25 Jun
2016
25 Jun
'16
12:36 a.m.
Hi All,
In bringing up and debugging my PDP 11/45, I found that one of my GRA
(M8101) spares has a failed ALU subsidiary ROM. It's a pretty standard
little 32x8 ROM in a 16-pin DIP, and the truth table is in the 11/45
print set.
I wonder what the replacement options are for parts like these? In
particular, given the 30ns micro-cycle on the KB11-A, and the fact that
the propagation time for the ALU downstream of this is roughly 20ns on
its own, I'd be worried that an off-the-shelf bipolar PROM might be too
slow here.
I'm still a little slow on reading the microcode flows, so its not clear
to me exactly how many micro-cycles there are on the critical path for
the E-class instructions where this ROM is used. Maybe its not an issue.
Anybody every try replacing one of these with a bipolar PROM? Any other
suggestions for how to repair parts like these?
cheers,
--FritzM.
25 Jun
25 Jun
2:28 a.m.
On 6/24/2016 4:36 PM, Fritz Mueller wrote:
Hi All,
In bringing up and debugging my PDP 11/45, I found that one of my GRA (M8101)
spares has a failed ALU subsidiary ROM. It's a pretty standard little 32x8
ROM in a 16-pin DIP, and the truth table is in the 11/45 print set.
I wonder what the replacement options are for parts like these? In particular,
given the 30ns micro-cycle on the KB11-A, and the fact that the propagation
time for the ALU downstream of this is roughly 20ns on its own, I'd be worried
that an off-the-shelf bipolar PROM might be too slow here.
I'm still a little slow on reading the microcode flows, so its not clear to me
exactly how many micro-cycles there are on the critical path for the E-class
instructions where this ROM is used. Maybe its not an issue.
Anybody every try replacing one of these with a bipolar PROM? Any other
suggestions for how to repair parts like these?
cheers,
--FritzM.
Almost 100% certainty the part already there is a small bipolar TTL PROM. What
would you think it otherwise might be?
For a lot of these logic replacement applications DEC used the open collector
version, but it might be tristate variation. Check schematic.
Also, the microcycle on the 11/45 (and 11/70 for that matter, basically the same
design) is 150ns, not 30ns.
There are various clock timing pulses (tp1, tp2, etc) but the datapath / control
unit microcycle is 150ns.
Don
3:41 a.m.
Thanks for the info, Don -- learning a lot about this stuff as I go...
I had wondered if the part might have been a mask ROM rather than a
PROM. And wrt. timing, I was certainly mistaken to call the nominal
interval between the clock pulses a microcycle.
So after staring at the flows and prints a little more closely, it looks
to me now like the IR will be latched at FET.10 t6 (which is really
IRD.00 t1?) then there is the rest of intervening IRD.00 during which
time control signals can propagate to and through decode logic and the
subsidiary ROM and ALU, then the ALU results are latched into the
shifter at EXC.80 t2 or EXC.90 t2. So that's a solid 150ns there minimally?
From the prints, it looks like this is an open-collector part -- I
don't see it called out, but the chip select is wired active and I can't
see that the outputs have any other drivers.
So that's good news for repairing my board! Which brings on the next
question: do folks here have a recommendation for a good programmer to
try and track down on eBay for programming these sorts of parts?
cheers,
--FritzM.
On 06/24/2016 06:28 PM, Don North wrote:
Almost 100% certainty the part already there is a
small bipolar TTL
PROM. What would you think it otherwise might be?
For a lot of these logic replacement applications DEC used the open
collector version, but it might be tristate variation. Check schematic.
Also, the microcycle on the 11/45 (and 11/70 for that matter,
basically the same design) is 150ns, not 30ns.
There are various clock timing pulses (tp1, tp2, etc) but the datapath
/ control unit microcycle is 150ns.
3:42 a.m.
On Fri, Jun 24, 2016 at 7:41 PM, Fritz Mueller <fritzm at fritzm.org> wrote:
Thanks for the info, Don -- learning a lot about this
stuff as I go...
I had wondered if the part might have been a mask ROM rather than a PROM.
And wrt. timing, I was certainly mistaken to call the nominal interval
between the clock pulses a microcycle.
So after staring at the flows and prints a little more closely, it looks
to me now like the IR will be latched at FET.10 t6 (which is really IRD.00
t1?) then there is the rest of intervening IRD.00 during which time control
signals can propagate to and through decode logic and the subsidiary ROM
and ALU, then the ALU results are latched into the shifter at EXC.80 t2 or
EXC.90 t2. So that's a solid 150ns there minimally?
From the prints, it looks like this is an open-collector part -- I don't
see it called out, but the chip select is wired active and I can't see that
the outputs have any other drivers.
So that's good news for repairing my board! Which brings on the next
question: do folks here have a recommendation for a good programmer to try
and track down on eBay for programming these sorts of parts?
A Data-I/O 29A or 29B will do the job nicely.
- Josh
cheers,
--FritzM.
