At my wit's end
I have spent hours and hours trying to get a consistent first layer and am ready to give up. If I remembered how everything came out of the box I would disassemble it all, box it back up and ship it back. Some background:
- Fairly early on one of the bed standoffs stripped out. No, I didn't tighten the bolts past snug but it stripped out anyway. I ordered a bunch of replacement standoffs from Digikey and wound up replacing all of the bed standoffs because the Digikey standoffs were 5mm exactly while the original standoffs were 5.08mm to 5.10mm.
- When replacing the standoffs I used an in/lb torque wrench to torque them all to the same value with the idea that this would hopefully make them as even as possible.
- The bed does not seem terribly level. If I place a straightedge diagonally in one direction then things seem relatively good, as in I can't get a piece of paper under the straightedge at any point across the surface of the bed. Going diagonally in the other direction however I am able to easily slip a sheet of paper under the straightedge for about 3/4 the diagonal length of the bed. Perhaps this is OK but with a sample size of one who knows. Supposedly the nine point calibration blah blah blah helps to correct for this, but...
- After adjusting the PINDA probe (I've tried this multiple times) and running Calibrate XYZ (again, multiple times, and every attempt yields the best possible "perpendicular! Congratulations!" message) I attempt to print the V2calibration.gcode file. Once I get live Z adjusted to the point that the entire print is sticking well, it's obvious that the right side and the left side are not the same height; the right side is too squished at the same time that the left side (while sticking) is not squished enough, and the difference is very obvious. This is true even if I use the bed correction to virtually raise the left side to the max and virtually lower the right side to the max.
- When I print a wider calibration file that uses a larger portion of the bed (again, bed correction settings at max) I find that at the same time that a line on one side of the bed is printing at 0.2mm, a line on the other side of the bed is printing at 0.07mm. I chatted with a support agent last night and he told me that "that's not that bad" and that a variance of up to 0.15mm is OK. How on earth am I supposed to print a consistent 0.2mm layer if a 75% variance is considered OK? The manual claims that +/- 20 microns on the bed correction can make quite a difference. Not for me it can't.
What can be going on here? It it possible that my PINDA probe is bad/inconsistent out of the box? I have been through the instructions more times that I can count looking for anything I might have missed or done improperly and I can't come up with anything. Every self test and calibration the printer has tells me that everything is great but it will not under any circumstances print an even first layer, even across +/- 2" from the center of the bed. Any suggestions would be greatly appreciated, I'm completely out of ideas.
Re: At my wit's end
Try pacing a shim between the standoff and frame on the low points. A piece of typing paper is about 0.1mm thick.
Maybe that can get it close enough for the software to correct?
Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage or loss. If you solve your problem, please post the solution…
Re: At my wit's end
I am having the exact same problem and have had no luck. I have tried everything from adjusting the PINDA height to shimming the bed to modifying the firmware to sample the probe point more and at a lower speed.
If you send a G81 command after the bed leveling you can see the bed profile the software generates from interpolating the PINDA measurements. Mine seem to never be consistent, even running them one after another without even touching the printer with the heat bed turned off (this isn't a heat issue with the PINDA). For example:
Measured points:
0.48900 0.63339 0.73156 0.78350 0.78922 0.74872 0.66200
0.46650 0.59258 0.67793 0.72256 0.72645 0.68962 0.61206
0.43117 0.55141 0.63277 0.67522 0.67878 0.64345 0.56922
0.38300 0.50989 0.59606 0.64150 0.64622 0.61022 0.53350
0.32200 0.46801 0.56780 0.62139 0.62877 0.58993 0.50489
0.24817 0.42577 0.54801 0.61489 0.62641 0.58258 0.48339
0.16150 0.38317 0.53667 0.62200 0.63917 0.58817 0.46900
Measured points:
0.43800 0.53267 0.59050 0.61150 0.59567 0.54300 0.45350
0.39839 0.48647 0.53877 0.55528 0.53601 0.48095 0.39011
0.35622 0.44517 0.49752 0.51328 0.49243 0.43499 0.34094
0.31150 0.40878 0.46678 0.48550 0.46494 0.40511 0.30600
0.26422 0.37728 0.44652 0.47194 0.45354 0.39132 0.28528
0.21439 0.35069 0.43677 0.47261 0.45823 0.39362 0.27878
0.16200 0.32900 0.43750 0.48750 0.47900 0.41200 0.28650
As you can see the center PINDA probe point has varied 0.156mm which would be equivalent to adjusting the live Z the same amount. Therefore any live Z adjustments I make on one print may as well be ignored on the next. I think there must be an issue with my PINDA.
