> I've been tossing around the idea of building my own GAL programmer for
> 16V8s and 20V8s, BUT!  It seems that the people who make them (NS & SGS
> Thompson) are keeping the programming specs a secret (lots o' money maybe).
> I've found bits of info but no where near enough to actually do anything
> other than increase curiosity and frustration.

I agree that this situation can be at times frustrating; the problem, though,
oddly enough is not that the semiconductor manufacturers are scared about
other companies stealing their code, but that they're concerned about either:

[1] Having lots of hobbyists taking up the time of their engineers; it's
    not that the companies are trying to be mean to hobbyists, but having
    a $40/hour engineer spend half an hour on each hobbyist who may buy
    perhaps $50 (retail) worth of parts is, frankly, not cost effective.

[2] Devices which are programmed improperly may fail in use, perhaps after
    many hours of service.  Chip vendors do not want any headaches or bad
    reputations that could come from such device failures.  Compared to
    microcontrollers and xxROM chips, many PLD's are downright touchy in
    their programming specifications.

[3] Manufacturers sometimes change the programming requirements for GALs
    and PLD's slightly, even without changing the part number.  In such
    cases, the variable details of the specification will usually be coded
    within the device and readable by the programmer.  The flexibility to
    change the specifications allows manufacturers to make improvements in
    their manufacturing process, but a hobbyist's programmer which, e.g.,
    uses a 4ms programming pulse because that's what lot#99223 required
    may slag chips if it doesn't read the required pulse duration from
    each device.

[4] Chip manufacturers who do make significant changes to programming requ-
    irements (even ones they've documented in advance that they MIGHT make)
    will notify programmer manufacturers so that they can ensure that the
    programmers in fact continue to work with the new devices.  If there are
    20 certified programmer vendors, this is practical.  If there are 500
    vendors selling development programmers, this becomes a nightmare.

[5] For most programmable logic families, there are devices available from
    selected manufacturers which are, in fact, easily programmable by well-
    documented means.  The 22V10isp from Lattice is an excellent example.
    Such chips cost more, because manufacturers are either limitted in the
    process improvements they can pursue, or else because they have to add
    more circuitry to the chip to aid in programming.  On the other hand,
    for small hobbyists, dollar or two per chip is often a non-issue.

Note that some of these issues occur in other types of devices entirely.
For example, a company of a popular PC peripheral made hardware specifica-
tions available for free to anyone who agreed to the following (from mem-
ory):

[1] The recipient could use the information in his/her own compiled soft-
    ware, and could release such software, but could not release the soft-
    ware in source-code form or otherwise reveal how the hardware was
    programmed.

[2] The recipient would not bother the company's tech-support people for
    any assistance regarding the information provided.

[3] The vendor was under no obligation to continue to have their hardware
    work the same way, and that he/she would add a disclaimer to any rel-
    eased software to that effect.

If a PC hardware vendor puts on those restrictions, it's reasonable to see
why a GAL vendor would be equally paranoid...