I have read it, it's on my bookshelf. PALs were still just
starting to be seriously used in the very late 70s with an
increase into the 80s, it was the cost vs board real estate
vs reliability issue. The testability/reliability issues took
a few years to gain engineers confidence.
Allison
-----Original Message-----
From: Pete Turnbull <pete_at_dunnington.u-net.com>
To: classiccmp_at_classiccmp.org <classiccmp_at_classiccmp.org>
Date: Tuesday, December 25, 2001 9:05 PM
Subject: Re: PALs (was Re: MITS 2SIO serial chip?)
>On Dec 25, 12:22, Ethan Dicks wrote:
>> --- ajp166 <ajp166_at_bellatlantic.net> wrote:
>> > -----Original Message-----
>> > From: Ben Franchuk <bfranchuk_at_jetnet.ab.ca>
>> > >> Sure its trivial to do now but we were talking 1981 when PALS
were
>> > >> expensive.
>> >
>> It depends on what your limited resource is and how costly _that_ is.
>> ISTR PALs were $2-$10 each c. 1981-1982. If someone remembers
>differently,
>> please supply more accurage data. Yes, a single PAL cost more than
>several
>> TTL popcorn chips, but given how much you could fold into a single PAL
in
>> the way of I/O select logic (a common use on the board I first
>encountered
>> PALs on), it wasn't so expensive then.
>>
>> > Yes, they were compared to random logic, but if board space was
costly
>> > they were cheaper.
>>
>> Or if board space was a fixed resource, then it's doubly costly. The
>> oldest example I can cite from personal experience is the COMBOARD-I
>> to COMBOARD-II design. Both were early examples of 68000 designs (an
>> intellegent Unibus serial periperal)... one with SRAM (2114s) and TTL
>> logic, the other with DRAMs and PALs. One similarity - about 1.5 sq
ft.
>> of board space to cram in a CPU, RAM, support logic, a sync serial
port
>> (based on the COM5025) a parallel port and Unibus DMA logic.
>>
>> In 1981, our designers went with older tech, but the DMA engine was
>> horribly complicated from the standpoint of the 68000, but it was in
part
>> because there wasn't enough room on the board for a better design with
>> TTL. The second revision (c. 1983-1984) had more onboard I/O, and
>> implemented the DMA engine as a bank of shared memory to the 68000
(i.e.,
>> read/write to a certain range of memory from your code and it
>automatically
>> generates a Unibus DMA cycle - most cool). I don't know for a fact,
but
>> from what I remember about the designs, I don't think a shared-memory
DMA
>> engine would have fit on a Unibus board if it had been made out of
>popcorn
>> logic.
>>
>> > >I never heard about pal's until about 1990...
>> >
>> > PALS are 1970s technology, really old to some of us.
>>
>> I didn't see PALs in use until the early 1980s. Yes, I know they came
>> out in 1978, but products designed with them didn't hit right away.
>
>You should read Tracy Kidder's book, The Soul of a New Machine. It
>describes the design of the Eagle inside Data General from early 1978 to
>early 1980, and mentions PALs a lot.
>
>--
>Pete Peter Turnbull
> Network Manager
> University of York
Received on Tue Dec 25 2001 - 21:36:39 GMT