Page MenuHomeAleph Objects Inc

Nozzle z offset doesnt seems to calibrate correct
Open, Needs TriagePublic

Description

Ive ran calibration on my printer and the z offset for the nozzles was .29. I recorded my values before running calibration and my offset from .27 was -00.31. I tried a print at the new calibration and it was far too squished. I changed to to 0-.31 and it was perfect. Also on other printers that I have worked on, the x and y are almost always perfect but i have to manually tune z

Event Timeline

west created this task.Jan 14 2019, 12:58 PM
logan added a subscriber: logan.Jan 15 2019, 2:11 PM

I've found this to be true as well. The default values retrieved by the auto calibration would vary from 0.20 to 0.38 and the print results would show the second nozzle attempting to print much lower than the first. Reducing this offset value on Q11 to 0.05 to 0.08 yielded the most accurate print results. Last tested on .67 FW. I noticed your numbers are negative while mine were positive, what FW version are you currently running?
Regarding X/Y offsets, these are typically much better but not perfect. Usually not off by more than 1-2 steps on a vernier scale print (0.1 to 0.2mm) and for me able to provide consistent results once adjusted accordingly.
@west Have you found an X Axis that is out of level during auto calibration to affect the z nozzle offset value reported?

Have you found an X Axis that is out of level during auto calibration to affect the z nozzle offset value reported?

I suspect this is why calibration on Z will always be imprecise. With the lateral movement of the toolhead, plus the up and down movement of the nozzle, I doubt we will ever get as good values in Z out of auto-calibration as we do from X and Y.

matth added a subscriber: matth.Jan 15 2019, 2:57 PM

I ran the calibration test 20 times in a row and recorded my results.

The x and y offsets were very consistent, with both the x and y offsets having a range of about 0.03. The z offset had a range of about 0.17, with the majority of results being in a 0.05 range. However, I had 4 outliers that were over 0.5-1.0 above the majority of other results, in addition to a few minor ones. I ran each test back to back, without manually cleaning the nozzles or adjusting temps, so my only guess would be fuzzies from the wiper pad, which I know has been mentioned in other tickets. Firmware v2.0.0.72

anolen added a subscriber: anolen.Jan 15 2019, 3:54 PM

I am on .72, these calibrations and start code were observed for the entire process.
For one of my machines T0's bead was good but I had to adjust T1 by +0.20mm.
For the other I had to adjust T0 by +0.40mm and T1 by +0.20mm.

I also notice that my calibration wipe (aka rewipe) for 2 of my machines seems to be falling to the right a bit, anyone else seeing this?

for my 3rd machine:
T0 was adjusted +0.40 and T1 was adjusted +0.20

To clarify, when I adjust T0 I leave "adjust nozzle together" ON, but when adjusting T1 I switch it OFF.

west added a comment.Jan 16 2019, 7:30 AM

When I calibrated at an earlier fw the results were consistent within a +-0.03 i believe. Now the range is much larger

tutley added a subscriber: tutley.Jan 16 2019, 10:41 AM

@anolen i also saw issues with the rewipe. Mainly that it is too deep.

I calibrated with .72 and got a pretty spot on calibration.

matth added a comment.EditedMon, Jan 21, 3:29 PM

For my machine, firmware 77 made it a lot worse. I've run the calibration more than 10 times in a row and have been getting very inconsistent results for the z offset. My min and max values so far have been -0.29 and 1.32, and even if I dismiss the extreme outliers I still have a very wide range of results, which then would require more testing to find the ideal z offset.

Link to a spreadsheet I've been recording my data on: https://code.forksand.com/matt/q_test/src/branch/master/ods

@matth: Has your printer been updated with two Z endstop switches? .77 gets rid of the old X-leveling routine so it probably will not work as well on unmodified printers.

west added a comment.Tue, Jan 22, 8:26 AM

@marcio mine is still exhibiting the problem. My printer has the parallel z end stops

DaniAO added a subscriber: DaniAO.EditedTue, Jan 22, 8:28 AM

@marcio there are only a few printers in R&D and MER with this update, and we haven't pulled the other printers back to update just yet. That will be done once all the things are approved.

marcio added a comment.EditedTue, Jan 22, 8:51 AM

@west: @matth: We need to determine exactly what firmware version made calibration go worse. To help narrow this down, I made a list of FW transitions that changed the calibration routine. Please install and compare the following versions. If it works with the FW on the left hand of the arrow, but not with the one on the right side of the arrow, then it means that commit introduced the problem.

.25 -> .26
.36 -> .37
.46 -> .49
.63 -> .64
.65 -> .66

karrad added a subscriber: karrad.Tue, Jan 22, 8:57 AM

Just a note, the FW changed names from TAZ7 -> TAZPro. If you have issues using the script, you will want to change that name locally.

