Found that the my circles still exhibit the exact same flat edges.
The flats are where the edges of the Y axis movement are most subtle.
If I watch the printer I can cleary see the extruder nozzle moving almost straight down the X axis (left to right) at those ploints.
Error is the same regardless of how I slice the cylinder model so the error is still in the printer somewhere.

I'm going to try to calibrate the extruder perhaps it's still extruding too much per step.
If that were the case as the printer slows down at the crest of the arc there would be some overage and it always prints the cylinder walls
the same direction so the error would be similar on each layer.

Turns out asking for 100mm of filament into the extruder gets 117mm of filament.
Time to start working on that calibration and probably upgrading the firmware in the process.

In theory that should reduce the backlash.

So your talking about the shaft on the Y motor not the drive rod above it? Mine is super tight and probably will strip if I try to remove. I like the idea of marking the pulleys for slippage

What are fairloc? Why are they better?

I removed the shaft above the Y motor (the shaft the Y motor turns and through that moves the axis).  However now that you mention it that stepper will be removed and a flat will be put in that output shaft as well before I am done.  I already (in another topic) ground a flat in my extruder stepper with excellent results (at least for me).

FairLoc(TM) hubs are designed to reduce the exact issue you pointed out earlier in this topic.  They are essentially a compression hub made out of aluminum but unlike the set screw they grab most of the shaft circumference.  SDP-SI sells them (this idea is patented) as alternatives for gears, pulleys and other applications where non-marring adjustment of that item will happen often.  There are other ways to do this, and FairLoc(TM) does require some regularity in the shaft dimensions, but in this case I can substitute these gear pulleys for the ones on the shaft without cutting a flat at all.  Effectively it is a drop in replacement.  As those 2 gear pulleys were obviously heavily tightened because without a flat they would otherwise slip.  My hope is with the FairLoc(TM) hubs I won't have that sort of issue.  Course as proprietary parts they cost a bunch more.

I bought 3 18 tooth FairLoc(TM) hub pulleys at basically $19 each.
I won't need one for the stepper because I will put a flat in that shaft, there is no rotational misalignment there that requires adjustment...just the need to stop it from slipping so the $5 existing stepper gear pulley is just fine.

What are fairloc? Why are they better?

I removed the whole 6mm shaft at the rear and of course discovered that the gear pulley set screws were turned down so tight they dug into the shaft itself.  This is unlike the extruder 'gear' which was unable to dig into the that stepper's output shaft.

What is interesting is that the drive gear pulley was apparently set into place 3 times.  It has 3 distinct landing area on the shaft.  It was interesting getting that drive gear pulley off as well as all those set screw landing spots made it hard to slide on the shaft.

If anyone needs the the MXL drive belt from the Y axis stepper is marked on my system.
It is an SDP-SI part number: SDP A6Z16 084 025/060
That should be actual part number: A6Z16M084060 (the M is implied, 84 is the number of teeth, the 025/060 are SAE/metric belt width).
This is just a fiberglass reinforced neoprene belt so a steel or Kevlar reinforced belt would be an improvement.
I'll set about ordering those today.

I will clean up this shaft and try to reuse it with my inbound FairLoc(TM) replacements.  Given this is a recommended and seemingly common enough adjustment the FairLoc(TM) should probably have been the first choice as they are the least likely to damage the shaft if properly tightened.  When I use them I will mark the shaft position with regards to the gear pulleys so I can visually detect any slippage.  I can infer from the lack of futher damage to this rear Y axis shaft that the previous owner did not actually achieve this calibration otherwise the shaft would have more landing areas from set screws or marks from the set screws slipping.  Probably stripped out the head and gave up on it.

If that doesn't work then I guess I set about making a better axis.

going from memory here....I get back in the other shop in a bit.

Nope... I have 18 too written down, so they are 18 tooth... well so much for my memory

Not sure why I even thought they would be different, because then they would be a different size???

They are idlers so it shouldnt matter anyway

It might impact the belt performance a bit as too much change from the stock tooth count might change the angle of the forces on the belt.

I will soon see if upgrading the 3 gear pulleys on the rear axis drive makes this better as I altered my order to get 3: 18 tooth, MXL, FairLoc(TM) hub, 2 flange pulley.

In theory they should not slip on that shaft if the set screw was almost good enough and I rid myself of having to flat the shafts to do it.  Course they did cost almost $20 each.

Correct on the 6 and 5 mm sizes I don not have part numbers with me right now but it really isn't that hard to figure most of them out... nothing is 'special' just about everything is 'off the shelf'

The front gear pulleys are also 18 tooth right?

true

I could see making a modification for this.  However it would make the machine taller and given how the frame is integral to the system I'm pretty sure some cutting would be required.  Though maybe there's a way to do it.  Will have to look at the system more closely.  I'm pretty sure the cover would need to have the top opened up to make room for this even if the frame could be worked around.  Unless some Z axis working range was sacrificed (what fun is that?)

Thanks for the info. yes I Agree driving both sides and reducing the work space is standard but for such a small area and to reduce the binding headaches and cantilevered system could apply here for such small loads and forces? Would you agree?

For such a small workspace I don't see why they did not pull and push the X carriage on the support rods from directly over the top in the middle.  If you think about it if you were directly over the center of the X carriage there would be no twist at all and the machines are not open at the top in the enclosure anyway.

It would not have been all that hard to fabricate a bridge that is high enough over the X carriage with the extruder and filament feed so that it could traverse under it and still attach to the axis above.  You could easily make it with polycarbonate or even laminated birch plywood.  Just print corner blocks.

Plus this way you replace the round rods on the sides with square stock.  Use real ball bearings to run up and down them.
The result of driving it this way is like pushing a little car from the top center.  So you put the little car on tires and those tires are the bearings.