Z offset irregularities across the build surface

Hello,

I have a new Workhorse printer that so far has printed test prints in the center of the bed well. However, when I attempted to put the printer into actual service (which involves printing an array of the parts we use across the bed) The print failed. The Z offset in some areas is so small that the first layer is thin enough to have holes dragged in it and cause nozzle clogs, while the Z offset in other areas is so large that the print cannot even adhere to the bed.

I used a feeler gauge to measure the Z offset in a 3x3 grid and found the Z offset map has sags and distortions that exactly match the areas of the print bed with failed first layers. See the map below.

Also attached are the pictures of test prints that I located approximately at each of those coordinates that shows the quality of the first layer there. Unfortunately my test print coordinate convention is different than the printer coordinate convention, but I have included a name reference in the grid below to look up the test print at that location precisely.

"Z = " measurement in the array is the printer Z setting required to just contact a .152 mm thick feeler gauge at that printer XY coord

Printer Y coord												
												
250		x-y-				x0y-				x+y-		
		Z = 2.075			Z = 2			Z = 1.775		
												
												
												
												
140		x-y0				x0y0				x+y0		
		Z = 2.025			Z = 1.975			Z = 1.825		
												
												
												
												
30		x-y+				x0y+				x+y+		
		Z = 2.175			Z = 2.125			Z = 2		
												
												
Y												
	Printer	X	40				150				260		

Anyone have any advice on how to resolve this problem? Or do I have a “lemon” so to speak? Taz support is giving some resistance to even acknowledging there is something wrong with this printer.

And here are some of the pictures of test prints at those locations:
Top right corner


Center of the plate

Bottom left corner

I own a Mini 2 and I have the same exact problem. From left to right on the bed, the z-offset ranges from -1.20 to -1.37 BUT it can vary by the day. In other words, the machine head cannot retain its offset. As a result, I am constantly having to level the x-axis and fine tune my prints as they are printing. I cannot currently print something that covers the entire bed. I have called Lulzbot tech support many times trying to resolve the issue. Since they haven’t provided me with a solution, I can only assume that there is no solution. This of course means that the Lulzbot machines are inherently defective. You are not the only one with this issue. If you have calculated the exact difference in the offset at each corner, one solution I’ve come up with would be to print small spacers to make up for the difference at each corner so that the plate surface is level. These can be installed under the plate holders. Just make sure that they don’t interfere with the metal connections in the corners. Otherwise the machine will not properly level.

That sort of solution has been done and with great results. See post SOLVED: Taz 6 auto leveling is >0.5mm off (NOW PERFECT TO 0.01mm!)

2 Likes

I have no idea what you guys did to fix this because when I put that gcode into my Mini 2, it goes haywire. It does several wipes, increases the temperature and changes the filament!! It does not provide a read out. Nice concept but it doesn’t help me.

You put it where? Those G commands are to be entered one at a time on the console interface built into Cura. You have to connect then you can run the commands. The examples there are for the bed size of the Taz 6, you will need to adjust the X & Y values of your mini’s bed size coordinates. Example command, G1 X-8 Y-8 F8000 ; move safely to font left. As the comment says, this will move the nozzle to coordinates X=8, Y=8, which will be right over the front left washer. The results of these tests will display values so you’ll know where to add the shims under the washer’s support. Cmidgley has included the shims in the thread. [Leveling corner.stl) (36.4 KB).

Thank you very much for responding. I had created a gcode file to run on the machine. I thought that the results would show in the LCD screen on the machine. Seemed logical at the time! I have now gone into the Cura console and have run the diagnostics. It will not be possible for me to fix this machine. The results change every single time I run the G29 V4 routine. Z is all over the place. Here are the results for the 3 runs. Any suggestions on how to fix this? The shim may help a little, but it won’t be level every time. In addition, once I scrub the nozzle head to remove debris, the numbers will change AGAIN. See Run 4

Run 1

< [07:35:01] Bed X: -3.000 Y: 168.000 Z: -1.633 (.14 from the highest corner)

< [07:35:11] Bed X: -3.000 Y: -4.000 Z: -1.493

< [07:35:21] Bed X: 163.000 Y: -4.000 Z: -2.674 (1.181 from the highest corner)

< [07:35:29] Bed X: 163.000 Y: 168.000 Z: -2.014 (.521 from the highest corner)

Run 2

< [07:52:14] Bed X: -3.000 Y: 168.000 Z: -1.631 (.15 from the highest corner)

< [07:52:24] Bed X: -3.000 Y: -4.000 Z: -1.481

< [07:52:34] Bed X: 163.000 Y: -4.000 Z: -2.641 (1.16 from the highest corner)

< [07:52:44] Bed X: 163.000 Y: 168.000 Z: -1.991 (.51 from the highest corner)

Run 3

< [07:55:54] Bed X: -3.000 Y: 168.000 Z: -1.634 (.146 from the highest corner)

< [07:56:04] Bed X: -3.000 Y: -4.000 Z: -1.488

< [07:56:14] Bed X: 163.000 Y: -4.000 Z: -2.654 (1.166 from the highest corner)

< [07:56:24] Bed X: 163.000 Y: 168.000 Z: -1.995 (.507 from the highest corner)

Run 4

< [08:17:15] Bed X: -3.000 Y: 168.000 Z: -1.683 (.201 from the highest corner)

< [08:17:25] Bed X: -3.000 Y: -4.000 Z: -1.482

< [08:17:35] Bed X: 163.000 Y: -4.000 Z: -2.661 (1.179 from the highest corner)

< [08:17:45] Bed X: 163.000 Y: 168.000 Z: -2.021 (.539 from the highest corner)

The numbers will never be exactly the same from test to test but should be relatively close in value from the previous test, and yours are. The values are a tenth, hundredth, a thousandth of a millimeter; the ultimate goal here is to be within a half a millimeter (.5) of each other. Of course the closer the better. When I leveled mine I think I got it within .1mm. It is a very manual and tedious process but is worth the time and effort to achieve a level as possible bed.

You’ll need to shim the first two corners at the very least and re-test. And repeat the process until you’re within .5mm or better tolerance Also keep in mind how much you tighten the screws on the corners as the bed holders are very flexible and is easy to compress and over torque thus changing the corner’s height influencing the measurements, try and keep it as consistent as possible.

There are too many variables to take into account when leveling the plate–rubber pads that hold the glass plate, screws in those pads that when tightened impact the height, the metal washers that hold the plate down and their screws that impact the height. I realize that this is how it’s made, but it is a poor design. I gave up on trying to level the plate using the gcode sequences because there was no way to get the numbers within .5mm of each other. I made over 30 attempts, with and without using shims, and the numbers were all over the place.

I went to using the ruler that came with the machine. I was able to get the plate leveled by sight and only needed to put a carboard shim under one corner. The machine came with a z:offset of -1.24 but this was been updated to -1.10 due to the leveling. However, this machine has been having the problem of probe failure, even before I worked on all of this leveling. It fails and stops working. I’ve contacted Lulzbot and all they tell me is that my washers are dirty which is not the case. In addition, whatever update Dell or Microsoft did in September screwed up the Cura software which caused filament to just pour out of the nozzle on any new compiled gcode. I’ve had to deal with uninstalling and installing that too and now my z-offset is no longer -1.10mm on the prints–it’s -.68! The cura software should have nothing to do with the z-offset!

This is not rocket science, and I would argue rocket science would be a lot easier because dealing with this machine has been plagued with illogical data! And this is the replacement machine that they sent me just over a year ago! The first one had problems within its first year too! Again, I appreciate your help and I’ll keep struggling with this thing until I can get a better one from a different manufacturer.