Can you try running a series of G80 and G81 commands and check how much your bed leveling varies.
Jordan
Re: At my wit's end
just printed two parts (dso138 case & back) and connected my printer to the computer to have a look at the meshbed levelling.
here are the results for the first print:
1st the details during G80 at the start of the print and then the values gathered by G81 after the print was finished.
No clamping for first calibration point.
Z jitter data from Z cal. valid.
0: no xy clamping
X: 12.00000
Y: -2.00000
mesh bed leveling: 0.15083
Bed leveling, point: 1, calibration z: 0.42083
1: no xy clamping
X: 114.00000
Y: -2.00000
mesh bed leveling: 0.47583
Bed leveling, point: 2, calibration z: 0.79083
2: no xy clamping
X: 215.00000
Y: -2.00000
mesh bed leveling: 0.53583
Bed leveling, point: 3, calibration z: 0.51083
3: no xy clamping
X: 215.00000
Y: 96.00000
mesh bed leveling: 0.47000
Bed leveling, point: 4, calibration z: 0.38083
4: no xy clamping
X: 114.00000
Y: 96.00000
mesh bed leveling: 0.58250
Bed leveling, point: 5, calibration z: 0.11083
5: no xy clamping
X: 12.00000
Y: 96.00000
mesh bed leveling: 0.29167
Bed leveling, point: 6, calibration z: -0.07917
6: no xy clamping
X: 12.00000
Y: 194.00000
mesh bed leveling: -0.02583
Bed leveling, point: 7, calibration z: 0.29083
7: no xy clamping
X: 114.00000
Y: 194.00000
mesh bed leveling: 0.41250
Bed leveling, point: 8, calibration z: 0.22083
8: no xy clamping
X: 215.00000
Y: 194.00000
mesh bed leveling: 0.01833
Mesh bed leveling while loop finished.
MESH_HOME_Z_SEARCH:
5.00000
clean up finished
babystep applied
Bed correction data valid
PJR A =-15 microns.
PJR D =-33 microns.
PJR E =-60 microns.
PJR H =-15 microns.
Bed leveling correction finished
Upsample finished
Mesh bed leveling activated
Go home finished
<.... snip ....>
>>>G81
SENDING:G81
Num X,Y: 7,7
Z search height: 5
Measured points:
-0.02583 0.21944 0.36556 0.41250 0.36028 0.20889 -0.04167
0.12426 0.32897 0.45419 0.49991 0.46613 0.35286 0.16009
0.22509 0.39874 0.50924 0.55657 0.54076 0.46178 0.31965
0.27667 0.42876 0.53070 0.58250 0.58415 0.53565 0.43700
0.27898 0.41902 0.51859 0.57769 0.59631 0.57446 0.51215
0.23204 0.36953 0.47289 0.54213 0.57724 0.57823 0.54509
0.13583 0.28028 0.39361 0.47583 0.52694 0.54694 0.53583
and here are the results during the second print.
No clamping for first calibration point.
Z jitter data from Z cal. valid.
0: no xy clamping
X: 12.00000
Y: -2.00000
mesh bed leveling: 0.14833
Bed leveling, point: 1, calibration z: 0.41833
1: no xy clamping
X: 114.00000
Y: -2.00000
mesh bed leveling: 0.47000
Bed leveling, point: 2, calibration z: 0.78833
2: no xy clamping
X: 215.00000
Y: -2.00000
mesh bed leveling: 0.64000
Bed leveling, point: 3, calibration z: 0.50833
3: no xy clamping
X: 215.00000
Y: 96.00000
mesh bed leveling: 0.54250
Bed leveling, point: 4, calibration z: 0.37833
4: no xy clamping
X: 114.00000
Y: 96.00000
mesh bed leveling: 0.51833
Bed leveling, point: 5, calibration z: 0.10833
5: no xy clamping
X: 12.00000
Y: 96.00000
mesh bed leveling: 0.22750
Bed leveling, point: 6, calibration z: -0.08167
6: no xy clamping
X: 12.00000
Y: 194.00000
mesh bed leveling: -0.05417
Bed leveling, point: 7, calibration z: 0.28833
7: no xy clamping
X: 114.00000
Y: 194.00000
mesh bed leveling: 0.39083
Bed leveling, point: 8, calibration z: 0.21833
8: no xy clamping
X: 215.00000
Y: 194.00000
mesh bed leveling: 0.15000
Mesh bed leveling while loop finished.
