[R-1051] Easy fix for sloppy code generator

[David Wise]

My MC knobs didn't always work.  I found that there was maybe five degrees of backlash;
after turning clockwise, I had to turn that far just to start moving the switch rotor back the other way.

The root cause was a design weakness in the mechanical coupling between knob and switch,
and a technician who didn't read the manual.

The Code Generator is a removable module, and like the A4 RF Amp and A6 Synthesizer, it uses
a slotted-flange mechanical coupler.  The flange has a slot in its outside edge, and a lozenge-shaped
hole in the center.  The disk that mates with this flange has a pin that goes in the outside slot, and
a projection in the center that goes in the lozenge.  These work together to transmit torque with good
angular accuracy even when the shafts are not collinear.  The A4 and A6 modules have a one-piece
disk with a locating boss in the center, and the disks mate with flanges in the mainframe.  The
CG module has flanges, while the front panel has disks..  The MC knobs are setscrewed to
shafts which extend through the front panel.  Behind the panel are detent star wheels, which are
mildly press-fitted onto the shafts.  Each wheel also acts as the disk side of a coupler, with a
pin, and using the shaft as locating boss.  And there's the problem, because the wheel is not
setscrewed or pinned in place; as I said, it's just a moderate press fit.

The manual's procedure for reinstalling the CG is easy to misinterpret.  It ends with "Reassemble
by reversing removal sequence."  It is clear to me that this induced someone to bolt the module
in place, then rotate the knobs to achieve registration, just like A4 and A6.
But this skips a critical step, to wit: You must get each boss into its corresponding lozenge.
If you don't, then when you tighten the mounting screws, the flange will press the shaft forward
until it's flush with the disk, leaving it without a locator.  As a result, when you turn the knob, the
flange will walk sideways until its outer edge fetches up against the recess in the star wheel.
Only then will the coupler begin to transmit torque.

Happily, it's easy to undo this gaffe.  Using a 3-inch C-clamp as a press, move the shaft
rearward until it once again projects from the wheel.  (You'll also need a block of something as
a spacer - a 3/4" socket is perfect.  Otherwise the clamp will mash the detent spring.)  And
when you put the module back in, maneuver it around with gentle finger pressure while rotating
the knobs, until you're *certain* that both couplers are fully mated.  Only then tighten the screws.

HTH,
Dave Wise

PS - The procedure has you dismount shield plate MP69 to reach captive screw A2A7H1.
I didn't have to do this, but it's best to use a screwdriver with a round shaft and no larger than
needed, to avoid gouging cable insulation.