Teensy 3.5: Pin 15 only works when I press on the CPU

[thinbits]

I'm having a weird problem with a Teensy 3.5 I recently purchased. Pin 15 only reads when I press on the top of the CPU chip. It (and about 40 other pins I'm reading from) are pulled up to 5v with 18k resistor packs. All of the other pins read as expected (normally high, read low when grounded).

Pin 15, however, seems to stay stuck in whatever state it was last in unless I press down on the top half of the CPU. Removing pressure from the CPU seems to leave it stuck in the last read state. All the other pins seem to read fine.

Is it possible the CPU has a bad solder joint to the board? It's only about 2 months old, been sitting on my bench until my boards from OSHPark arrived last week...

 

[tonton81]

how would the pin have a state if it wasnt connected? if the processor leg was not connected, neither would be a voltage on the pin... i doubt the pcb has what, 60 pullups inside its layers lol

 

[thinbits]

Sorry, I just realized I wasn't 100% clear in my post that I'm reading from the pins, not writing. I've edited the post to make that more clear.

In any event, what I'm observing is what you normally see when reading from an unconnected pin. DigitalRead returns whatever the last state the pin was externally driven to.

I wrote some code to show read states on my ILI9341, and it's obvious that pin only reads correctly when I press on the top of the CPU. Scope probe on that pin shows it's being driven correctly.

 

[GremlinWrangler]

That sounds very much like a bad joint, you may want to get hold of Robin on the customer service email address if she and/or Paul don't pop up here. The read thing will be because the suddenly unconnected input will act as a capacitor and hold the charge for a bit. If you got all CSI on it and found where the surrounding pins on the IC are and tried toggling them you could probably manipulate it capacitivly it to not very useful end.

Looking at the pin while it operates as an output may be interesting to confirm solder vs on die problem, possibly by removing the supply from the resistor pack and running blink (don't touch the buttons or whatever grounds the pullups or it really will be busted)

It might be possibly to very carefully reflow it but would suggest avoiding that move until you hear from PJRC, they way want to autopsy it and/or have better advice than random dude on the internet.