27 Nov
2011
27 Nov
'11
3:35 a.m.
Hi all --
Picked up a Vaxstation 3100 today at the SRCS meeting. There is some
minor corrosion on the memory and motherboard PCBs, some due to
prolonged exposure to moisture, and some due to a leaking clock
battery. I've cleaned up the corrosion and it doesn't appear to have
done any serious damage.
At POST it identifies itself as a KA42-A, which I believe makes it a
Model 30.
I'm getting error messages from the POST and I'm unsure how to interpret
them -- the manuals have not been 100% clear. Here's what I get:
KA42-A V1.3
F...E...D...C...B...A...9...8...7...6...5...4...3_..2_..1?..
? E 0040 0000.0005
? C 0080 0000.4001
? B 0010 000C.0081
?? 1 00C0 0011.700E
The (E) error I expect since there's no clock battery. The (C) is also
expected since I don't have a keyboard/mouse plugged in (just using the
serial console). Similarly, (1) is also expected since I don't
currently have the ethernet interface hooked up to anything.
(B) is what's confusing me -- my understanding is that a single question
mark indicates a "soft" error, described in the manual thus: "Usually, a
question mark (?) in the power-up summary indicates a soft error and the
system may be usable without replacing any FRUs." OK. What does that
mean for memory, and how do I interpret the error code? I know that the
"000C" part indicates the amount of memory installed -- what does
"0081"
mean in this context, and should I be concerned?
For further reference, "test 50" gives the following for memory:
? MEM 000C.0081
00400000
The machine does boot (VMS 6.0 off of two very noisy 104MB drives) and
it seems to work OK, though I haven't stressed it much yet.
Any suggestions here would be much appreciated.
Thanks!
Josh
28 Nov
28 Nov
1:18 a.m.
To (sort of) answer my own question:
Looks like a "soft" error w.r.t. memory for this particular case means
that it's a fault found on a memory expansion board -- the system can
run (in a degraded state) in this situation by simply ignoring the
expanded memory. Checking "SHOW MEMORY" under VMS confirms this
suspicion. Further, removing said memory expansion allows the memory
self tests to pass with flying colors.
So, my 8MB expansion board appears to be toast. On closer inspection it
looks like there's some corrosion around one of the memory chips that
may have destroyed some traces underneath it. Anyone have any of these
boards going spare? (it's labeled as a "4/8 MB Memory Option") 4MB is
a little tight :).
Thanks as always,
Josh
On 11/27/2011 12:35 AM, Josh Dersch wrote:
Hi all --
Picked up a Vaxstation 3100 today at the SRCS meeting. There is some
minor corrosion on the memory and motherboard PCBs, some due to
prolonged exposure to moisture, and some due to a leaking clock
battery. I've cleaned up the corrosion and it doesn't appear to have
done any serious damage.
At POST it identifies itself as a KA42-A, which I believe makes it a
Model 30.
I'm getting error messages from the POST and I'm unsure how to
interpret them -- the manuals have not been 100% clear. Here's what I
get:
KA42-A V1.3
F...E...D...C...B...A...9...8...7...6...5...4...3_..2_..1?..
? E 0040 0000.0005
? C 0080 0000.4001
? B 0010 000C.0081
?? 1 00C0 0011.700E
The (E) error I expect since there's no clock battery. The (C) is
also expected since I don't have a keyboard/mouse plugged in (just
using the serial console). Similarly, (1) is also expected since I
don't currently have the ethernet interface hooked up to anything.
(B) is what's confusing me -- my understanding is that a single
question mark indicates a "soft" error, described in the manual thus:
"Usually, a question mark (?) in the power-up summary indicates a soft
error and the system may be usable without replacing any FRUs." OK.
What does that mean for memory, and how do I interpret the error
code? I know that the "000C" part indicates the amount of memory
installed -- what does "0081" mean in this context, and should I be
concerned?
For further reference, "test 50" gives the following for memory:
? MEM 000C.0081
00400000
The machine does boot (VMS 6.0 off of two very noisy 104MB drives) and
it seems to work OK, though I haven't stressed it much yet.
Any suggestions here would be much appreciated.
Thanks!
Josh
1:37 p.m.
New subject: Expansion memory boards (was: Interpreting Vaxstation 3100 memory error)
In article <4ED3279F.30009 at mail.msu.edu>,
Josh Dersch <derschjo at mail.msu.edu> writes:
So, my 8MB expansion board appears to be toast. On
closer inspection it
looks like there's some corrosion around one of the memory chips that
may have destroyed some traces underneath it. Anyone have any of these
boards going spare? (it's labeled as a "4/8 MB Memory Option") 4MB is
a little tight :).