On 06/24/2016 06:28 PM, Don North wrote:
Almost 100% certainty the part already there is a
small bipolar TTL PROM.
What would you think it otherwise might be?
For a lot of these logic replacement applications DEC used the open
collector version, but it might be tristate variation. Check schematic.
Also, the microcycle on the 11/45 (and 11/70 for that matter, basically
the same design) is 150ns, not 30ns.
There are various clock timing pulses (tp1, tp2, etc) but the datapath /
control unit microcycle is 150ns.
3:47 a.m.
On 6/24/16 7:42 PM, Josh Dersch wrote:
A Data-I/O 29A or 29B will do the job nicely.
and a Unipak I or II like this
www.ebay.com/itm/301995688339
4:11 a.m.
On 2016-06-24 22:47, Al Kossow wrote:
On 6/24/16 7:42 PM, Josh Dersch wrote:
I have a Data-I/O Series 22 which could do these, plus another burner
that may do them. I haven't checked the whole thread, but I'm sure some
others on here have the equipment to do these too.
Is there a file containing the image ?
You'll need to identify if that chip uses tri-state or open collector
bipolar PROM.
If DEC didn't completely obliterate the original markings, those would
be very helpful (i.e. Atari tended to stamp on top of the original chip
#s, so with some careful examination, you can usually figure out the
original chip #).
The bipolar PROMs aren't cheap like an EPROM.... depending on the
source/brand/timing they cost anywhere from $5 to $20+ each. Then most
people will usually ask for $5 to $10 to burn one.... or if you are
really lucky... free....
-- Curt
A Data-I/O 29A or 29B will do the job nicely.
and a Unipak I or II like this
www.ebay.com/itm/301995688339 
4:36 a.m.
On 2016-Jun-24, at 7:41 PM, Fritz Mueller wrote:
On 06/24/2016 06:28 PM, Don North wrote:
> Almost 100% certainty the part already there is a small bipolar TTL PROM. What would
you think it otherwise might be?
>
> For a lot of these logic replacement applications DEC used the open collector
version, but it might be tristate variation. Check schematic.
>
> Also, the microcycle on the 11/45 (and 11/70 for that matter, basically the same
design) is 150ns, not 30ns.
>
> There are various clock timing pulses (tp1, tp2, etc) but the datapath / control unit
microcycle is 150ns.
Thanks for the info, Don -- learning a lot about this
stuff as I go...
I had wondered if the part might have been a mask ROM rather than a PROM. And wrt.
timing, I was certainly mistaken to call the nominal interval between the clock pulses a
microcycle.
So after staring at the flows and prints a little more closely, it looks to me now like
the IR will be latched at FET.10 t6 (which is really IRD.00 t1?) then there is the rest of
intervening IRD.00 during which time control signals can propagate to and through decode
logic and the subsidiary ROM and ALU, then the ALU results are latched into the shifter at
EXC.80 t2 or EXC.90 t2. So that's a solid 150ns there minimally?
Many/most of the common bipolar fusible proms are Schottky class, so are quite fast.
Take a look at 74S188 / 74S288 as a starting point. Down in the 20-30ns range.
From the prints, it looks like this is an
open-collector part -- I don't see it called out, but the chip select is wired active
and I can't see that the outputs have any other drivers.
Are there pull-up resistors anywhere along the output/data lines? If not, it is more
likely a tri-state device.
So that's good news for repairing my board! Which
brings on the next question: do folks here have a recommendation for a good programmer to
try and track down on eBay for programming these sorts of parts?
Probably more than you want to bother with, but blowing fusible proms generally isn't
all that difficult, I've hacked a burner on a breadboard with an RPi (substitute other
microcontroller as desired), 2-3 common TTL ICs and a few transistors.
5:11 a.m.
On 6/24/2016 7:41 PM, Fritz Mueller wrote:
On 06/24/2016 06:28 PM, Don North wrote:
Device E74 on GRAA is a 32x8 bipolar PROM and there are no pullups on any
outputs, so it is a tri-state device. The outputs directly drive the inputs of
several TTL logic gates.
Even if it were a mask ROM (not likely in that timeframe for that size device
and in the volume DEC would have used it) the timing would still not be that
different from a fusible link PROM. Access time would be in the 25ns to 50ns range.
The 11/45 timing generator has five timing pulses TP1 thru TP5; each one in
sequence at 30ns intervals, 150ns from TP1 to next TP1. In the flows they use
nomenclature of TP1 thru TP6, and as you indicate TP6 is really the next TP1.
Don
Almost 100% certainty the part already there is a
small bipolar TTL PROM.
What would you think it otherwise might be?
For a lot of these logic replacement applications DEC used the open collector
version, but it might be tristate variation. Check schematic.
Also, the microcycle on the 11/45 (and 11/70 for that matter, basically the
same design) is 150ns, not 30ns.
There are various clock timing pulses (tp1, tp2, etc) but the datapath /
control unit microcycle is 150ns.
Thanks for the info, Don -- learning a lot about this stuff as I go...
