[GreenKeys] Another Model 33 Puzzle

[Jones, Douglas W]

I'm having trouble with my Model 33.  The serial to parallel conversion done by the
selector mechanism seems busted:

As typed  as received
10000000  10000000
10000010  11111110
10000011  11111111
10000100  11111100
10000101  11111111
10000110  11111110
10000111  11111111
10001000  11111000

I checked the As typed with a scope monitoring the voltage across the resistor that's in series with the selector magnet.  It shows the right data going to the magnet.  I read the "as received" by watching the code bars while holding down the repeat key.

In summary, the least significant zeros get through, and then the first one that shows up causes all the rest to be read as ones.

It's really difficult to watch the mechanism between the selector cams and the code bars, but here's how things work for each bit:

a) the selector cam turns and at the time that bit is to be sampled, an indent lets the
b) selector lever rise, testing the state of the selector armature.
c) depending on the armature state, the push lever either catches or doesn't catch, so
d) when the cam pushes the selector lever down, the push lever raises the blocking lever
e) which determines whether the code bar for that bit will or will not rise to the left.

The above is a summary of page 53 of (hand numbered 24) of the following PDF:
-- http://www.navy-radio.com/manuals/tty/M33-32_Training.pdf

Cool!  Rube Goldberg couldn't have designed a better system.  Looking with bright light through the hole on the side of the machine (with paper tape reader removed), I can see that the push levers are part of the problem.  They are misreading the data just as much as the code bars are.

This makes me suspect the adjustment of the selector magnet and armature -- I've found the instructions for that.  The work we did over the past weeks involved repeatedly lifting the TU out, adjusting bits of the control character decoding, dropping it in, and testing.  After the last "drop in", suddenly, things didn't work.  Perhaps we bumped something out of adjustment in the vicinity of the selector magnet?

But it also makes me wonder about the potentiometer that's in series with the selector magnet.  I can't find instructions for that, but if it's misadjusted, it would probably make the armature movements marginal.

Suggestions about this area of the machine are welcome.

          Doug Jones
          jones at cs.uiowa.edu