I updated the FW install script to cover all versions, to make the job easier.

@karrad: The script should now take care of all cases, including TAZ7 and TAZPro, and Quiver_TAZ7

matth added a comment.Tue, Jan 22, 9:32 AM

@marcio It was 72 -> 77, but it's probably the lack of z-endstop switches.

@matth: Once you get your endstop switches, please confirm whether the problem has been resolved in .79

matth added a comment.Thu, Jan 31, 2:43 PM

I had my machine upgraded with the endstop switches. With .79 I ran several calibrations and got very consistent results.

However, with .80 I'm once again getting a wide range of results, not just with the z offsets but with the x and y as well. I don't know if it's the firmware or the fact that I've been transporting the machine back and forth between AO Mountain and Jeff's house the past few days and maybe I knocked something out of whack?

However, with .80 I'm once again getting a wide range of results, not just with the z offsets but with the x and y as well. I don't know if it's the firmware or the fact that I've been transporting the machine back and forth between AO Mountain and Jeff's house the past few days and maybe I knocked something out of whack?

@matth: The only way to know would be to return to .79 and see if you get better results again.

marcio added a comment.EditedThu, Jan 31, 3:12 PM

@matth: There also is another very useful test you can run. Open a console and run the following commands:

M425 S0 F1 ; Enable full backlash compensation
G425 ; Run auto-calibration on the cube

T0 ; Switch to toolhead 1
G425 V ; Reprobe cube and print measurements

T1 ; Switch to toolhead 2
G425 V ; Reprobe cube and print measurements

What you want to look for is the "Tn Positional Error". After calibration, this should be less than 0.1 mm on each axis. This is the results on my printer when running .81. Notice that my positional errors were reduced to virtually nothing after the auto-calibration:

Cube Sides:␊
  Top:   2.98␊
  Left:  258.07␊
  Right: 269.96␊
  Front: -28.00␊
  Back:  -16.01␊
␊
Cube Center:␊
  X: 264.01␊
  Y: -22.00␊
  Z: -2.02␊
␊
Backlash:␊
  Left:  0.02␊
  Right: 0.02␊
  Front: 0.02␊
  Back:  0.02␊
  Top:   0.01␊
␊
Nozzle Tip Outer Dimensions:␊
  X: 1.89␊
  Y: 1.99␊
␊
T0 Positional Error:␊
  X: -0.01␊
  Y: 0.00␊
  Z: 0.02␊
␊
T1 Hotend Offset:␊
  X: 42.93␊
  Y: 0.29␊
  Z: -0.12␊
␊
ok P15 B4␊
echo:Active Extruder: 1␊
ok P15 B4␊
␊
Cube Sides:␊
  Top:   3.01␊
  Left:  258.00␊
  Right: 270.05␊
  Front: -28.01␊
  Back:  -16.01␊
␊
Cube Center:␊
  X: 264.02␊
  Y: -22.01␊
  Z: -1.99␊
␊
Backlash:␊
  Left:  0.02␊
  Right: 0.02␊
  Front: 0.02␊
  Back:  0.01␊
  Top:   0.01␊
␊
Nozzle Tip Outer Dimensions:␊
  X: 2.05␊
  Y: 2.00␊
␊
T1 Positional Error:␊
  X: -0.02␊
  Y: 0.01␊
  Z: -0.01␊
␊
T1 Hotend Offset:␊
  X: 42.93␊
  Y: 0.29␊
  Z: -0.12␊
␊
ok P15 B4␊
matth added a comment.Thu, Jan 31, 3:46 PM

I updated to .81 and ran those commands. My output:

[15:40:18] T0

< [15:40:20] echo:Active Extruder: 0

[15:40:29] G425 V

< [15:41:29]
< [15:41:29] Cube Sides:
< [15:41:29] Top: 3.12
< [15:41:29] Left: 257.80
< [15:41:29] Right: 270.18
< [15:41:29] Front: -27.96
< [15:41:29] Back: -15.85
< [15:41:29]
< [15:41:29] Cube Center:
< [15:41:29] X: 263.99
< [15:41:29] Y: -21.90
< [15:41:29] Z: -1.88
< [15:41:29]
< [15:41:29] Backlash:
< [15:41:29] Left: 0.02
< [15:41:29] Right: 0.02
< [15:41:29] Front: 0.02
< [15:41:29] Back: 0.01
< [15:41:29] Top: 0.21
< [15:41:29]
< [15:41:29] Nozzle Tip Outer Dimensions:
< [15:41:29] X: 2.38
< [15:41:29] Y: 2.11
< [15:41:29]
< [15:41:29] T0 Positional Error:
< [15:41:29] X: 0.01
< [15:41:29] Y: -0.10
< [15:41:29] Z: -0.12
< [15:41:29]
< [15:41:29] T1 Hotend Offset:
< [15:41:29] X: 43.86
< [15:41:29] Y: 0.69
< [15:41:29] Z: 0.10
< [15:41:29]

