Repeatable positional accuracy of z-axis

I'm having a heck of a time getting my Mendel90-based printer calibrated, specifically in the z-axis. I can get the bed level in the X and Y axis +/- .02mm or so which is within the limits of my Harbor Freight digital dial indicator (rated accuracy +/- .001" if the literature is to be believed). I can manually lift and release the indicator's plunger and the readings are at least repeatable from the indicators use.

As soon as I raise up and then lower back down the z-axis, the indicated distance changes. In theory, I'd expect it to be the same at least to the resolution of the indicator. For a cycle of going up 10mm and back down, it changed to .05mm (so it went down .05mm further) when I expected it to be 0 since I zeroed it out before the test. For an cycle of 30mm up then down, it was .17mm. It seems to be linearly repeatable for a given distance, so for instance it didn't matter if I told it to go up and down 10mm once, or if I did 5 cycles of up and down 10mm resetting inbetween, it was always .05mm. But if I went up 30 in one shot, then back down, then I would get the .17mm result.

Since the Mendel90 design homes the z-axis at z-max, I think this is contributing to my difficulty of poor bed adhesion because it's anyone's guess what the actual height is after it's homed and then travel down ~200mm. I've tried traveling very slowly, medium speeds, and reasonably fast z-axis travel speeds. I've tried having steppers hooked up in parallel via the RAMPS z-axis connector, in series, or dual z-axis drivers using the unused 2nd extruder driver on RAMPS. I've tried the computed full current for the driver(s), I've tried 70% of the rated current in case the motors were overheating, and I've tried over driving the drivers.

The one thing that I haven't tried is just running the stepper without any load on them to see if they are positionally accurate in a repeatable manner, eliminating any other mechancial issue such as with the M6 threaded rod binding, backlash, whatnot..

I'm kind of at a lost as to what else to try next. Any ideas? What is a reasonable and expectable z-axis accuracy?

Edited 2 time(s). Last edit at 04/03/2014 02:24PM by cdru.

Check your firmware for steps per mm on the Z axis. If this is not a whole number it may be the reason you're getting inconsistent results. Try the test again with a whole-number multiple of the steps per mm setting.

What type of Z-couplers do you use? If there is any twist-flex in these it would account for the issue.

Quote
3dkarma
Check your firmware for steps per mm on the Z axis. If this is not a whole number it may be the reason you're getting inconsistent results. Try the test again with a whole-number multiple of the steps per mm setting.

No, it's currently set to the nophead's default of 200*16/1. 200 steps per revolution, 1/16th microstepping, 1mm per revolution of the M6 threaded rod.

Quote

What type of Z-couplers do you use? If there is any twist-flex in these it would account for the issue.

The couplers are printed as part of nophead's design. They are pairs of these that sandwich the threaded rod and motor shaft. The motor shaft currently is cushioned to allow for some alignment inaccuracies by tight fitting nylon tubing. I actually have some replacement lead screws, nuts, and real couplings coming from Misumi in the next day or two as part of their free $150 promo. I guess that could be it, but it seemed to repeatable. If it twist flexes in one way, why not the other? Or why does the problem seem linear the farther I travel the more it's off when it returns? I'd think a twisting flex issue would show the same error margin whether it traveled 10mm or 30mm.

It sounds to me like your z-axis is sliping within the coupler. I have had this issue myself before. Make sure you tighten it as much as you can.

You should be able to observe whether the steppers are positionally accurate. If your couplers are tight and do not slip, make a note of their angle prior to moving up and down. If the angle is not the same after the test, they're not positionally accurate.

Do you have any anti-backlash measures installed?

I'm having a similar issue with inconsistencies in Z-height. I was curious if you figured out what was wrong. In the meantime, I'll try some of the suggestions in your post.

I switched from metric threaded rod to a true leadscrew, nut, and non-printed couplers from Mitsumi. That went a long way to helping however life and a few othe projects have gotten in the way since then so I haven't had much time to work on it to say if the problem is fixed. I'm trying to get back to it in the next week or two so I'll try to remember to post back.

z height constantly changes depending on temperature and length of lever for z axis switch. to increase accuracy remove the lever on the z axis switch and use the button contact on it. also only home z with bed heated up. This will increase accuracy. also make sure that the switch is reinforced so it does not flex. printing fails can happen if z is off by more than 0.05mm, another issue is stepper speed on z axis, check to make sure z axis is lubed and not binding, if all else reduce speed by 50% on z axis and this should help, also make sure that the z axis is not sliding out of place on the threaded rods, mark the point where it goes into the coupler with a sharpy. if it slides you will notice it. this is less of an issue with mendel 90, just make sure that z axis is all the way in. also mark z axis with a line that goes thru couper to see if position is shifting. then coupler will likely need to be tightened. i also add locktite or superglue to make sure it does not move around.

also it is possible that the stepper may be sliding downwards on its shaft, especially if it is just in place by its bearings. you can check this by marking the stepper shaft on the outside.

best calibration is with nozzle heated up and bed heated to temp.


make sure all of machine is mechanically secure, the hot end and the sides are tightened. loose parts can wiggle free and mess things up.

as always go ahead and do a complete inspection of the machine, there are several times that people spend time solving complex issues when in the end it is just a loose nut or screw!

If you have "free" nuts in Z carriages try to load left carriage with a weight and repeat tests. Left carriage is very lightweight and sometimes it even does not press low endstop (in older versions).