MESH_HOME_Z_SEARCH:
5.00000
clean up finished
babystep applied
Bed correction data valid
PJR A =-15 microns.
PJR D =-33 microns.
PJR E =-60 microns.
PJR H =-15 microns.
Bed leveling correction finished
Upsample finished
Mesh bed leveling activated
Go home finished
<.... snip ....>
>>>G81
SENDING:G81
Num X,Y: 7,7
Z search height: 5
Measured points:
-0.05417 0.17704 0.32537 0.39083 0.37343 0.27315 0.09000
0.07315 0.26313 0.38970 0.45287 0.45263 0.38899 0.26194
0.16204 0.32058 0.43170 0.49537 0.51161 0.48041 0.40178
0.21250 0.34941 0.45135 0.51833 0.55035 0.54741 0.50950
0.22454 0.34960 0.44867 0.52176 0.56886 0.58998 0.58511
0.19815 0.32115 0.42365 0.50565 0.56714 0.60813 0.62861
0.13333 0.26407 0.37630 0.47000 0.54519 0.60185 0.64000
printer was my genuine i3 mk2 with firmware 3.0.12 & hyperfine bedlevelling amendment and verbosity level set to 20 (that's why you can see what the prusa does during the 9-point calibration sequence.... and how i adds afterwards the bed correction levels (at four of eight points)).
take in account, that I've printed two objects that needed to be removed from the pei sheet with some effort (petg and decent first layer surface).
here now two consecutive G80/G81 sequences, right after each other:
Connecting...
start
Printer is now online.
echo:echo: Last Updated: Jul 23 2017 01:57:38 | Author: (none, default config)
Compiled: Jul 23 2017
echo: Free Memory: 2547 PlannerBufferBytes: 1280
echo:Stored settings retrieved
echo:SD card ok
>>>G80
SENDING:G80
echo:Enqueing to the front: "G28 W0"
No clamping for first calibration point.
Z jitter data from Z cal. valid.
0: no xy clamping
X: 12.00000
Y: -2.00000
mesh bed leveling: 0.15000
Bed leveling, point: 1, calibration z: 0.42000
1: no xy clamping
X: 114.00000
Y: -2.00000
mesh bed leveling: 0.38833
Bed leveling, point: 2, calibration z: 0.79000
2: no xy clamping
X: 215.00000
Y: -2.00000
mesh bed leveling: 0.77417
Bed leveling, point: 3, calibration z: 0.51000
3: no xy clamping
X: 215.00000
Y: 96.00000
mesh bed leveling: 0.49583
Bed leveling, point: 4, calibration z: 0.38000
4: no xy clamping
X: 114.00000
Y: 96.00000
mesh bed leveling: 0.35083
Bed leveling, point: 5, calibration z: 0.11000
5: no xy clamping
X: 12.00000
Y: 96.00000
mesh bed leveling: 0.07000
Bed leveling, point: 6, calibration z: -0.08000
6: no xy clamping
X: 12.00000
Y: 194.00000
mesh bed leveling: -0.06417
Bed leveling, point: 7, calibration z: 0.29000
7: no xy clamping
X: 114.00000
Y: 194.00000
mesh bed leveling: 0.31750
Bed leveling, point: 8, calibration z: 0.22000
8: no xy clamping
X: 215.00000
Y: 194.00000
mesh bed leveling: 0.29667
Mesh bed leveling while loop finished.
MESH_HOME_Z_SEARCH:
5.00000
clean up finished
babystep applied
Bed correction data valid
PJR A =-15 microns.
PJR D =-33 microns.
PJR E =-60 microns.
PJR H =-15 microns.
