Where are you located? would loaning you an extra board set help?
On Sat, Mar 1, 2014 at 7:07 PM, Rick Bensene <rickb at bensene.com> wrote:
It probably doesn't matter that much, but I
meant take it to the bare
minimum- no EAE, no timeshare, 4K only, etc. Sorry it didn't help.
No worries -- I appreciate the input!
I did think that you meant stripping it down to the bare minimum, but
having OS8 running makes loading the diags and testing much easier and
faster -- and all that needs 8K, Mem Ext/Timeshare, and the RX8E.
I don't want to remove the EAE..I just am scared of messing with the CPU
At least I was able to eliminate the high speed paper tape interface,
the MOS RAM, and the second serial port, and get it all down to one
Omnibus, which is a step in the right direction.
I could strip it down to the bare, but then I'd be fussing with loading
RIM, and then BIN, then pushing papertape images of the diags to the
machine over serial -- it's just tedious and slow. I may give this a
try later, but in light of the information below, I will probably try
something else before I strip it bare.
I just now put the machine back to the original configuration, and
nothing changed -- the problematic behavior with the RK8E still exists.
David Humphries had mentioned in a posting here that he had observed a
transient on a signal related to loading the current address register in
his RK8E that exhibited similar symptoms to mine.
While I didn't know the specific signals that David was scoping, I
guessed, from looking at the schematics, that perhaps he was monitoring
the LOAD and CLK signals that drive the three 74161 counter ICs on the
M7105 board that make up the Current Address Register.
So, while putting the system back to its original configuration, I
tacked in wire extensions to the LOAD and CLK pins of one of the 74161s
to bring them out so I could scope them. Since the M7105 board is the
middle board of the three-board RK8E sandwich, the only way to get
access to the signals is use wire extensions.
Once I got the system put back together, I toggled in a small program to
repeatedly load the current address register
in the RK8E with zero, e.g.,:
0200 7200 CLA CLL
0201 6744 DLCA
0202 5200 JMP 0200
I started this program running and watched the scope.
There is a definite spike showing up on the CLK line.
This spike could indeed cause the counter to increment after the initial
address is loaded into the counter using the 6744 IOT (Load Current
The photo of the scope traces I observed is posted here:
The top trace is the LOAD signal, and the bottom trace is the CLK
signal. The negative-going spike on the CLK line is clearly obvious.
The traces look quite similar to the scope image that David posted.
I am hoping that David will read this and find his notes on what he did
to fix the issue (installing a 7474 flip flop somewhere to eliminate the
spike) and post it so I can make a similar modification and see if it
fixes my problem. It isn't intuitively obvious to me where a flip
flop/latch tacked into the logic would get rid of the spike.
It's my hope that this is what is going on, and that David's fix will
fix my RK8E/RK05 problems.