• Please review our updated Terms and Rules here

Repair of a Clearpoint DCME-Q4E-F

C

czunit

Veteran Member
Joined
Aug 7, 2015
Messages
530
Well, this is interesting. Bought a Clearpoint 4mb PMI memory board on Ebay from a reputable seller. Got the card, plugged it in, didn't work. Changed one jumper that appeared to be the +5 battery/+5 line, and the card came up. Probably was used in a pdp11 with battery backup memory.

Anyway, the card worked well, passed all diagnostics, however it was only coming up as a normal Q Bus memory card.

Checking around I found a picture of another users' memory board, he had additional jumpers on the C/D jumpers, mine did not. That's not a good sign: If a PMI card is plugged into a normal Q bus slot that will usually blow out the board. But I ordered a set of jumpers, put them in, and now I get this with the board in slot 1, CPU slot 2 (CPU PMI works, I have a 1mb PMI board for testing):

Error 46
Memory Error

See troubleshooting documentation


Error PC = 173256 PCR page = 15 Program listing address = 015256

R0 = 060002 R1 = 125252 R2 = 125772 R3 = 052525
R4 = 000100 R5 = 040002 R6 = 172300 Par3 = 062400

Expected data = 125252
Bad data = 125772
Address = 06240002

Command Description

1 Rerun test
2 Loop on test
3 Map memory and I/O page
4 Advance to the next test

Time to start figuring out what's up... However at least it does report as Parity PMI...
Memory Map
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type

00000000 - 17757776 4088 17772102 Parity PMI
 
One thing that is interesting: The board has a total of 148 256k chips. Which makes no sense as a parity board should have 144 chips (18 bits * 8 banks of memory). So what are the extra 4 chips for?
 
Hm. Powered down, up,
Error 46
Memory Error

See troubleshooting documentation


Error PC = 173256 PCR page = 15 Program listing address = 015256

R0 = 064000 R1 = 125252 R2 = 000042 R3 = 052525
R4 = 000100 R5 = 044000 R6 = 172300 Par3 = 020600

Expected data = 125252
Bad data = 000042
Address = 02064000
 
Interesting. On a lark I tried to use this system (11/73 cpu, quad width) with a PMI memory board. Passes tests, but does not work right.

Testing in progress - Please wait
Memory Size is 1024 K Bytes
9 Step memory test
Step 1 2 3 4 5 6 7 8 9

Message 04 Entering Dialog mode

Commands are: [Help, Boot, List, Setup, Map, Test]
Type a command then press the RETURN key: BOOT DU0


Trying DU0

Starting system


000002
@

(System is an RT11 drive) Memory is as follows:
Starting Ending Size in CSR CSR Bus
Address address K Bytes address type type

00000000 - 03777776 1024 17772100 ECC PMI

Maybe I should try my 11/83 in this configuration. Or maybe the memory board is damaging PMI boards. Thoughts?
 
As far as I remember, a KDJ11-A worked with PMI memory. Moreover, this memory worked with other processor boards that do not know anything about PMI. I'll try today a little later to assemble a stand with a KDJ11-A.
 
It does, boots RSXM+. I must have screwed up the boot block on my RT11 disk, easy fix.

So I know the CPU runs fine with real PMI memory (and compiles Adventure 30% faster) now we can figure this board out.

Fact: Board works fine as Q bus memory
Fact: Board blows up at random places as PMI memory
Fact: Board does not boot M+ properly as PMI memory (parity faults)
Fact: Board does boot M+ properly in Q Bus mode.

Test 1: See if those last 4 chips are bad?
Pulled the last chip from the board (all chips socketed, thank you Clearpoint!)
Run board in Q Bus mode.
Results:
Board comes up. Yellow light in center is on.
Board passes boot diagnostics as 4mb memory.
Board passed 10 continious loop self tests with no error.
Try booting M+
System started to boot then stopped before displaying anything.

Backout:
Put in DEC PMI board as Q Bus memory
Booted M+
M+ Came up.


Thought:
Last 4 chips may not be needed to access as Q bus memory, need to remove other 3 and check.
They do seem to do something, maybe they are scratchpads for DMA requests or something
Will try more stuff later, bedtime.
 
czunit said:
Yellow light in center is on.
Click to expand...
It seems to me that the yellow LED is a signal of a correctable error. Prehistory. The board I bought was half full of memory chips. I bought similar memory chips from aliexpress and completely filled the board with memory chips. After turning on, the yellow LED began to light up, but the board worked - both RSX and RT loaded without problems. As it turned out, I carelessly inserted one chip into the socket - one contact was bent and did not hit the right place. After the problem was fixed, the yellow LED stopped lighting up.
 
Do you remember which chip?

