From my previous post, when I first built my Jubilee my mesh showed a twist, I used a granite tile as a known flat surface; the resulting mesh can be seen in the photos below.
I had resolved the issue mainly by physically pulling the front frame into alignment for more detail please read my previous post. This worked but after a while the misalignment would return, it was not a permanent solution.
To try to put this issue to bed the first thing I did was acquire a steel plate for the bottom of the machine. I am not sure how much the plate contributed to the solution or not. But I had to completely drop off the bottom frame to get in installed which may been the primary contribution.
After reassembling the machine I ran a bed mesh I was surprised to find that largely the twist has been resolved. I also designed and had a custom acrylic panel cut for the "Front" of the machine which when finished with the adjustments I use it in attempt to lock the squared frame after getting a desirable mesh.
Due to some abuse over the lifetime of the machine the mesh run on the granite shows a variety of issues that can be seen below. Looking at this image can tell us a lot. Points to remember when analyzing this mesh is that the 3 points used to tram the machine should be 0 in this mesh. With the origin in the front left corner in this image, that will make middle point closet to the camera and the corner points in the back the tram positions. When debugging the mesh I like to set the grid to 5x5 its a good mix of quick and high enough resolution to see the issues.
The most notable feature here is "Taco Bed", but we can also see there is some twist about the center point as the right side is noticeably higher. Since we know the granite tile is flat all of these defects are in the motion system due to a couple crashes. I had a Z switch fail during a homing sequence, the wiring to the switch would create an open circuit if it flexed to much I found this out the hard way.
To resolve this, (Note: its important to remember that the mesh is inverted, defects are essentially a reflection of the motion system, so a low spot like this really means the bar is bent up because the bed has to travel up further to hit the switch.) I moved the cross bar to the center of the machine then being carful where and how I put my weight onto the machine I pressed down.
Do this slowly try not to solve it in one shot instead go easy then rerun the mesh and analyze the change repeat this until the taco is gone. If you go too far the taco will turn into an arch. I lost my intermediate mesh captures but after resolving the taco I had this as my mesh
This mesh is turned 90 so we can more easily see the next most notable feature. It appears that rail on the right side of the machine is bent up as the mesh sinks low here by a slight amount. I did not install the acrylic panel on this side of the machine, I opted to for corner brackets as I like to access the machine from this side, my best guess is I think that allowed this side of the machine to get bent slightly along side the cross bar during the collision event.
To resolve this I did the same as the cross bar, put my weight dead center on the rail after moving the gantry out of the way, I think it took a little extra effort to straighten this side out.
After a few more meshes I managed to flatten out the right rail pretty well (left from the rear viewing angle here) Next I attempted to use a cable and a turnbuckle to tweak the last little bit of twist out of the frame. I used an extrusion nut with a wire attached and slid it into the ends to the aluminum frame with a turnbuckle in the middle. I will try to add better picture of the jig in the future but I attached the nuts and wire like this.
With the panel loose I slowly tweaked the turnbuckle and ran a mesh after each adjustment. There wasn't much of a twist at this moment to tune out but it did work. Arguably there is not a lot of difference but this is the final mesh it at least did prove the concept worked. At the end of night I locked in the panel.
Final thoughts. I think this method works well but I would be hesitant to use it often if your frame is get getting out of alignment frequently there may be root causes to look into, for me there was at least 1 big crash and a few smaller ones that got me here. It is a good idea to make sure your frame is tight periodically and to routinely check over any wires that are flexing especially those as important as a the z switch.
Also I think I will be adding the turnbuckle onto the machine as a permanent feature. After mostly sitting for a the last few months as I dealt with some firmware and personal matters I was surprised to fine this as my mesh.
The panel did not lock in the frame as well as I had hoped. I think a little of what is happening here is my cats found machine to be a good highway to higher places and after getting jumped on repeatedly here we are again. (Note the ripples are do to the fact this was off a mag sheet with a print surface not the granite tile.) I will add an addendum to this post once the job is done.
