How exactly do PAL chips get programmed?

[Tony Duell]

> Presumably there are all sorts of 'standard' formats for storing read-out PAL 
> information, so:
> 
> 1) What's the most sensible format to use, and

The most standard, which should be supported by any PAL programmer and/or 
design software, is called 'Jedec'. 

> 
> 2) Is it possible for a PAL to fail reading in such a way that the data you 
> get out is junk - but not *obviously* junk to the user? (if I go around 

Alas yes. A chip can fail in all sorts of 'interesting' ways.

> backing PALs up I'd like to know for sure that the backups are good - but 
> obviously testing them at backup time by writing a new device for every single 
> one isn't much of a solution! :)

The best way I've found is to turn the jedec file back into the equations 
(there used to be a free (binary-only) MS-DOS program for this provided 
by National Semiconductor which covered most PALs and GALs of the time, I 
think it was called JED2EQN.EXE) and see if the equations make sense. 
Obviously if you get terms with both an input and its inverse ANDed in, 
you start to get suspicious. If you have a scheamtic of the device the 
PAL comes out of, you could see if the equations made sense for the 
signals that go into the PAL 

-tony