19 Aug
2014
19 Aug
'14
7:21 a.m.
I spent some time tonight debugging my problematic 11/05. The current
issue is that Exam and Dep do not increment the address when depressed,
and the PC does not increment when executing a series of NOPs.
I started with the 74181 ALU for bits 0:3, checking the inputs, outputs,
and the function control lines during an "Exam" operation. The inputs
are correct but the outputs are always just the contents of A. The
control inputs (mode and function) are all high, meaning that the ALU
operation is set to just pass A through unchanged, which explains the
behavior.
The control inputs to the ALU are generated by two bipolar ROMs, (named
A11A2 and A20A2) on the M7261 board. The addresses and Enable lines
going into the ROMs look to be correct but the data coming out is all
1s, leading to the incorrect ALU control.
I'm unsure whether these might be faulty ROMs or something else
(incidentally, anyone know what kind of ROMs these are so I might hunt
down a datasheet? The markings on the chips are all different and have
lead me nowhere and the engineering drawings just refer to them as
A11A2, etc. in the parts lists).
Additionally, I cannot find a listing of the contents of these anywhere;
the print sets on Bitsavers and elsewhere have listings for most of the
other ROMs but of course not these two (unless they're tucked away
somewhere apart from the schematics where I've somehow missed them...)
Anyone spent any time debugging the microcode logic on one of these?
Any tips?
Thanks,
Josh
19 Aug
19 Aug
9 p.m.
On Tue, Aug 19, 2014 at 1:21 AM, Josh Dersch <derschjo at gmail.com> wrote:
I spent some time tonight debugging my problematic
11/05. The current issue
is that Exam and Dep do not increment the address when depressed, and the PC
does not increment when executing a series of NOPs.
So a single NOP *is* a series of NOPs?
9:13 p.m.
On Tue, Aug 19, 2014 at 5:00 PM, Eric Smith <spacewar at gmail.com> wrote:
On Tue, Aug 19, 2014 at 1:21 AM, Josh Dersch
<derschjo at gmail.com> wrote:
NOP-and-BRANCH-TO-SELF?
-ethan
I spent some time tonight debugging my
problematic 11/05. The current issue
is that Exam and Dep do not increment the address when depressed, and the PC
does not increment when executing a series of NOPs.
So a single NOP *is* a series of NOPs?
9:33 p.m.
On Tue, Aug 19, 2014 at 2:00 PM, Eric Smith <spacewar at gmail.com> wrote:
On Tue, Aug 19, 2014 at 1:21 AM, Josh Dersch
<derschjo at gmail.com> wrote:
Yeah, it's pretty efficient space-wise that way.
- Josh
I spent some time tonight debugging my
problematic 11/05. The current
issue
is that Exam and Dep do not increment the address
when depressed, and
the PC
does not increment when executing a series of
NOPs.
So a single NOP *is* a series of NOPs?
20 Aug
20 Aug
3:06 a.m.
On Tue, Aug 19, 2014 at 5:33 PM, Josh Dersch <derschjo at gmail.com> wrote:
On Tue, Aug 19, 2014 at 2:00 PM, Eric Smith
<spacewar at gmail.com> wrote:
Finally, a use for my box of 1x8 PROMs!
-ethan
On Tue, Aug 19, 2014 at 1:21 AM, Josh Dersch
<derschjo at gmail.com> wrote:
Yeah, it's pretty efficient space-wise that way. The current issue is that Exam and Dep do not
increment the
address when depressed, and the PC does not increment
when executing a series of NOPs.
So a single NOP *is* a series of NOPs?
10:20 p.m.
To return to the original question, I beleive the contents of those PROMs
are in the printset. Just not obviously.
The APROMs are part of the microcode, the format of the microcode word is
in the pritnest I looekd in on bitsavers. As is the microcode binary
listing. So you have to extract the appropriate bits from that listing.
It's only 256 locations...
The PROms are 256*4 bipolar PROMs. Various companies made them, they all
should be equivlaent when reading, but nor when programming. So if you
get some, make sure your programmer can handle that brand.
-tony
21 Aug
21 Aug
12:18 a.m.
On Wed, Aug 20, 2014 at 3:20 PM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
To return to the original question, I beleive the
contents of those PROMs
are in the printset. Just not obviously.
The APROMs are part of the microcode, the format of the microcode word is
in the pritnest I looekd in on bitsavers. As is the microcode binary
listing. So you have to extract the appropriate bits from that listing.
It's only 256 locations...
Thanks, Tony. I actually realized that late last night (and I feel like an
idiot for overlooking it). I've transcribed the ALU column over to a text
file and now I just need to reorder it by address (rather than by symbol)
and convert it into a binary.
The PROms are 256*4 bipolar PROMs. Various companies made them, they all
should be equivlaent when reading, but nor when programming. So if you
get some, make sure your programmer can handle that brand.
Worked that out last night too (it was a very productive evening :)). I
found some NOS MMI 6300s for a reasonable price and they're on their way.
My ancient Data I/O Model 29 w/Unipak 2 should be able to program them. I
hope I don't need to replace too many of these...
- Josh
-tony
7:30 p.m.
