?

5

Hi Trumany,

You are right. we are talking about our custom board. And it is related chip not board.
the value of new chip is more than 1kohm…

so far, i guess our tester made shorting +1.05v_RUN to another voltage to break those board during the electric relays switching. but i just guess, because i am not easy to get evidence.

so i am not sure if this case is recoverable by software? or physical damaged?

6

It should be physical damage, SW can not change it. Looks like what you guess, it is caused by some shorting that might due to the test or the too much solder paste.

7

Ok. Thanks.