At 22.49 2005.02.23 +1300, you wrote:
>> I fully agree that the proper way to deal with unused pins is to set
>> them as output and leave them unconnected. If something else than
>> the program changes the TRISx registers, then I think that we've got
>> a much more serious problem anyway..
>
>The trouble is that the level of sophistication built into a product is
>probably going to be optimal, perhaps for economic or practical
>reasons
>
>If someone happens to place one of your control boxes right next to
>a radio transmitter or a noisy relay bank, which you might not have
>foreseen them doing, and it goes ape, do you then have to think about
>putting "Do not put next to transmitters or relay banks" stickers on ?
>Or blow the budget by changing from a plastic to a metal box ? Or
>add things like tying resistors so unwanted external noise can't make
>it go ape ?
>
>Obviously every product is different in its design, cost restrictions and
>how important it is for it not to go wrong and the consequences if it
>does, but it pays to at least consider risk management. Both of the above
>instances happened to my products. Recalling units and adding literally
>a couple of cents worth of resistors and a bit of aluminium foil made
>them so much less susceptible to future problems and, touch wood,
>they've all behaved themselves for some time now
>
>To quote a list member's signature, which I keep in the back of my mind
>with everything I do,
>
>"There's never enough time to do it right - there's always enough time
>to do it again"

Ok, but specifically, why should this RFI only change TRISx and not a
GPR with even more catastrophic consequences?

That's why I say that unused pins can be left unconnected and set as
outputs.. because if TRISx changes, chances are that a lot of even more
"delicate" GPR's have changed already.

Metaphorically, what's the real usefulness of a parachute in a submarine
whose only serious risk is to sink? Sure, the parachute will help in case
an UFO picks up the submarine and then drops it from 10000 ft altitude..
but I think that adding a resistor to every unused pin is almost as
useless as above..

On a 18F8720 we have almost 4K of GPR's that may be changed by EMI/RFI,
and only a few of TRISx registers. I would concentrate my efforts for
reliability elsewhere than on the lone TRISx..

And about the firmware/software, well.. while I don't think to produce
bugless code all the time, I'm extremely confident that I won't write
a routine that changes TRISx improperly for no reason..

One thing I've always believe in, regardless of TRISx issues, is to
debug code also *before* you notice any bug.

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist