woodelf said:
Ethan Dicks wrote:
If you are designing your own circuits, a 18 or
24-pin GAL can replace
several square inches of TTL, with the benefit of being able to make
changes later that don't involve cutting traces and adding blue wires.
You lose the ability to poke around in the middle of select circuits,
and to make substantial changes, but it all depends on what your goals
for the project are if that's a worthwhile tradeoff or not. You do
have to start with a GAL programmer, though. Not all hobbyists have
them, so if you make a product to sell, as Bob does, you should
consider selling programmed parts for those that can't burn their own.
That is the downside , the programmer for them.
I suspect 18 pin GAL is a typo. They come in 20 and 24(thin) pin packages.
There are also 28 pin versions.
There is a very good design on the internet, namely the GALBLAST of
Manfred Winterhoff (check google).
But although they are handy, IMHO they are meanwhile outdated. I prefer
CPLDs (xilinx 9536 or 72 for instance) for the same purpose even if these
require PLCC44 or 84 pin sockets, because they are even easier to program
through their boundery scan interface, and do no longer need the obscure
programming voltages. GALs are good for any kind of decoding logic, but
are rather weak when it comes to multilevel sequential logic. So I'd say:
while a GAL might replace square inches of TTLs (I'd think about the
graveyard of 7400, 7404, 7402, 7432, 7451, 7454 chips frequently found in
old machines, where often a well-specified 74188 ROM decoder would have
been sufficient), a CPLD will replace some inches of GAL glue logic. Plus
the CPLD usually doesn't suck that much current as the old PAL/GAL
technology.
Regards
--
Holger