This morning before work I tried removing another chip, bottom row, one to the left of the last chip. This caused the board to fail in Q bus mode with memory errors, so it looks like I hit a low memory bank. Put that chip in and took out the chip above the first one. This time it came up failing the CSR test, but the memory itself did pass diagnostics.

I think I see a pattern. Possible the last column of chips at the bottom of the board are the four used for board housekeeping. I'm going to try replacing them with 150ns 256k memory chips (all I have) and see what's up.
 
Well, worst case I can just test the chips 16 at a time in my Pro/350. If you can remember which slots were populated vs empty that would help.

I guess it has some sort of ECC memory but presents as parity. Weird.
 
czunit said:
If you can remember which slots were populated vs empty that would help.
Click to expand...
Fortunately - there was a photo - http://www.kpxx.ru/dec/PDP-11/Hardware/DCME-Q4E-F/ClearPoint DCME-Q4E-F-Front-before.jpg

Fully filled with memory chips - http://www.kpxx.ru/dec/PDP-11/Hardware/DCME-Q4E-F/ClearPoint DCME-Q4E-F-Front.JPG

But you still need to set the jumpers - the top four (размер памяти в Мб) determine the size of the memory - http://www.kpxx.ru/dec/PDP-11/Hardware/DCME-Q4E-F/ClearPoint DCME-Q4E-F-switches.JPG
 
Well, bad news: I set the card to 1mb using the jumpers, still doesn't work.

Pulled every chip in the first 2 megabytes (based on Hunta's picture) and tested them in my Pro/350 by plugging them 16 at a time into the Pro memory board and running the power up diagnostics.

All pass, which means the problem is in the ASIC or the PLA's. I guess someone plugged it into a Q/Q slot with the jumpers in, then removed the jumpers so it works.

Drat. Any other thoughts? Worst case it's a 4mb memory board, just good old Q bus slow..... :)

Next up: Adding 44 chips to my 1mb PMI board to turn it into a 2mb. Will the heat from the preheater damage the chips already on the board?
 
czunit said:
Next up: Adding 44 chips to my 1mb PMI board to turn it into a 2mb. Will the heat from the preheater damage the chips already on the board?
Click to expand...

I'll be following that one because I have an MSV11-JD (1MB) card that I'd like to upgrade. Looks like it's more than just adding the RAM chips (and bypass caps) though - there are some resistor SIPs and two AM2965 chips that need to be added as well. Then hopefully it's just a matter of removing W2 to tell it the board is fully populated. But I'll let you go first and report back (maybe start a new thread for it?) :)
 
Yah. The good thing is I have a MSV11-JB (2mb, 11/84 only) that I can loot for parts. I could do sockets, but the problem is DEC boards have chips placed so close that sockets tend to jam against each other with no clearance. Besides, it's ECC, right?
 
 
In my case, I have a PDP11 system with 4MB Clearpoint memory. When I try to load the program with 4 floppy disks (3 1/2"), the error always occurs on the fourth floppy disk. (MORE DATA THEN MEMORY).
Could you help me with the reason for the failure? Is it related to a bit error?

System Log:
Halt/Reset/Enable

MEMORY 0 -> 16770000-1 OR 035760 256. BYTE BLOCKS
Type Load/Dump: L

Type drive unit number: 1

Insert disk xxxxx A into drive 1. type <CR>:
Insert disk xxxxx B into drive 1. type <CR>:
Insert disk xxxxx C into drive 1. type <CR>:
Insert disk xxxxx D into drive 1. type <CR>:

MORE DATA THAN MEMORY

163576
@

Note:
When I performed the first memory dump the system reported:

GOOD DUMP

MEMORY 0-16766000 OR 035754 256. BYTE BLOCKS
 
czunit said:
Hm. What are you loading and how?
Click to expand...
I'm loading a dedicated Measurex system with 4 3 1/2" floppy disks. This floppy disks is an image (dump) of the memory. When I go to restore the memory fails to write. I believe it is a damaged memory chip in writing, as in reading it works normally.

When I used the your jumper configuration (http://www.kpxx.ru/dec/PDP-11/Hardware/DCME-Q4E-F/ClearPoint DCME-Q4E-F-switches.JPG), I was able to load the program because the reserved area of the window was ignored. The Measurex system uses only 3.75MB of memory. The rest of the 256KB is window reserved.

Measurex jumper configuration:
MEMORY 0 -> 16770000-1 OR 035760 256. BYTE BLOCKS

Your jumper configuration
MEMORY 0 -> 17700000-1 OR 037600 256. BYTE BLOCKS

However, I need to load the program with the original memory card configuration.

I want to replace all the memory chips on this pcba.
 
You must log in or register to reply here.
Back
Top