[15:42:40] T1

< [15:42:44] echo:Active Extruder: 1

[15:42:45] G425 V

< [15:43:29]
< [15:43:29] Cube Sides:
< [15:43:29] Top: 2.55
< [15:43:29] Left: 257.83
< [15:43:29] Right: 270.81
< [15:43:29] Front: -28.39
< [15:43:29] Back: -15.75
< [15:43:29]
< [15:43:29] Cube Center:
< [15:43:29] X: 264.32
< [15:43:29] Y: -22.07
< [15:43:29] Z: -2.45
< [15:43:29]
< [15:43:29] Backlash:
< [15:43:29] Left: 0.02
< [15:43:29] Right: 0.02
< [15:43:29] Front: 0.13
< [15:43:29] Back: 0.01
< [15:43:29] Top: 0.42
< [15:43:29]
< [15:43:29] Nozzle Tip Outer Dimensions:
< [15:43:29] X: 2.98
< [15:43:29] Y: 2.64
< [15:43:29]
< [15:43:29] T1 Positional Error:
< [15:43:29] X: -0.32
< [15:43:29] Y: 0.07
< [15:43:29] Z: 0.45
< [15:43:29]
< [15:43:29] T1 Hotend Offset:
< [15:43:29] X: 43.86
< [15:43:29] Y: 0.69
< [15:43:29] Z: 0.10

My positional errors are way off.

I think we need to determine what average values are like to say whether yours is particularly bad. My printer seems to *extremely* well tuned.

I notice your backlash on Z is horrendous. That's a bad sign.

matth added a comment.Thu, Jan 31, 3:51 PM

I can bring it back to mountain on Friday for the day if needed

logan added a comment.Thu, Jan 31, 3:52 PM

@marcio
Also the z backlash is double on the 2nd nozzle:

Top: 0.21
Top: 0.42

I've heard offhandedly that Z pulleys were loose on some of the printers. You may want to check that.

Also the z backlash is double on the 2nd nozzle:

@logan: Yeah. Could you run this test on a couple of your printers in MER and see what kind of values we should be expecting? Like I said, my printer seems to be an outlier. It has always performed very well.

marcio added a comment.Fri, Feb 1, 8:01 AM

@matth: I had a thought as to why you are getting poor results. It could be your nozzle was dirty. Instead of running G425 ; Run auto-calibration on the cube in the sequence above, try select "Measure automatically" from the UI under "Nozzle Offset" or "Axis Backlash". The difference between that button and G425 is that the former will heat up the nozzle and do a wipe, which may help with the calibration.

matth added a comment.Fri, Feb 1, 2:52 PM

Well I removed the toolhead and put in a new one, and I've gotten far more consistent results. I'll be doing some more testing but so far my issues seem to be related to that particular toolhead.

matth added a comment.Mon, Feb 4, 12:50 PM

I should also note that the first calibration after a factory reset is always bad. I do a factory reset after each firmware update, and every time, the first calibration is always way off compared to the calibrations I do afterward.

I'm not sure if this requires an adjustment to the calibration process or just an extra step that will be done when calibrating machines after building.

Steven added a subscriber: Steven.Tue, Feb 5, 7:12 AM

@matth can you please bring that tool head, that you're seeing issues with, to R&D the department the next time you come to the AM building?

matth added a comment.Tue, Feb 5, 9:27 AM

@Steven Kent has it, I'm currently using the newer one that I installed last Friday.

@matth ok, we have it in R&D now. @tutley is testing it.

tutley added a subscriber: kent.Tue, Feb 5, 12:54 PM

@Steven @matth I received the tool head (it was disassembled when I got it) and reassembled everything. I found no mechanical issues, no binding, etc.
I installed the tool head on my machine running Marlin .81 and ran through a few calibrations making note of the offsets.

I got the following data on the x4 calibrations that I ran,

Nozzle Offsets:                             Lash:
X: 43.98                                    X: 0.25
Y: 0.23                                     Y: 0.14
Z: -0.67                                    Z: 0.01
 
Nozzle Offsets:                             Lash:
X: 43.99                                    X: 0.25
Y: 0.22                                     Y: 0.14
Z: -0.68                                    Z: 0.02
 
Nozzle Offsets:                             Lash:
X: 43.98                                    X: 0.25
Y: 0.23                                     Y: 0.14
Z: -0.66                                    Z: 0.01

Nozzle Offsets:                             Lash:
X: 43.98                                    X: 0.25
Y: 0.23                                     Y: 0.15
Z: -0.69                                    Z: 0.01

These are very consistent and repeatable, so i do not think there are any issues with the toolhead in its current state.
I gave this tool head back to @kent for further testing on the machine @matth was seeing issues with