[GreenKeys] BRPE wiring

[Jones, Douglas W]

In going over plans to make a new cable from the U of Iowa's PDP-8 Type-75 high-speed punch interface to our BRPE punch, I started worrying about the current flowing.
 
Pin 18 on the Amphenol micro-ribbon connector on the back of the BRPE punch carries the entire -28V supply to all 9 solenoids (on an 8-level machine -- all 6 on a 5-level machine).  Each solenoid draws 1A (on the 28V version of the punch) during its 5ms feed pulse, and feed pulses come at 110 per second, or about 9ms apart.

So, the peak current through pin 18 is 9A on an 8-level machine, and the average current is about 2.5A assuming random data.  On a 5-level machine, the peak is going to be 6A and the average will be much lower, particularly if Baudot code is used -- Baudot tries to minimize the number of holes punched -- common characters are coded with fewer holes than uncommon ones.

The problem?  Amphenol's data sheet for the micro-ribbon connectors says that the maximum current through any one contact position is 5A.  Punching a run of rubouts on an 8-level machine at 110 per second pushes the average current to just about this.  It's always better to have a safety margin.

The story is even worse at the computer end of the cable, where a card-edge connector is used.  There, it seems the maximum curent is closer to 3A

Note that the averages drop by almost a factor of 2 on a punch geared (or belted) at 60 characters per second, and if you have the older vacuum-tube version, the currents fall by a factor of 6, meaning there is no problem at all with pin 18.

As far as I can see, nobody used pins 15, 16, 17 and 19 on the BRPE data connector.  DEC used pins 13 and 14 for a modification they made for remote control of the motor, used on the PDP-7 and perhaps some other early machines.

I've modified my BRPE to put pins 16, 17 and 18 in parallel for the -28V line.  This makes it easy to put 3 wires in parallel in the cable in order to lower the resistance -- 10 feet of 22AWG ribbon cable really needs wires in parallel to handle 9A pulses, and each pin on the Amphenol connector can fit just 1 22AWG wire.

I would recommend this change to anyone contemplating using a BRPE punch for 8-bit data.  Does anyone know if my proposed use of pins 16 and 17 conflicts with anything anyone else has done with BRPE punches?

               Doug Jones