MK22 Temporary Internal Short?

[myke_predko]

I'm working with some prototype boards and occasionally, the MK22 on them will get very hot and they cannot be reset or programmed until power is removed for some time.  

We're respinning the board because the selected motor driver won't be available until late next year and we're testing the TMC2225 in its place.  The changes require a new motor boost (I've added an LTC3114-1 eval board for the testing because 24V is required for the driver and 12V was designed into the product) along with power to external TMC2225-BOB boards and some fly wires for the connections.  

The base product is rock solid and we have a dozen of the same boards (with the original motor driver and 12V supply) that have been working for months without any issues.  Could the fly wires (which are only terminated at the MK22) be the problem?  

Any ideas what could be the problem?  It happens when the MK22s are programmed with an application and when they are "Erased" using MCUXpresso.  

Thanx!

[bobpaddock]

"Could the fly wires be the problem? "

If it is similar to a problem I ran into years ago, yes.

We solved the problem by putting a transorb, as a terminator, on the far end of the wire.

The scope showed a lot of ringing, which could have a voltage higher than the chip allowed.
That would lock up the chip and turn it into a heater.
Burning fingerprints into chip lids are not a good way to test I've found.

If the wire is long starting thinking transmission line and termination.

[myke_predko]

@bobpaddock 

Thanx for the comment - you describe what I'm seeing here. 

As for it being induced voltages that's kind of what my thinking is but the wires aren't terribly long - from 15cm to 25cm - and there don't seem to be any huge electrical fields nearby.  

The fly wires are temporary, they're going to be plugged into devices very shortly, any simple solutions (setting the pins to outputs, soldering the untethered ends temporarily to GND) as adding a TVS is not going to be trivial.  

[bobpaddock]

In my case it was not induced fields, it was ringing on a clock line that was about a foot long.

Someone thought it was good idea to put a Dallas Semi Tag (packaged battery backed RAM with SPI like bus) on the front panel of the unit.  This was used by the customer to configure the unit.

Adding termination solved the problem.

It took a while to figure out because it didn't happen that often that the Tag would fall.

[myke_predko · Accepted solution]

Hey @bobpaddock & @jingpan 

I relocated the fly wires from one set of UART pins to another and things have been running fine and the problem seems to have gone away.  This involved quite a bit of moving other functionality around but it's a prototype board so nothing is really set in stone.  

The fly wire length went from 170cm to 100cm and they no longer run over the space taken by a boost supply on the other side of the board (which I guess was the problem, but I thought I was okay with a large ground plane in between).  

I tried terminating the lines with 100k resistors to GND (also a fly wire), but that didn't fix the problem - changing the pins used for the function so that the wires could be picked up on another part of the board seemed to do the trick.  

Lesson learned: keep unterminated wires away from a switch mode power supply on the other side of the board.  

[jingpan]

Hi @myke_predko ,

Maybe high voltage attack some pins and cause latch-up problem?

Regards,

Jing

[myke_predko]

@jingpan 

That's my thought but I can't see any occasions where it is happening.  

That's why I'm asking about the fly wires and could there be situations where they are picking up induced voltages.  

Does it make a difference if the chip is programmed (and running) or not?  When the chip is programmed various pins (including the fly wires) are sent to input/outputs where as in an unprogrammed state they are undetermined?