#Memory HELP!
13 messages in this thread
Jay,
If you could tell me the two values (entered and actual) I could probably
tell you which chip was bad. From your description so far I can narrow it down
to U15-U29 and U47-U54. The address indicates one of these 16 chips, the data
entered and what was actually stored will tell me which of them is the culprit.
-Dean
Dean, The first thing I should mention is that this 2052 is being used in a
Subsystem on a 500, since I have 1 meg internal could we assume that the memory
allocation routines would be the same as on the 2000? If not I can test it in a
friends 2000 and try to get the addresses. If they are the same then I will try
and explain it in further detail.
Ok, first thing I tried was two memory test programs, MEMTEST and
MTEST-2. Memtest gives me an error at address 002019A2, MTest-2 gives the error
at 002019A3. To try and clear this up I got MEd, a memory editor. Going to
address 002019A2 it would display;
002019A2: 00 FF 00 FF 00 FF 00 FF (or some combination of 00 and FF)
Now, I know it starts out this way when the address is not in use. then
I'll go in and put in all 0's, and this is the result;
002019A2: 00 01 00 00 00 00 00 00
You'll notice that address 002019A3 is in error, now I can continue with
other values, all 1's comes out ok. All 2's comes out 22 23 22 22 22 22 22 22,
again 002019A3 is in error. All 3's, ok. All 4's comes out 44 45 44 44 44 44 44
44. All 5's ok….etc. All A's comes out AA AB AA AA AA AA AA AA. All B's is
ok. All C's comes out CC CD CC CC CC CC CC CC ……etc.
<continued>
Jay,
The fact that the board is in an A500 expansion chassis shouldn't affect
anything. I assume though, that removing the 2052 makes the machine work
properly?
Also, is the total amount of memory in the machine 3Meg? If not, what
other devices are installed.
-Dean
Dean, I will remove chip U54, and have it shot, court martialed, and sent to
the Russian Front! Really though, I appreciate your help. I hope that my
somewhat vague description of the problem didn't cause you too much bother.
With the 2052 installed there is a total of almost 3 meg when I type
"avail", so evidently the system is recognizing all of the ram. The only other
device installed is a C-Ltd interface to my HD, and the FIRST thing I tried was
disconnecting the interface, same problem. When I was doing the memory tests
and when using the memory editor, they both were run from a cold boot from
empty (except for the programs themselves) floppies. That way I could be
(relativly) sure that the address wouldn't be in use. The thing that was
confusing me was that evidently the allocation routine puts the external ram
(or some portion of it) BEFORE the internal ram (512k), and something I was
running at boot-up was using that address and screwing up other tasks. So at
first I didn't believe that the new ram was being used at all, but I guess that
was the problem from the beginning.
Again, THANKS -Jay
Jay,
No problem at all, I'm used to recieving much less consice description of
problems, so your's was actually easy. All I did was to look at the 2052
schematics and trace the circuit to the faulty chip.
If there is no other FAST RAM installed in your machine, then U54 should
be the correct chip. And, yes, FAST memory is placed before CHIP memory in the
system memory lists so that FAST mem is the default memory type.
BTW, if you need a witness at the court martial, I'm available! 😉
-Dean
<continuation>
I hope this means something to you, because all I know is there definitly
is a problem, but, I haven't been able to track it down any further. Maybe it's
tracked down as far as it can be, it would be great if you could determine
which chip it is from the info I gave you. I'll be waiting (and hoping 🙂 for
your determination.
Thanks for your help, Jay
Jay,
The bad chip should be U54. From your description, bit zero of the lowest
bank of chips has one cell stuck high. If you replace that one chip, the
problem should disapear.
-Dean
Dean,
I've seen a diagnostics program for the AT&T 6300 — it does a ram test
and if anything fails it brings up a graphic with the offending chip outlined
in red. Anything like that for Amiga?
-sja
Steve,
Not that I've heard of. Of course it'd only work for CHIP memory, and it'd
have to know which machine it was running on in order to point out the correct
chip.
-Dean
Steve,
I remember seeing a program like that, but can't remember which memory
expansion it was for.
A test that will call out a chip is a specific test for one particular
make/model of board, and getting the right board even, could be a bit tricky.
The ASDG boards can be configured to not configure (huh?), if you know what I
mean, and the diagnostic software can locate an unconfigured board. It can also
callout the failing chip (in most cases, as is usual, because of the inability
to narrow down certain types of failures), though it does not do so
graphically.
-larry
I realized after I'd left the msg that it's not too likely such a program could
be done on the Amiga given the wide range of 3rd party boards and various
address ranges of Amiga memory. On MessyDOS it's easy — ya got 640k in a
contiguous space.
I just opened up a A2058 box today. Inside, along with the board, was a disk
with an A2058 ram test program on it (written by Rick Sterling). It looked to
be nicely done.
-sja
Steve,
Rick told me he was hoping to get that ram test into a product. He was very
proud of it, and tells me he wrote it in FORTH.
For those that almost recognize Rick Sterling's name, he used to be half of
Microtechnic Solutions, and wrote most of Sixth Sense, a beautifully done
telecomm program for the C64. He also did Common Sense, a stripped down version
supplied with C64 modems.
-larry