I had wondered if the part might have been a mask ROM rather than a PROM. And
wrt. timing, I was certainly mistaken to call the nominal interval between the
clock pulses a microcycle.
So after staring at the flows and prints a little more closely, it looks to me
now like the IR will be latched at FET.10 t6 (which is really IRD.00 t1?) then
there is the rest of intervening IRD.00 during which time control signals can
propagate to and through decode logic and the subsidiary ROM and ALU, then the
ALU results are latched into the shifter at EXC.80 t2 or EXC.90 t2. So that's
a solid 150ns there minimally?
From the prints, it looks like this is an open-collector part -- I don't see
it called out, but the chip select is wired active and I can't see that the
outputs have any other drivers.
So that's good news for repairing my board! Which brings on the next
question: do folks here have a recommendation for a good programmer to try and
track down on eBay for programming these sorts of parts?
cheers,
--FritzM.
4:10 a.m.
On 06/24/2016 06:36 PM, Fritz Mueller wrote:
Hi All,
In bringing up and debugging my PDP 11/45, I found that
one of my GRA (M8101) spares has a failed ALU subsidiary
ROM. It's a pretty standard little 32x8 ROM in a 16-pin
DIP, and the truth table is in the 11/45 print set.
I wonder what the replacement options are for parts like
these? In particular, given the 30ns micro-cycle on the
KB11-A, and the fact that the propagation time for the ALU
downstream of this is roughly 20ns on its own, I'd be
worried that an off-the-shelf bipolar PROM might be too
slow here.
I'm still a little slow on reading the microcode flows, so
its not clear to me exactly how many micro-cycles there
are on the critical path for the E-class instructions
where this ROM is used. Maybe its not an issue.
Anybody every try replacing one of these with a bipolar
PROM? Any other suggestions for how to repair parts like
these?
Well, you could use a scope to find out what the good ROMS
do for access time.
Jon
5:15 a.m.
On Fri, Jun 24, 2016 at 4:36 PM, Fritz Mueller <fritzm at fritzm.org> wrote:
Hi All,
In bringing up and debugging my PDP 11/45, I found that one of my GRA
(M8101) spares has a failed ALU subsidiary ROM. It's a pretty standard
little 32x8 ROM in a 16-pin DIP, and the truth table is in the 11/45 print
set.
http://www.bitsavers.org/pdf/dec/pdp11/1145/1145_System_Engineering_Drawing…
If you are talking about he ALU CNTL ROM on page 40 of the PDF and the
truth table for it listed on page 48 of the PDF the part is listed as
DM8598-AD, where a DM8598 is a 256-bit (32x8) tri-state bipolar mask
ROM.
Some substitute T.S. PROMs include the Signetics 82S123, AMD 27S19,
National DM74S288.
My BP Microsystems programmers support the 82S123 and 27S19, but maybe
not the DM74S288.
6:21 a.m.
Thanks much for all the info and help, folks!
I'll see about tracking one of the programmers mentioned. I'd like to
have one around in case I run into more parts like these. Was helping a
friend with an old arcade game last weekend, and it had a few parts like
these as well.
--FritzM.
4:40 p.m.
Here is an ebay link to a pretty good (IMHO) guide to the data I/O programmers.
http://rover.ebay.com/rover/1/711-53200-19255-0/1?icep_ff3=2&pub=557493…
Sent from my iPhone
On Jun 24, 2016, at 10:22 PM, Fritz Mueller <fritzm at fritzm.org<mailto:fritzm at
fritzm.org>> wrote:
Thanks much for all the info and help, folks!
I'll see about tracking one of the programmers mentioned. I'd like to have one
around in case I run into more parts like these. Was helping a friend with an old arcade
game last weekend, and it had a few parts like these as well.
--FritzM.
10:16 a.m.
On Fri, 24 Jun 2016, Fritz Mueller wrote:
In bringing up and debugging my PDP 11/45, I found
that one of my GRA (M8101)
spares has a failed ALU subsidiary ROM. It's a pretty standard little 32x8
ROM in a 16-pin DIP, and the truth table is in the 11/45 print set.
Some years ago I had to replace one such PROM in my 11/45 that is
responsible for the ALU conditional codes. Since the PROM contents is
quite simple, I replaced that with a GAL with the corresponding logic
equations. Not a pretty solution, but I can replace it with a new PROM
when I get one.
A similar fault was in my RK05 "Cylinder Address and Difference" board,
also changed the failed PROM with a GAL. The symptom was that the drive
never "found" the desired track and instead, bumped the carriage to the
inner stop.
Christian
2886
days inactive
2887
days old
12 comments
10 participants
participants (10)
-
aek＠bitsavers.org -
cc＠informatik.uni-stuttgart.de -
derschjo＠gmail.com -
elson＠pico-systems.com -
fritzm＠fritzm.org -
glen.slick＠gmail.com -
hilpert＠cs.ubc.ca -
north＠alum.mit.edu -
rescue＠hawkmountain.net -
wsudol＠freedom.com