Thanks, Tony. I actually realized that late last night (and I feel like an
idiot for overlooking it). I've transcribed the ALU column over to a text
I'd better not start a thread pf the number of times I've felt like a
complete idiot. We'd all run out of disk space :-)
file and now I just need to reorder it by address
(rather than by symbol)
and convert it into a binary.
DEC (obviously) didn't wnt you to program your own PROMs, they expected
you to buy spares from them :-). So the listings wre not designed as PROM
programmign files, there are there, esentially, to expalin the operation
of the machine. And very useful they are too.
Hve yuo made sure no oher microcode PROM gives all 1's? IIRC these are
open-collector outptu PROMs, so if the enable cirucitry fails, all
outputs float (And are puleld to 1's bu the external resistors).
-tony
22 Aug
22 Aug
5:52 a.m.
On 8/21/2014 12:30 PM, Tony Duell wrote:
:).
Yes, they make debugging of the machine actually possible to
accomplish. The 11/05 manuals are very thorough, I'm finding them very
easy to follow.
Thanks, Tony.
I actually realized that late last night (and I feel like an
idiot for overlooking it). I've transcribed the ALU column over to a text
I'd better not start a thread pf the number of times I've felt like a
complete idiot. We'd all run out of disk space :-) file and now I just need to reorder it by address
(rather than by symbol)
and convert it into a binary.
DEC (obviously) didn't wnt you to program your
own PROMs, they expected
you to buy spares from them :-). So the listings wre not designed as PROM
programmign files, there are there, esentially, to expalin the operation
of the machine. And very useful they are too.
Hve yuo made sure no oher microcode PROM gives all 1's? IIRC these are
open-collector outptu PROMs, so if the enable cirucitry fails, all
outputs float (And are puleld to 1's bu the external resistors).
So far, it looks like the A20A2 ROM may also be faulty. Amongst other
things, it controls the "Mode" bit for the ALU and it's always coming
out 1. So I suspect there are probably others but I figure I'll take it
one at a time. If they were socketed I'd just pull them and dump them
out but as it is I'll wait to see what else is broken first. There are
a lot of these things on these two boards!
The enable circuitry does seem to be OK, and the faulty A11A2 ROM does
have some non-zero data in it (verified while debugging by having my
logic analyzer trigger on a non-zero output and later by dumping it
after removing it from the board). I haven't examined it closely to see
if there's a pattern to the bad data or not, I suspect there probably is.
- Josh
-tony
4 Sep
4 Sep
9:08 a.m.
So, just a quick follow up on this -- the MMI 6300s arrived last week
and I finally had time this evening to put them to use. I transcribed
the data for the faulty A11A2 and A20A2 PROMs (ALU control plus a few
other bits) from the engineering drawings, had my wife double-check my
work (she's very patient with me :)) wrote a quick tool to turn my
transcriptions into binaries and used my Data I/O 29A to burn those
binaries onto the PROMs.
I'd previously installed sockets for the PROMs I'd pulled, so I
installed the new PROMs in the sockets, and fired the 11/05 up, and...
it was behaving completely incomprehensibly. Doing anything from the
front panel sent the machine off into the weeds.
Drat.
After futzing around for awhile, I took a closer look at the schematic
(Page 91 in the PDF on bitsavers if you're following along at home).
Let's see... the address lines for the A20A2 PROMs are labeled "CONF ROM
ADRS 00 (1) H" and the address lines for the A11A2 are... "CONF ROM ADRS
00 (1) *L*." So my A11A2's contents are backwards because the address
lines for that PROM are active low. Why? Good question. To make my
life miserable, I suppose.
Burned a new A11A2 with the data in reverse order and hey, Exam and
Deposit work correctly! Simple programs also seem to be running
correctly but I'm guessing that I'm not out of the woods yet. At any
rate it's now rather late and so more vigorous testing will commence
tomorrow...
Thanks all for the tips,
Josh
On 8/22/2014 4:57 PM, Tony Duell wrote:
Yes, they make
debugging of the machine actually possible to
accomplish. The 11/05 manuals are very thorough, I'm finding them very
easy to follow.
Agreed. DEC manuals of this vintageare very clear and complete. I
learnt a
lot from the PDP11/45 and PDP8/e manuals sets.
-tony
6:57 p.m.
Let's see... the address lines for the A20A2 PROMs
are labeled "CONF ROM
ADRS 00 (1) H" and the address lines for the A11A2 are... "CONF ROM ADRS
00 (1) *L*." So my A11A2's contents are backwards because the address
lines for that PROM are active low. Why? Good question. To make my
life miserable, I suppose.
My guess (without checking the printset) is that the microcode program
counter register has both Q and Q/ available. By using both (some ROMs on
Q, some on Q/ ) you can meet the fanout requirements, there may be too
many ROMs for the address pins of all of them to be driven by a single
set of TTL outputs.
At least they didn't feed soem ROMS with a mixture of Q and Q/ signals --
did they???
-tony
3742
days inactive
3758
days old
12 comments
5 participants
participants (5)
-
ard＠p850ug1.demon.co.uk -
cisin＠xenosoft.com -
derschjo＠gmail.com -
ethan.dicks＠gmail.com -
spacewar＠gmail.com