23 Mar
2009
23 Mar
'09
6:32 p.m.
Hi
My pdp-8/e has developed a nasty fault.
I have all the manuals and a scope but before delving in too deep I thought
I'd see what the 'Hive mind' knew.
Its straight forward enough of a fault. I can no longer toggle in values via
the switch register.
The address sets up ok and the switch register value appears to transfer to
core on raising DEP.
The location counter increments normally.
However going back and checking the locations shows nothing there.
It does not appear to be a problem with the core as replacing the low
address 4k with a higher one makes no difference.
Experienced PDP-8 fixers please comment.
Rod Smallwood
23 Mar
23 Mar
9:52 p.m.
Rod Smallwood wrote:
My pdp-8/e has developed a nasty fault.
I have all the manuals and a scope but before delving in too deep I thought
I'd see what the 'Hive mind' knew.
Its straight forward enough of a fault. I can no longer toggle in values via
the switch register.
The address sets up ok and the switch register value appears to transfer to
core on raising DEP.
The location counter increments normally.
However going back and checking the locations shows nothing there.
It does not appear to be a problem with the core as replacing the low
address 4k with a higher one makes no difference.
Experienced PDP-8 fixers please comment.
When you say you replaced the core did that include the address and inhibit
drivers, as well as the actual core and sense amps? I'm not familiar with the
8/e specifically but web info indicates the 4K mem is composed of 3 distinct
boards, probably distributed as core and sense amps on one board, address
drivers, and inhibit drivers.
Drivers are a common point of failure in core mem systems. All the bits in a
word failing would suggest an address line driver, a single bit across words
could be an inhibit driver.
If it's an address line driver, either a block of sequential words, or a
sequence of individual words separated by the same offset, in either case
coincident with the core matrix dimensions, will be faulty. Determining the
failure range by examination can lead you to the specific driver.
If it's not the drivers what about the core power supplies, as distinct from
the logic supplies?
24 Mar
24 Mar
2:55 p.m.
Rod Smallwood wrote:
Hi
My pdp-8/e has developed a nasty fault.
Its straight forward enough of a fault. I can no longer toggle in values via
the switch register.
Does that mean you can read the current contents of the core or is it
all 0000?
Can you step through memory examining successive locations?
If you re-read them, do they change or is it the same?
When you deposit to a location it reads back as 0000?
My meager understanding of the -8/e makes me think we will need this
kind of information to narrow down the source of the problem. Since core
memory reads are destructive, we can eliminate some core issues if you
say that core can be read repeatedly without it's contents changing. It
could be a problem with the path between the switch register and the
memory data bus.
My two cents worth...
-chuck
4 Apr
4 Apr
4:29 a.m.
Since core
memory reads are destructive, we can eliminate some core issues if you
say that core can be read repeatedly without it's contents changing. It
could be a problem with the path between the switch register and the
memory data bus.
But he wrote that the data displays correctly. The 8/e display
shows the
state of the memory data bus. My first idea would be to check for bad
contacts. And think about the memory control.
And? Can you read out memory? Does it change?
14 Apr
14 Apr
7:36 a.m.
Thanks for the reply.
The list backlog and subsequent mail tsunami means it's a while since I last
had a go at the system. I'll run it up tonight and verify what happens.
Here's what I can remember. In the manual there's a simple test which counts
up in the accumulator. It's entered through the switch register.
First you set the switch register to 0000 (Octal) and press the load
address. Next you set up the first instruction in the SR and then lift DEP
(deposit). The address is auto incremented by one and you enter the next
instruction. After the last instruction is in then you load the start
address and hit clear and continue and it should count up.
It was working but it stopped. If you do examine rather than start you find
the memory locations are (I think) still zero.
The memory is core with four sets of three boards (4K) making 16K total.
Swapping one set of boards with another makes no difference.
The switch reg. must be working because memory addresses can be set up.
I suspect the memory data bus or associated buffers may have something to do
with it.
I'll do a detailed check to-night
Regards
Rod Smallwood
I collect and restore old computer equipment with this logo.
[digital]
-----Original Message-----
From: cctech-bounces at classiccmp.org [mailto:cctech-bounces at classiccmp.org]
On Behalf Of Philipp Hachtmann
Sent: 04 April 2009 12:29
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: pdp-8/e fault
Since core
memory reads are destructive, we can eliminate some core issues if you
say that core can be read repeatedly without it's contents changing. It
could be a problem with the path between the switch register and the
memory data bus.
But he wrote that the data displays correctly. The 8/e display
shows the
state of the memory data bus. My first idea would be to check for bad
contacts. And think about the memory control.
And? Can you read out memory? Does it change?
5736
days inactive
5757
days old
4 comments
4 participants
participants (4)
-
chd_1＠nktelco.net -
hachti＠hachti.de -
hilpert＠cs.ubc.ca -
rodsmallwood＠btconnect.com