Processor Technology 16KRA Docs?

From: Richard Erlacher <richard_at_idcomm.com>
Date: Mon Sep 18 12:08:58 2000

Hi, Bill!

If you go carefully, starting with noting the current dip-switch settings,
I'd recommend you do the following.

(1) set the dip-switches for the non-functional (suspect) board so they
match the functional one. Install the suspect board with its new dipswitch
settings and perform the same operations on it that you used with the
seemingly functional one. If the behavior matches what you got with the
seemingly functional board, you're home free. The hardware is probably
working OK. You need to figure out your dip switch settings from their
prior settings. It doesn't take a rocket scientist. Fear not!

(2) attempt to isolate 4K blocks on the board's map, i.e. fill 0x4000
through 0x7FFF with zeroes, then read 0x4000-0x4FFF. Are the bytes all
ZERO? If so repeat for 0x5000-0x5FFF, etc for each 4KB block. If one or
more, including all, of these blocks fails to read all zeroes, fill 0x4000
through 0x7FFF with 0xFF. Read each block, making note of the results.

If one of the 4K blocks fails to clear a single bit, that's cleared in
another block, there's probably a failed IC or possibly a broken or
intermittent enable or write command strobe in that RAM block. Make note
and proceed. If one of the 4K blocks produces 0xFF but failed to produce
zeroes, you've likely got a defective decoder or an intermittent or open
write command, output enable command, or chip select to that block. If one
bit is consistently wrong, it may be the memory device itself, but the stobe
lines should be checked nevertheless. Memory IC"s are not much more likely
to be at fault than logic devices or PCB taces.

(3) now get out your meter and verify that all the memory IC's in a given
bank have continuity on their address, select, write, and enable lines. If
that seems correct, verify that the DQ or D and Q (if they have separate
I/O) of all the memory IC's on a common data data bit have continuity as
well. It's unlikely they'll be buffered separately. If these strobes are
hooked up similarly to each device in a bank, and the data bits are hooked
up similarly for each data bit in all four banks, then you should look for
the TTL buffer to which they're connected. On boards of this age, these are
likely to be 7436x's where x can be 5-8. it's most likely they'll all be the
same, so I'd look for '367's using some combination of the groups of two and
four bits that are enabled by each device enable. The S-100 has separate
data in and data out busses, so there will be two sets of buffers. These
buffers can be inverting or non-inverting, just so the same type is used on
both busses.

(4) now get out your oscilloscope and set up to examine coincidence of pWR*
( that signal was used pretty early on, though it had several names.) and
the write strobe on your memory devices. If you see no activity on the
write line when executing an infinite loop that writes to only one location
in one of the banks of memory on the board. Find the right one by 'scoping
the write lines on each row of 8 memory IC's until you have the one that
goes LOW when the pWR* is low. Since you're in an infinite loop doing only
one write to memory and a subsequent read of the same location, you should
have no trouble finding the strobe if it's there. If you can't find it on
the suspect board, you've probably found the problem, but now you should
substitute the "working" board to ensure you've found the right signal.
After ensuring you know where the select, output enable, and write command
strobes are (which you can verify from a diagram of the device pinout) you
should be able to verify that each of the strobes in question shows up on
the working board as it should. Now go back to the suspect board and check
each memory device to see which ones get the appropriate signals and which
don't. Once you've established that the strobes appear where they should,
look for the data to see that the correct data arrives and departs as it
should. If that fails, you've probably found a defective memory IC.

(do you see a pattern here?)

If the strobes arrive at the correct time and the data arrives at the
correct time during writes, (check setup and hold times, but those shouldn't
be an issue if the board design works at all) and the correct data doesn't
arrive at the CPU, then either the data-in bus buffers are defective, in
which case the board won't work at all, or the select/enable logic is
defective. If these parts are socketed, you can learn quite a bit by
swapping them with parts from the board that works.

Once you've got two boards that work all you have to do is find out how the
switches work. As stated earlier, it doesn't take a rocket scientist to
establish that.

Dick


----- Original Message -----
From: Bill Sudbrink <bills_at_adrenaline.com>
To: <classiccmp_at_classiccmp.org>
Sent: Monday, September 18, 2000 9:12 AM
Subject: Processor Technology 16KRA Docs?


> Hi,
>
> The Sol I just got came with two 16KRA memory boards.
> I didn't get any documentation for these boards.
>
> One seems to be occupying the address range 4000-7FFF.
> This board seems to be OK. I can enter values and dump
> them back out.
>
> The other board seems to be addressed from 0 to 3FFF.
> Without the board in the system, dumps from this address
> range show all FF. With the board installed, dumps from
> this address range show values other than FF. Entering
> values into this address range does not seem to work.
> The command works, the machine does not hang, but the
> values displayed by the dump command don't change.
>
> There are two, eight bit dip switches on each board.
> The settings are very different on the two boards.
>
> Can anybody help me out here?
>
> Thanks,
> Bill Sudbrink
>
>
Received on Mon Sep 18 2000 - 12:08:58 BST