Adjusting the rotate character selection on the 1053 - final round part 3

BALANCE ARM ISSUE AND CORRECTION

The parts didn't slide far enough when I tried to use the screwdriver on the slots on the top of the arm, but by pushing the two sides I was able to get the two versions of zero rotate to match. 

REDOING OTHER ADJUSTMENT STEPS

I retimed the print shaft and tweaked the lever point of the rotate arm as shown below.

What I discovered is that the amount of rotation is enough to reach all eleven positions, from +5 through to -5, but the detent which locks the ball in place was not working reliably. Part of that is a timing adjustment I have to tweak, as currently the detent stops the ball while the machine is still moving the rotate tape to its final position. This can lock the ball into the wrong column.

NAIVELY ASSUMED THE TILT ADJUSTMENTS WERE CORRECT

I had been assuming all along that the tilt selection was correct. First, the machine came with the rotate tape broken but the tilt tape was intact thus presumably set up properly. Second, I could see that the machine moved the ball to four distinct tilt positions as I selected characters during my time working on rotate, so I just assumed it was doing it correctly.

I had been working on a flaky detent during rotation, where sometimes the rotation detent didn't enter the tooth fully and would snap in place when I lightly rocked the ball. I had been inspecting for burrs and lubricating it when I realized that in fact this was caused by a tilt misadjustment, 

The detent system first locks in the tilt and then continues on to lock in rotation. Tilt is simple, just four teeth inside on a pivoting arm, but if the tilt arm doesn't fall into a tooth then the rotation detent can't pull up into the ball teeth. 

I found that the selection was falling right at the high point between teeth, so that sometimes the tilt locked in properly but other times it hung up on the high point. The ball seemed to be tilted but it was in between the rows. 

The adjustment was super easy and now the tilt locks in properly, thus the detent for rotation is reliable. 

GOING BACK TOMORROW TO READJUST ROTATION 

I am feeling confident that I will get this 1053 dialed in with proper character selection. It is just a matter of iterating through the settings. In spite of step by step procedures published by IBM, which imply that these don't interact with each other, I have found that I have to repeat adjustments or loop through sets of adjustments if I want to get things working as they should.