Well, in the category of "it ain't gonna fix itself" I fixed it.
Testing in progress - Please wait
Memory Size is 4088 K Bytes
9 Step memory test
Step 1 2 3 4 5 6 7 8 9
Message 04 Entering Dialog mode
The problem with the memory was that one of the 256k*1 chips was bad.
Easy to find, it was bit 4000 in the top of the second 1mw memory half.
Problem is while the table on the gunkies.org
site is accurate for bits
within a word, it's totally inaccurate when it comes to describing the
(Note: I'm using 16 bit words as the board has the chips laid out in 18
bit stripes. 16 for data, 2 for parity)
The banks of memory are laid out as follows:
going across the board. Thus bank 1 is 0-256kw and is across the top.
The second bank for 256-512kw is right underneath it, then you skip the
next two rows to get to 512-768kw. Then the row under that is 768-1024kw
To go to the second half of memory you go back up then to row 3 which is
bank 5 (1024-1280), then 6, then skip 2 to get to bank 7 and 8.
Realized this when I found a picture of an MSV11-P memory that had only
half the chips. DEC put them in that same order, and they did the same
on the MSV11-Q memories.
So I RULE!
And now I have a full 4mb memory card. Not sure what to do with it, I
think my next step is going to be to take this 1mb msv11-JD memory board
and see if I can put on enough chips to make it a 2mb JE board. Because
PMI memory is just faster... :-)
On 12/31/2022 6:34 PM, Chris Zach via cctalk wrote:
So I have an MSV11-QC (the 4gb Q Bus memory board)
that has a failed
bit. On an 11/73 it throws the following error:
Expected data = 125252 (1010 1010 1010 1010
Bad data = 121252 (1010 0010 1010 1010)
Address = 10015140
Which puts the error in the top 2mb (1mw), bit 4000. No biggie,
consulting the matrix at:
Shows this to be bank 4, bit 4000 or E85.
Cut out E85, popped the board in (to double check before replacing) and
I see this:
Expected data = 125252
Bad data = 121252
Address = 02015140
Ooops. That would be decimal 531,040 which puts it in bank 1? Which
should be E88, where did I go wrong here?
Or is the table for the MSV11-Q only valid for the 64k ones and not the
256k ones? If so the bits may be right, but the banks are different.
Anyone want to cross-check this or my maths?
PS: Sorry if duplicates are getting sent to the list. I'm retiring
cz(a)alembic.crystel.com and replacing it with cz(a)beaker.crystel.com. If
the list manager could update my email I would appreciate!