Issue: MCU failing taking excessive current

Nigel,

What is your setup for powering the rails (VDDIO, VDD18 & VDD12)?  Are you using internal VREG or external LDO for VDD18 & VDD12 rails?

Are the boards new? Are the devices new?

Regards,

Adam

Adam,

The power scheme is the same as the development board, both the 1.8 and 1.2V are from the internal regulator.

Each VDDIO pin is sourced from the 3.3V via a ferrite bead and has a 2.2uF bypass capacitor.

Each VDD12 pin has a 0.47uF capacitor to 0V.

Each VDD18 pin has a 2.2uF capacitor to 0V.

The boards are new and the devices are RevB.

Regards

Nigel

Nigel,

Is this the first version of your boards or do you have older custom boards in which the devices function correctly?

Have you checked continuity from GND to the VDDIO, VDD12 and VDD18 rails (with a device on the board and without)?  

Do you have the ability on your board to measure the current from your power source to the VDDIO rail?

Regards,

Adam

Adam,

This issue has manifested itself on the second revision of this board. The initial 6 prototype boards did not have this problem. The main changes between the prototypes CCAs and this batch was the addition of a JTAG port and some minor corrections to pull certain pins high or low (all white wire tested on the prototypes). The major component change was from a Rev0 device to a RevB device.

Continuity checks from GND to VDDIO, VDD12 and VDD18 are as follows:

The only way I have of checking with the device out of circuit is on blank pcbs to which all lines are open circuit. And I do not have the ability to separately measure the VDDIO current as all those pins are connected to the 3.3V power layer.

Nigel 

Nigel,

There definitely seems to be an issue with the resistance/lack-there-of on the first five boards in the table.  All measurements that I have taken on a controlCard from GND to rail are on the order of kΩ.  

If you have a microscope, you can look at the soldering of the device to the board to see if any of the pins are shorted together.

The only way that I can think of determining if it is a board issue or a device issue is by desoldering the device from one of the bad boards (listed above), measuring the resistance without the device in, then soldering a new device to the board and measure resistance with the new part.

There are a few possibilities:

1. Device pins could be shorted together due to soldering

2. Could have shorts inside the device

3. Could have shorts on the PCB

Regards,

Adam

Adam,

I am working with Nigel on this issue.  He is out of the country for about a week.

We pulled the micro off one of the bad boards that measured less than 1 ohm from VDDIO  (3.3v) to Ground.  

What we found is that a number of the pins on the micro now measure about 2 ohms to Ground.  Also, the resistance between the 3.3v power to ground has now gone back to around 800-900 ohms, which is the same as the boards with no issues.

Pin    Ohm     MCU         Circuit
---    ---     ---         -------
67      2      VDD10        3.3V
74      2      VDD10        3.3V
91      2      NC           nc
92      2      VDD10        3.3V
93      2      X1           X1
94      2      VSSOSC       nc
117     5      ADC11NB0     nc
118     5      VSSA1        0V
134     5      VDDA2        3.3V
135     5.8    VSSA2        0V

Any idea what could cause some of the power input pins on the micro to suddenly be virtually shorted to ground?

It may be a coincidence, but the X1 input is the only signal input that is now also 2 ohms from ground.  We do provide the X1 oscillator signal before applying the 3.3v to the micro power pins.

Casey

(working with Nigel)

Casey,

Are you saying that the resistance from VDDIO rail to GND on the board without the device is 2 ohms?

If that is the case, then you have a short on your board which could damage the device.

Regards,

Adam

Adam

No, we removed the micro from the board.  The specified pins on the micro itself are showing 2 ohms from some of the VDDIO pins to ground. Once the micro was removed, the board resistance went to back up to about 750 ohms from 3.3v supply to ground.

Basically, the problem followed the micro chip.  Here is a "corrected" list of pins:

Pin    Ohm     MCU         Circuit
---    ---     ---         -------
67      2      VDD10        3.3V
74      2      VDD10        3.3V
92      2      VDD10        3.3V
94      2      VSSOSC       nc
119     5      VDDA1        3.3V
134     5      VDDA2        3.3V

The million dollar question is how did this happen?  Our board functions correctly during many power cycles before this happens on the next power up cycle.  Once it happens, our 3.3v DC-DC converter goes into current limit (around 1 amp) and can only bring the 3.3v line up to about 0.80 vdc because the board 3.3v line is only 0.5 ohms or so above ground.  Remove the micro chip and we see the above.

Thanks,

Casey

Casey,

Unfortunately, we don't know if the device was already shorted before it was soldered to the board.  The damage could have been caused by a faulty board or even by ESD when the devices were being handled.

Would it be possible to solder a new device to one of the boards which a damaged device was removed then go through the same troubleshooting process?  Also check the continuity of the device (which you did above) before placing on the board so you have a reference.

Let me know what you find.

Regards,

Adam

Adam,

The micro chip was fine when it was soldered in and worked for quite a while afterward.  I think we can rule out ESD because another one failed in the same manner in a sealed aluminum enclosure.  It was just fine on power-up and then died on the next power-up.  There was no handling of the box nor the cables between those two power ups.

We have now pulled another micro that died and the same pins are shorted to 2 ohms.  This board functioned correctly for a while.

Do the pins that we found shorted to 2 ohms (3 VDDIO, VDDA1, VDDA2) go to the same location on the die or have something in common as opposed to the other pins that are not damaged?  We are monitoring the pins during power-up and everything looks normal.

There is a lot of head-scratching here.

Casey

Casey,

It has been brought to my attention that applying 3.3V to the X1 pin prior to powering the rails is not good for the device.  The clock should be powered from the same rail as VDDIO.

Regards,

Adam

We have an external oscillator feeding into X1.  X2 is left open. The external oscillator (ASEMPC-20mhz) is powered by the same 3.3v rails as the micro.  However, the oscillator does start before the MCU comes out of reset.

What is really odd is that now we have two MCU's that were working fine and then died sometime later during a random power-up event.  Both MCU's have the same pins that are measuring between 2 - 5 ohms with respect to ground after the event.  We are trying to put new MCU's on the boards but there are a few logistical problems at the moment.

VSSOSC is also 5 ohms to ground but that may be normal.  It is a "no connect" in the circuit.

It appears that these pins have common junctions inside the MCU die.  It also appears that these lines can somehow be shorted without affecting the numerous other VDDIO pins on the MCU.

Pin  Ohm    MCU     Circuit
---  ---    ---     -------
67    2     VDD10   3.3V
74    2     VDD10   3.3V
92    2     VDD10   3.3V
119   5     VDDA1   0.0V
134   5     VDDA2   3.3V

94    2     VSSOSC  nc