Bed leveling correction finished
Upsample finished
Mesh bed leveling activated
Go home finished
>>>G81
SENDING:G81
Num X,Y: 7,7
Z search height: 5
Measured points:
-0.06417 0.11444 0.24167 0.31750 0.34194 0.31500 0.23667
-0.02009 0.13752 0.25360 0.32815 0.36116 0.35264 0.30259
0.01963 0.15738 0.26393 0.33926 0.38338 0.39629 0.37798
0.05500 0.17404 0.27265 0.35083 0.40859 0.44593 0.46283
0.08602 0.18748 0.27976 0.36287 0.43680 0.50156 0.55715
0.11269 0.19771 0.28527 0.37537 0.46801 0.56320 0.66093
0.13500 0.20472 0.28917 0.38833 0.50222 0.63083 0.77417
>>>G80
SENDING:G80
No clamping for first calibration point.
Z jitter data from Z cal. valid.
0: no xy clamping
X: 12.00000
Y: -2.00000
mesh bed leveling: 0.13917
Bed leveling, point: 1, calibration z: 0.40917
1: no xy clamping
X: 114.00000
Y: -2.00000
mesh bed leveling: 0.38000
Bed leveling, point: 2, calibration z: 0.77917
2: no xy clamping
X: 215.00000
Y: -2.00000
mesh bed leveling: 0.76833
Bed leveling, point: 3, calibration z: 0.49917
3: no xy clamping
X: 215.00000
Y: 96.00000
mesh bed leveling: 0.49000
Bed leveling, point: 4, calibration z: 0.36917
4: no xy clamping
X: 114.00000
Y: 96.00000
mesh bed leveling: 0.35250
Bed leveling, point: 5, calibration z: 0.09917
5: no xy clamping
X: 12.00000
Y: 96.00000
mesh bed leveling: 0.07083
Bed leveling, point: 6, calibration z: -0.09083
6: no xy clamping
X: 12.00000
Y: 194.00000
mesh bed leveling: -0.07167
Bed leveling, point: 7, calibration z: 0.27917
7: no xy clamping
X: 114.00000
Y: 194.00000
mesh bed leveling: 0.31417
Bed leveling, point: 8, calibration z: 0.20917
8: no xy clamping
X: 215.00000
Y: 194.00000
mesh bed leveling: 0.29083
Mesh bed leveling while loop finished.
MESH_HOME_Z_SEARCH:
5.00000
clean up finished
babystep applied
Bed correction data valid
PJR A =-15 microns.
PJR D =-33 microns.
PJR E =-60 microns.
PJR H =-15 microns.
Bed leveling correction finished
Upsample finished
Mesh bed leveling activated
Go home finished
>>>G81
SENDING:G81
Num X,Y: 7,7
Z search height: 5
Measured points:
-0.07167 0.10907 0.23769 0.31417 0.33852 0.31074 0.23083
-0.02259 0.13678 0.25369 0.32815 0.36014 0.34968 0.29676
0.01991 0.15911 0.26612 0.34093 0.38353 0.39394 0.37215
0.05583 0.17607 0.27496 0.35250 0.40869 0.44352 0.45700
0.08519 0.18766 0.28022 0.36287 0.43560 0.49842 0.55131
0.10796 0.19388 0.28190 0.37204 0.46428 0.55863 0.65509
0.12417 0.19472 0.28000 0.38000 0.49472 0.62417 0.76833
at all I'm not very conviced if comparing the values really helps you to improve your setup. take in account that during the meshbed levelling procedure (G80) you are "sampling" the analog dimensions of a real object (the heatbed) with limitations in room and temporal resolution. so it's obvious that you won't get the same results twice.
dem inscheniör is' nix zu schwör...
Re: At my wit's end
I'm still new to the Prusa, but here's my thoughts: If you are getting variable Z values during calibration, either your bed is loose and moving, or your Z axis is shifting around.
1) Make sure all your bed anchor bolts are tight and make sure the Y carriage is tight to the frame with no wobble.
2) Make sure there is no backlash in either of your Z axis motors and check your bearings, make sure you get nice smooth motion on both Z slides and you have no backlash when moving up and down.
