MK3 Titan Aero mount  

MK3 Titan Aero mount

Hello,
After a bit of printing with my MK3 I tried to remove the "trailing waves" after sharp direction changes you can see on almost all prints.
The biggest problem with the stock layout is the distance between the extruder motor and the frame, which means that a fairly large momentum wil twist the frame on every direction change. My goal was to move the cog of the extruder assembly as close to the frame as possible,while still being able to use the pei sheet entirely.
All parts were designed and modelled in openScad, sources here.
My design pushes the x frame rails 70mm apart and mounts the stepper motor between them.

For some reason Slic3r doesn't want to slice the carriage at 0.2mm layer height, as it either fills it completely with bridging or leaves out a layer depending on the version. 0.25mm works fine though.

These are the parts I thought were needed, but the cable guide doesn't work:

With the belt mounted on the rods. The inside is a bit flimsy but it will be supported when the stepper is installed:

With the Stepper installed:

This is how it looks with the complete extruder:

Clearances are a bit tight, but everything fits. I notched out x-end-motor and the x-carriage to be able to move the nozzle to x=0

My belt tensioner;

Obviously this isn't finished, I still have to adress cable routing and parts cooling, as the fan is just moving a bit of air in about the right direction.
Also I have to redesign the heatbed cable cover, as it interferes with the carriage during calibration. You also have to file down the centering pins for the steel sheet and the solder of the heatbed cable, to provide clearance during calibration.

In the firmware I had to make the following changes:
#define TMC2130_USTEPS_E 16
#define X_PROBE_OFFSET_FROM_EXTRUDER 21
#define Y_PROBE_OFFSET_FROM_EXTRUDER 9
#define DEFAULT_AXIS_STEPS_PER_UNIT {100,100,3200/8,838}

Here are the first results. In the first picture the left benchy is the old one, in the second one the right. Both were printed with the gcode supplied with the printer and original prusa PLA:

You can see that part cooling is insufficient now, but the trailing oszillations are alost gone entirely, even in y direction, which i dint't think would happen.

I will continue testing and updating here.