Re: Blank CMS SCSI Card PCBs

[mjmahon@aol.com (Michael J. Mahon)]

David Wilson replied:

>mjmahon@aol.com (Michael J. Mahon) writes:
>>Such a conversion is by its nature quite ambiguous--revolving around
>>the issue of what canonical "simplified" form is desired.
>
>>In this case, you will want to factor the truth table to match the structure
>>of some programmable logic chip that you want to use to duplicate the
>>function.  Unless the program you have is easily "steerable" in the kind
>>of factorings it prefers, it may be easier to start with another program.
>
>I think SOP (Sun of Products) seems fairly standard. eg:
>
>F(A,B,C,D,E)=A./B.C + /B./C.E + /B.C./E + /B.D.E + C.D./E
>
>The program I have is called BOOZER and was written in 1977. I have it
>printed in the back of William I. Fletcher's "An Engineering Approach to
>Digital Design" (1980). It is 18 pages of Algol 60.
>
>My only worry is a circuit like:
>
>O1 = F(I1..IN)
>O2 = F(IJ..IK,O1)
>
>My scanning program will have to recognise that O1 is an input to O2
>and scan O2 with O1 both asserted and de-asserted.

This is exactly what I referred to as "factoring".  In this case, O1 (whether
it appears outside the package or not) is a useful subexpression, which
greatly simplifies the expression for O2.

Of course, given a combinatorial PAL, exhaustively enumerating the
input vector will guarantee that O1 is both asserted and de-asserted
in the scan.

The trick is recognizing what O1 is!  Hopefully, the factoring into
sub-expressions is not too bad, since the form of expressions that
the PAL can evaluate directly is known from its structure.

I'm sure that PAL manufacturers have software to easily convert a
truth table into fuse/anit-fuse settings--hopefully easily available.


-michael

 Email:  mjmahon@aol.com
 Home page:  http://members.aol.com/MJMahon/