PDP-12 Restoration at the RICM

[Jay Jaeger]

On 10/12/2015 12:40 PM, Michael Thompson wrote:
>>
>> Date: Sun, 11 Oct 2015 18:44:54 -0500
>> From: Jay Jaeger <cube1 at charter.net>
>> Subject: Re: PDP-12 Restoration at the RICM
>>
>> Don't forget about the other more remote possibilities:  cables,
>> backplane, bad wrap, supply voltages at the actual card(s) for the
>> mis-behaving channel, etc.
>>
>> JRJ
>>
> 
> We used different control and data cables for the TU55 and the TU56 drives
> and observed the same track 3 bad behavior.
> 
> The backplane appears to be in good shape.
> 
> My scope had a little trouble looking at 10-15mV signals in differential
> mode using the math functions, but we looked at the head signals going into
> the track 3 amplifier, and they looked reasonable.
> 
> The power supply voltages at the cards are within spec. The track
> amplifiers are supposed to be differential, so they should be fairly immune
> to power supply noise. We plan to connect a lab supply to the backplane
> near the track cards and adjust it slightly higher than the PDP-12 power
> supply. That should clean up any 60Hz noise on the power. Maybe that will
> help?
> 

I had in mind the power to the cards in the PDP-12 itself, not the tape
drive(s).  No local sense in those days, with considerable voltage drop
in the wiring.  A bad wrap might add to that drop on the backplane itself.

> We have swapped everything else between the tracks, including the logic
> analyzer probe, and the issue always stays with track 3. Maybe it is a
> backplane wiring problem?
> 

It is a possibility not to ignore.  The reason I asked is that at Wis.
DOT we had an Amdahl 580 that was acting flaky - odd crashes without any
apparent connection.  The machine was new, so Wis. DOT threatened to
send it back.  Amdahl flew in something like 3 engineers from the
Sunnyvale plant, and they along with the 2 local FE's and one CE went
over the machine with magnifying glasses (literally), looking at
everything.  They finally found one amazingly tiny coax wire (imagine a
coax cable about the diameter of a normal wire wrap wire) that had been
nicked/kinked by being pinched against a frame at some point - took them
almost all night to find it.  They were happy but pretty "ripe" when I
came into work the next morning.

Based on that story, it occurred to me that corrosion on a wrap on the
PDP-12 backplane could be an issue.

The other possibility that occurs to me is a timing glitch - where two
signals change state at the same time and then are sampled too close to
that time, resulting in a signal being clocked into a latch the wrong
state.  Those can be really hard to track down.

JRJ