Jeff Smith

letting go of phantom SCI problem

Discussion created by Jeff Smith on May 30, 2006
Latest reply on Aug 28, 2006 by Jeff Smith
Some people might have been trying to follow a thread where I was questioning if SCI errata MUCts00510 was really fixed in maskset 2K79X. Sorry if it confuses others that the thread was deleted (it did me). They mention (unofficially) that it was deleted more for a moderators incorrect response than for my statements. I admit though that I was speaking too quickly, and they were not trying to ignore me.

I think the trail has gone cold for me. My notes aren't good enough. I finally found another MCU of same maskset but was not able to recreate the exact problem I remembered. There are two prob that have software solutions. I miss some chars on SCI1 input, probably because SCI0 output interrupts take too long. That might be improved just using HPRIO in this case. I realize now that HPRIO helps not only when the two interrupts occur within one instruction operation (less likely), but when multiple interrupts are pending because MCU is already servicing another interrupt.

As for the actual time when my interrupts were left pending but the ISR was not called, I don't remember the details. I was doing extensive breakpoints and stepping. Maybe active BDM increases chance of the phantom SCI problem occuring. Maybe I didn't notice the I mask bit got set. Maybe a combination of things I tried solving at different times but not all at once.

Go figure. -- official phrase of the engineer --