Re: At my wit's end
at all I'm not very conviced if comparing the values really helps you to improve your setup. take in account that during the meshbed levelling procedure (G80) you are "sampling" the analog dimensions of a real object (the heatbed) with limitations in room and temporal resolution. so it's obvious that you won't get the same results twice.
I agree that they would obviously not be exactly the same but the large variations that I am getting cannot be blamed on room temperature fluctuations and the bed changing. Before my print even starts and I see the G81 output I can already tell if my print layer is going to be under or over squished.
2) Make sure there is no backlash in either of your Z axis motors and check your bearings, make sure you get nice smooth motion on both Z slides and you have no backlash when moving up and down.
How can I check for backlash?
Jordan
Re: At my wit's end
How can I check for backlash?
I mis-spoke using the phrase backlash in the motor, I meant to check for backlash in the Z axis gear train. The simplest way would be to spin each motor by hand back and forth in small increment and make sure that the X-Axis carriage moves and doesn't "stick" either up or down before moving in the required direction. Any "sticking" or hesitation in the z-axis motion could translate into variable z height readings when sampling the probe.
Having just put my Prusa together I don't see how there could be that much slop in the Z, but it's the kind of thing we look for in the gantry systems my company makes (although our machines are about 100 times bigger than the Prusa).
Re: At my wit's end
Yea that makes sense, I did try making sure all my axis freely move and they all seem to be fine except for the X axis being a bit stiff. I believe it has to do with the belt/pulley rubbing which seems to be a common issue. Also noticed that tightening the X tension screws helped a bit with consistency.
I think I may have finally gotten my Z height to be consistent by turning on the nozzle fan fully while heating and doing the mesh bed calibration. I find that this makes a huge difference as the fan helps keep the PINDA at room temperature. Also I turn off the heatbed just before the mesh bed leveling just in case the magnetic fields from the bed heating current interfere with the PINDA. Next I want to try reversing the hot head fan direction as I noticed it also blows warm air right onto the PINDA.
Here is my modified G code in case anyone else is having similar issues. BTW I always keep the extruder fully up on the Z axis before starting prints.
M115 U3.0.10 ; tell printer latest fw version
M106 ;Fan On
M83 ; extruder relative mode
M104 S[first_layer_temperature] ; set extruder temp
M140 S[first_layer_bed_temperature] ; set bed temp
M190 S[first_layer_bed_temperature] ; wait for bed temp
M109 S[first_layer_temperature] ; wait for extruder temp
M92 E167; set esteps
G28 W ; home all without mesh bed level
M140 S0 ; turn off heatbed
G80 V0; mesh bed leveling
M140 S[first_layer_bed_temperature] ; set bed temp
G81
M107 ;Fan Off
M190 S[first_layer_bed_temperature] ; wait for bed temp
Thanks for all the help guys, I feel like I'm finally starting to get things working 🙂
Jordan
Re: At my wit's end
...Next I want to try reversing the hot head fan direction as I noticed it also blows warm air right onto the PINDA.
...
just to mention it here again: at my self designed clone, I've reversed the fan direction of the hotend fan. so the fan "sucks" the air out of the extruder body.
for cooling the coldend the efficiency it's absolutely the same as vice versa, but it definitely keeps the p.i.n.d.a. colder (because I don't blow warm air onto the probe).
dem inscheniör is' nix zu schwör...
Re: At my wit's end
just to mention it here again: at my self designed clone, I've reversed the fan direction of the hotend fan. so the fan "sucks" the air out of the extruder body.
for cooling the coldend the efficiency it's absolutely the same as vice versa, but it definitely keeps the p.i.n.d.a. colder (because I don't blow warm air onto the probe).
Yea exactly what I was thinking. So you have tried this? Any noticeable improvements?
Re: At my wit's end
...
Yea exactly what I was thinking. So you have tried this? Any noticeable improvements?
sorry, but reporting an improvement wouldn't be fact based, because I can't compare "before" and "after".
as I build my clone I just clamped a temperature sensor between the fins of the cooler and checked the temperature at the coldend while heating up the hotend, to make sure that the fan direction doesn't influence the cooling capabilities of the system.
as both directions delivered the same result, my decision was in favour of the reverse direction.... having in mind that this might avoid heating up the p.i.n.d.a. .
so my printer was operational @ reverse direction right at the start of the first print.
dem inscheniör is' nix zu schwör...