Several people who have used a dial indicator have noticed that the bed appears to be bowed in the middle. We can set the corners to be the same height, but in between the extremes of the X-axis the bed seems to come up about 7 thousandths of an inch. This apparent bowing is believed to be due to sagging of the extruder support. To test the assumption and to compensate, I installed counter weights that offset the weight of the extruder assembly. With the counter weights, the apparent bed bowing came down to 3 thousandths. With the counter weights in place, I was able to print a large, thin flat surface that had not turned out well without the weights.
The counter weights are very low-tech, but they work. I used pulleys and fishing weights suspended from an enclusure that surrounds the TAZ. To balance the lateral forces in the X-direction, I split the weights and mounted pulleys on the etreme left and right. The extruder assembly is suspended with about 11 ounces of weights on each side. I used fishing line to suspend the weights, so it is hard to see in the attached pictures.
In looking at the design of the Taz 5, I see no reason why we can’t add a second set of X-axis rods and bearings in front of the extruder to remove the cantilever and reduce sag.
I’ve been toying with the idea of replacing the X rods and carriage with a piece of 20mm x 40 or 60mm openbuilds V slot extrusion to counter the middle bow. I don’t know if it’s any more dimensionally accurate than the rods though. Your test there definitly shows there is something going on in the middle though.
What about changing how the set screws holds the X-axis rods, such that two pairs of set screws are used? e.g. An outboard top set screw, and an inboard bottom set screw on each end of the X axis. When you tighten the set screw pairs, you would introduce a counter moment in the X shafts through a force couple, just enough to counter the sag (which you could measure).
When the extruder approaches one end of travel or the other, the sag will decrease (because the rods will want to bow), but so will the counter deflection from the moment as the supports will keep their height at the 0 relative deflection.
See sketch below (greatly exaggerated for clarity):
Your idea is workable, the problem you run into is the scale between the bolts. For a smooth rod installation as a glide surface, the rods are usually either under tension (which won’t work here because the required tension would distort the frame unless you had a brace of some sort) or you would typically have at least 1/4 the length of the rod buried in a rod mount exactly as shown to counter gravity. Or a shorter amount with weights, etc. For a particularily long segment the tension method works better. Going with a box rail, or going with thicker rods like the taz 6 prototype will probably be the most effective fix with the least amount of effort. The thicker rods is definitly going to cost less than the boxed rail approach.
