I'm repairing a board which requires replacement of an 82S129/74S287
PROM. I found a NTE 74S287 part at Fry's, which I assume is
compatible with the National part for programming purposes. My Data
I/O 3900 lists the National 74S287 in its database.
The ROMs both failed the blank check, and reading them with the
programmer produces a pattern that I can verify manually with a
breadboard/LED circuit. Is it possible that the Data I/O somehow
fried the chips while trying to read them? It still correctly detects
if the chip is place incorrectly or reversed.
Has anyone had success programming these parts with the National
74S287 algorithm, or is the part closer to one of the other
"read-compatible" versions of this PROM; the 3900 supports most of
these, I believe. Also, has anyone ever had the experience of a new
PROM not being blank?
Thanks,
David.
The programmer shouldn't have destroyed the chip by a read? They are
hard to "accidentally" program, as you have to raise VCC to +8.75v, then
apply +17.5 volts on the data pin you want to program while you've got
the address selected, and giving the CE/ a blip.
All bipolar proms (AFAIK) use the same programming algorithm:
http://www.leopardcats.com/bprom/bprom.htm
Was the part brand new, or a NOS? It's possible it's been sitting for so
long, it's just gone bad?
Good luck! Yours, Mark.
PS. That info on my website, I found somewhere else, but for the life of
me I can't find the URL, so if it's copyright by anyone, apologies :-)
A few datapoints for you:
I've used a lot of bipolar proms in the past and have seen:
1) Parts purchased from Jameco in the last year or two with *really* old
date codes ('79-'82, all Signetics). I had massive fallout on some of
those-- like 15-25%. They'd appear blank, but wouldn't program correctly in
my EMUP programmer.
2) Of the chips that failed on my EMUP from those 'bad' lots, my BP-1200
programmer would program maybe half of them OK. (So it looks like for
bipolar proms the quality of the programmer can make a difference-- the
BP1200 when new was probably ~10 times the price of the EMUP.) The EMUP is
within calibration, so I'm not real sure what's up there. Maybe an
algorithm difference...
3) I've bought a lot of parts from IC brokers that were pre-programmed.
They never put up a fuss about returning them, so it must happen on
occasion. (I suspect that they were old parts that someone had warehoused
and were bulk programmed, but were not marked afterwards-- so half a dozen
years later someone buys them surplus and assumes they're blank.) I've also
had some tubes of PROMs that were *partially* programmed (like the first 5
worked fine and then the next 20 were all already blown with the same data).
-Clay
hi, these parts are very reliable as a rule, the chances are that you
accidently programmed it with whatever was in your proggrammers
memory.next time fill your programmers memory with 0xff before you
start.
> All bipolar proms (AFAIK) use the same programming algorithm:
>
I've found reports on the net of some of the compatibles requiring
different programming conditions; apparently the compatibility is for
reading only.
> Was the part brand new, or a NOS? It's possible it's been sitting for so
> long, it's just gone bad?
>
It was new, but I believe all these parts must technically be NOS,
since they're discontinued (I can't even find an NTE listing for the
NTE replacement). I brought the parts back to Fry's to exchange; at
the service desk, I took one of the replacements out of the bubble and
placed in my breadboard. Sure enough, it too contained contained the
same zeroed bits (I checked the first 8 words.) I took a refund for
the parts, and suppose I'll be ordering them from Jameco for $7 a
pop+shipping :-( I suppose finding a needed part at Fry's--especially
an obscure one--was just way to good to be true :-)
The pattern I read manually on all 3 roms was: 0 2 4 6 1 3 5 7.
That's also what the Data I/O reported. Weird, isn't it? That's why
I posted--I thought there may be some systematic error. They *should*
read 7 7 7 7. . . etc in their pristine state, shouldn't they?
> http://www.leopardcats.com/bprom/bprom.htm
>
Thanks for the link--very interesting.
Regards,
David.
Oops, for these parts, 0 0 0 0 . . .
David
> hi, these parts are very reliable as a rule, the chances are that you
> accidently programmed it with whatever was in your proggrammers
> memory.next time fill your programmers memory with 0xff before you
> start.
When I returned the parts, I tested another in the store and found it
also pre-programmed with the same pattern (0 2 4 6 1 3 5 7 . . .); I
now therefore suspect that the parts were pre-programmed. Of course,
Fry's is still selling the rest of the batch :-)
David.
Just a thought, as you make all sorts of wonderful stuff for us
arcadies....
As these Bipolar proms are not made anymore, would it be possible to
progam a PIC, one that had enough memory, to hold the data of an 82s129,
and a little program to read the address lines and dish the data out
when needed? I know the pins are in the wrong place (why the HELL did
Microchip put the VCCs and GNDs where they did???), but you could do a
small PCB board that would adapt a PIC to the pinouts of the PROM? Not
the best of solutions, but it would get boards working.
Would that be a viable alternative? I've in the past made up a
cumbersome breadboard arrangement to simulate an 82s129 with a 2716, as
I don't (yet) have a programmer for them.
Yours, Mark.
Hum, try a different vendor.
Speed is the real problem there. Most little MCU's aren't too quick
relative to PROM speed. Maybe a fast micro with edge interrupts on all the
inputs would work, or something *really* fast like a Scenix (or whatever
they're called now ;-). Of course you'll need a clock and everything with
it. In general I'd think you'd need probably in excess of 18MIPS depending
on the instruction set and micro. (That'd be just three cycles to detect a
change on the inputs, lookup the new value, and write out the response in
less than ~160ns)
In general I'd just say go for an EPROM adapter. Fast/small EPROMs (like
57C291's) turn up a lot surplus (because nobody knows what to do with them
;-) and would be a good candidate. For video games that use the bipolar
PROMs as color and/or palette lookup tables you really only have to beat the
dot clock which in general is on the order of ~160ns or slower. (Figure
~53us of active video and maybe 320 pixels in that time-- probably more like
256 for older games.) Any EPROM in the 150ns range should be fine.
-Clay
My one experiment with a 2716 to prom adapter for pacman was not successful.
It was almost there, but a few random pixels were white. Maybe I just had a
bad eprom.
Dave
Could have just been a too-slow eprom...
--
Mark Spaeth msp...@mtl.mit.edu
50 Vassar St., #38.265 msp...@mit.edu
Cambridge, MA 02139
(617) 452-2354 http://rgvac.978.org/~mspaeth
Dave
"Mark C. Spaeth" <msp...@wizard-of-wor.mit.edu> wrote in message
news:3db88f20$0$3933$b45e...@senator-bedfellow.mit.edu...
No. There were at least a half a dozen. (I've got a number of data books
from that era). TI alone had (as I remember) 3 different ones. (Although
National and early parts TI did use the same algorithm).
Mark Zenier mze...@eskimo.com Washington State resident
Ubicom
Wouter van Ooijen
-- ------------------------------------
http://www.voti.nl
PICmicro chips, programmers, consulting