Hi people,

Long rundown on specific device latch up condition.

Just a thought for those going to try and use the 40 Pin PICs for the
first time.

The C64 (and others) have a parallel slave port on port D using
port E as control signals.

Well I used port E as control signals (outputs) for a alpha
numeric LCD display and used port D to read in a simple DIP switch
(the LCD data was on port B).  All this is great.

The problem comes up when you don't read the data sheet and check
out what the remaining TRISE bits do.  I am used to setting
unimplemented bits in a port into the input state (bits in TRIS set)
and using the low 3 implemented bits of port E as output I set the
rest high and used 0xf8 for the TRIS value.  This causes some realy
hard to detect problems.  The data sheet says one should program the
port E pins to input if using the parallel slave mode which is
selected from the same TRIS register.  I had inadvertently turned
on the parallel slave mode and set the pins to outputs.  Usually there
was no problem but if there was a intemediate power brown out or
noise on the supply when coming up it was likely that port D would
get stuck in some kind of wierd mode between normal and parallel
slave mode and not read properly.  Using a hard reset did not
clear the condition and the usual drop the voltage to 4V so the reset
generator kicked in was also not sufficient and I had to wait for the
supply to go to ground before this funny mode was cleared.  The only
reason I noticed this was because I was writing the dip switch value
onto the LCD when debugging which was working fine but the value
read from the switches was badly corrupt.

I understand that I did put the device into a mode that was not
allowed but even so when it occasionally did latch up port D the
condition was not cleared by a firm reset.

It is starting to look like there is a pattern with quite a few
strange modes and periferal latchups that are impossible to clear
without a deep power down cycle.  This is very hard for a program
to detect or remedy, having a watchdog driven powersupply crowbar
circuit does seem a bit extreme.

Cheers
--
Kalle Pihlajasaari   kalle@ip.co.za   http://www.ip.co.za/ip
Interface Products   P O Box 15775, DOORNFONTEIN, 2028, South Africa
+ 27 (11) 402-7750   Fax: 402-7751    http://www.ip.co.za/people/kalle

DonTronics, Silicon Studio and Wirz Electronics uP Product Dealer