I've been wondering how hard it would be to start building "new"
expansion memory boards for various machines.
Surely the timing of modern memory chips is fast enough to meet the
constraints of older systems and one-off PCB manufacturing should
allow you to make the necessary daughter boards for a variety of
machines.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 version available for download
<http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/>
Legalize Adulthood! <http://legalizeadulthood.wordpress.com>
1:49 p.m.
New subject: Expansion memory boards (was: Interpreting Vaxstation 3100 memory error)
On Mon, Nov 28, 2011 at 1:37 PM, Richard <legalize at xmission.com> wrote:
Surely the timing of modern memory chips is fast
enough to meet the
constraints of older systems and one-off PCB manufacturing should
allow you to make the necessary daughter boards for a variety of
machines.
I remember reading somewhere in a memory compatibility guide that the
refresh rate of the memory modules must be within a certain threshold of
the nominal refresh rate, otherwise the cells loose their charge and you
end up with memory corruption. I believe people were trying to use 133MHz
SDRAM DIMMs in a machine that clocked it's memory at 66MHz, and it wouldn't
work.
I'm not well versed in memory design, but you might get it to work with
some buffer circuitry that kept the memory refreshed properly, even if it's
doing nothing for 7 out of 8 cycles.
2:19 p.m.
New subject: Expansion memory boards (was: Interpreting Vaxstation 3100 memory error)
In article <CAHG8iGU8kEN-WCv7D=vzNUGUDfNZVhaLTD_Hkquv=3d=5gYwEA at mail.gmail.com>,
Jason McBrien <jbmcb1 at gmail.com> writes:
I remember reading somewhere in a memory compatibility
guide that the
refresh rate of the memory modules must be within a certain threshold of
the nominal refresh rate, otherwise the cells loose their charge and you
end up with memory corruption. I believe people were trying to use 133MHz
SDRAM DIMMs in a machine that clocked it's memory at 66MHz, and it wouldn't
work.
Sure, there's an impedance mismatch between the cycle times of the
host system and the required refresh rate of the modern (faster)
memory.
I'm not well versed in memory design, but you
might get it to work with
some buffer circuitry that kept the memory refreshed properly, even if it's
doing nothing for 7 out of 8 cycles.
Yes, sorry I wasn't explicit about what sort of design I was
imagining. I was definitely feeling that there would need to be some
sort of controller on the memory board that mediates between the host
and the modern memory parts -- something to match the timings between
the two worlds.
I wasn't thinking that just hooking the modern chips up straight to
the vintage busses was going to do the trick.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 version available for download
<http://legalizeadulthood.wordpress.com/the-direct3d-graphics-pipeline/>
Legalize Adulthood! <http://legalizeadulthood.wordpress.com>
7:44 p.m.
New subject: Expansion memory boards (was: Interpreting Vaxstation 3100 memory error)
On Nov 28, 2011, at 2:19 PM, Richard wrote:
Absolutely. For something newer, like a VAX with the direct memory bus (I haven't
actually looked at it, so I may well be wrong), it should be fairly simple. For
UNIBUS/QBUS boards (including PMI, as far as I know), the drivers for 120-ohm
open-collector buses are a bit of an issue. You can hack them together with 74AS641 and
74AS760, but I haven't done much testing on either of these to see if they're
reasonable for actual buses (especially really long ones). The thresholds on them are
also not 100% compatible with the DEC standard, though they'll probably do for most
real-world situations.
One of these days, I need to make up a simple test board to stick in my backplane to run
some simple bus compatibility tests. There was a thread either here or on comp.sys.dec a
while ago about some other possible ICs with built-in series terms which looked
interesting... anyone remember what they were?
- Dave
I'm not
well versed in memory design, but you might get it to work with
some buffer circuitry that kept the memory refreshed properly, even if it's
doing nothing for 7 out of 8 cycles.
Yes, sorry I wasn't explicit about what sort of design I was
imagining. I was definitely feeling that there would need to be some
sort of controller on the memory board that mediates between the host
and the modern memory parts -- something to match the timings between
the two worlds.
I wasn't thinking that just hooking the modern chips up straight to
the vintage busses was going to do the trick.
4619
days inactive
4621
days old
5 comments
4 participants
participants (4)
-
derschjo＠mail.msu.edu -
fraveydank＠gmail.com -
jbmcb1＠gmail.com -
legalize＠xmission.com