behaviour of classic PDP-8 frontpanel

[Klemens Krause]

On Tue, 30 Oct 2018, Brent Hilpert via cctalk wrote:

>
> If you haven't tried it already, a good exercising with contactr
> cleaner (stay away from the stuff with oil in it) or isoprop will often
> clear up excessive contact bounce on old switches. Getting contact 
>
May be, but this would only solve one of the problems. In many cases
contact cleaner works some time an then problems come back, because in
the past 50 years the original silver platening of the contacts has gone
away.


> For a proper design fix, if the switch has both NO and NC contacts, a
> solution might be to build an SR flip-flop from two transistors and a 
> few R and interposing that between the switch and logic as a de-bounce 
> circuit.

Yes this would solve one problem. But I don't want revise the logic of
such a historic machine.
And the other problem wouldn't be solved: the machine stops if any of
the keys <START> <DEP> <EXAM> <CONT> is pressed while the machine is
runnig.
A look behind the panel shows:
<START> <STOP> <SINGLE STEP> <SINGLE INSTR> are single wired with the
N.O. contacts of the switches. So they have open inputs with a
potential of -1.9V <START> or -3.9V <ALL OTHERS>. Activting one of
these keys makes contact N.O. to ground.
The other keys <CONT> <EXAM> <DEP> are connected to the middle of the
N.O - N.C switches. Inactivated N.C. is connected to -15V, bringing
this level to the inputs. Activated, the N.O. makes connection to
ground.
<PANEL LOCK> is simply a switch, which interrupts the common ground-
line of these switches.
The better solution would be a feed back line, which holds the former
ground line to 0V if the "RUN-FF" is reset and puts it to -3,9V or
-15V if "RUN-FF" is set. The first connect of the start key would
start the machine and set "RUN-FF", thus locking the switches by the
processor itself. Subsequent contact bounces would be ignored.
Inadvertent pressing one of the other controlswitches would be ignored
too.
Naturally <SINGLE INSTR> <SINGLE STEP> and <STOP> remain hard wired
to ground, to allow stopping a running machine.
But this improvement would need a siginficant rewiring of the front-
panel which I don't want to do for historical reasons.

Other writers in this thread wrote, that:
> On our Straight-8 at the museum, hitting START appears to cause the 
> running program to hiccup (I assume it's re-STARTing but who knows from 
> where).  I was
Nice word "hiccup", :-) probably the same effect than at our machine.
So this seems to be normal.
May be that was corrected in later machines with higher serial number?
Our machine has serial number #768

More behaviour, that I tested:
<DEP> stops the running machine and puts contents of SR in the actual 
address that is in this moment in the MA-register. 
<LA> stops the machine and loads SR in MAR.
<EXAM> stops the machine and shows content of current address in MAR.
<CONT> has no effect.
No one of theses switches ever cleared a memory word. Eventually they
are synchronised with the memory timing chain, whereas <START> works
asynchronously and clears most or all FFs.

Klemens