Long-ish post, so here's a summary:  how I'm self-testing a PIC circuit now, 
how else I'll be self-testing, and what/how else can I do more self-testing?

For a specific circuit that's in production now, I have a PIC16F913 connected 
to some discrete and 7-seg LEDs.  Considering the most common failures I've 
had after assembly (solder bridges, bad joints, etc), I've modified the code 
to run self-tests to quickly identify some of those issues -- IE: on initial 
power-up, the PIC code sequences thru all LED segments individually, then 
cumulatively in reverse order.  This has worked very well, as specific 
failure patterns on the display can quickly tell me where the problem should 
be -- a row, column, blown LED, etc.  After the self-test, a value is changed 
in EEPROM to tell the code not to run those tests ever again.

In the next version (very soon), I will have all V+ and Gnd wires on a panel 
(of 10 units) connected together, so I can run the self-test on all 10 units 
simultaneously.

But I want more ... This circuit also has 2 analog inputs, and an external (to 
the PIC) voltage reference.

I believe I can test the voltage reference by switching that I/O to be a 
regular analog input and sampling that value vs. the supply voltage, then 
verifying that it's in a certain range.

For the analog inputs, I'm going to setup the panelization so that all inputs 
are connected together and a single voltage applied to the panel will supply 
all the analog inputs during testing.  For simplicity, I think I should be 
able to put a 2-resistor voltage divider on each panel (ie: one voltage per 
panel of 10 circuits) to supply all the analog inputs.  Or maybe a 3-resistor 
voltage divider to supply the 2 analog inputs per circuit with different 
voltages.  This would ensure shorts between the 2 analog pins can be 
identified.

I also have a tact-switch on each circuit, and to test that I'll display a 
prompt for the operator (me for now) to press the switch to continue.

Now the problems... With the visual tests for the LED segments, the operator 
has to watch the LED's carefully.  It takes no more than a minute to run the 
full sequence, but it adds up to a bunch of time when 500+ of these are 
assembled.  And if I do run the self-test on a full panel of 10 at once, I 
would need to slow down the test-display sequence so the operator can see it.  
What I'm wondering is if there is some way to have the PIC check itself for 
shorts/opens.  Shorts should be simple -- the code can toggle the RB3 output 
for example, and check RB2 and RB4 to ensure that they don't follow RB3.  But 
checking for shorts at other points on the board (where pins are not 
necessarily adjacent to each other, but may run to PCB traces very close to 
each other) may be tedious.  Is there any clever way to check for opens?  Is 
there any clever way to check if the LED's actually light up (by checking 
their voltage drop), but keeping in mind that I don't have any spare I/O's.  
I doubt, but I'll ask anyway since I still periodically get fascinated by 
some trick I see on here. :-)  If there is some way, then the operator won't 
need to sit thru the self-test, but just wait a few seconds until they see 
"PASS" on the displays on a panel.

Cheers,
-Neil.






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