There was a rumor floating around that the IMI EPROM programmer passed 
an EPROM that the DIO failed. I'd like some confirmation of this and
sample parts if possible. I spoke to the designer of the IMI yesterday
to see if he's heard of this happening before. He said no, but it has
happened where the IMI fails a part that the DIO passes.

To clarify what the IMI does on verfiy, it grounds the OE and CE lines
(enables the chip for DC reads) and reads the EPROM by changing only
the address lines. The DIO wiggles the OE and/or CE lines in-between
each byte read (one of the reasons it is so S-L-O-W). The IMI verifies
the part at 5v only not at 4.75 and 5.25 as the DIO does. The part is
loaded with a resistor (10k I think) and it is unkown how the DIO loads
parts. 

Whichever programmer you use to make master EPROM's, you must check those
parts by inserting them in a working game board before placing your stamp
of approval on them. Neither the DIO nor the IMI will check the EPROM for
correct access time. A working game board performs a more comprehensive
test than either programmer.

So you thought that the EPROM's were static ROMs, well read on...

He told me of a bug that is present in a number of EPROMS, especially the big
ones (27256 and up) where if the OE or CE lines are not changed in an
unspecified period of time (its a bug, remember), all outputs will go to 0 and
stay there, perhaps because the output drivers are more dynamic in nature than
static. In any case, beware (he mentioned AMD as a major culprit). He also
mentioned that the EPROMS are VERY sensitive to the voltage on the Vpp pin. If
it is not exactly what it is spec'd to be, the outputs may oscillate.

ds