While drawing frame in last of the big drawing attempts i observed the paper shifting. I guess movement being close to edge thus biggest leverage, having opposite movement directions on the sides, and doing so multiple times was the worst case scenario.

To minimize this I wanted to try increasing the force from top rollers. This made me return to making springs, but this time extension springs.  Using a 3d printed jig after a little bit of practice I got somewhat decent looking springs. https://www.printables.com/model/468530-extension-spring-jig To better accommodate increased forces I beefed up some other parts of top roller mechanism, which accidentally made them look like birds.

I didn't test yet how much it helped with that kind of paper shifting. But it seemed like the increased force made the problem of slightly misaligned top rollers steering paper sideways and thus bunching up between multiple rollers.  I am not fully sure that the bunching up is caused only by misaligned top rollers and not the spiral wrapping of bottom rollers, or bottom roller misalignment.

First test to try confirming that the problem comes from top rollers was replacing them with plastic spoons. Turns out in the last few years single use plastic utensils have been more or less banned. I don't mind that, but it made this step a bit more trickier than I expected.

Next step was making 3d printed arms with a spherical sliding pads, covered with HDPE tape for less friction.  After testing it properly that was a failure. Unless I set the spring force very low, motor would skip steps. With pads pressing sufficiently lightly not to stall motors, even lightly holding the paper with a finger was enough for it to shift and rotate.

Iteration v10 was wider rollers wrapped in HDPE tape.  Somewhat worked but bunching up was still a concern.  In one of the tests I switched to a new crisp sheet of paper, from the old crumpled and flattened one. It started to work much better. This made me realize that there is a lot more to the paper rigidity than just g/m^2.  Since the paper was cut from 700mm roll (and the plotter can handle maximum of ~690-680) I had to feed it sideways so that paper want's to up from left and right sides, instead of front and back like you would get if roll was sufficiently narrow to be fed directly.  With the fresh paper things went much better than before, but as I was moving it back and forth at one point suddenly combination of way paper was curling and folding when hitting the wall behind caused paper to be pulled in folder over itself.  The second lesson here was managing how ends of paper fold is important. Minor sideway pull from the paper folding badly can cause paper to be pulled in wrong and make a crumpled mess.

So I switched from having the plotter on my table, to placing it on pair of stools. That way paper can freely hang down on both sides without anything getting in the way. I guess I will also have to take into account direction in which paper wants to roll. This is slightly annoying because material often comes in 700mm wide rolls which is a tiny bit more than what plotter can handle and I will have to cut of a centimeter or two from one side.

In parallel with previous tests I had order some PTFE tubing. So that I can replace HDPE wrapped rollers with PTFE. That would not only reduce coefficient of friction but also eliminate seam which the tape wrapping had.

Things to try out next:

• Actually test if changes had any effect on shifting while repeatedly drawing the big rectangle
• Make little shelves before and after rollers to help guiding the paper.
• Experiment more with using only one row of rollers when using a pen.
• Try increasing the spring force again
  

Bonus content:

An exploded view and one of schematics drawings for Tektronix 2215. Paper size 68x54cm. Margins could have been smaller but it was late already late and at that point I didn't want test how close to the sides I can get reliably.