Kossel xl custom built. Ir sensor coding fixed
-
That is a pretty good deviation but on the final line in your Bed.g
this G30 P9 X0 Y0 Z-99999 S7
Change to G30 P9 X0 Y0 Z-99999 S8 (this will calibrate for any tilt but will leave your Diagonal rods untouched which you will be able to calibrate later with a test cube)
And not sure if this is strictly required but add M500 after it to save the params especially if you know the precise length of your rods (This may need to be reset in your M665 line to the actual value (in Config.g) and M502 running manually first)
Doug (David will probably be along to comment soon enough)
-
Thank you Doug this is the results I got when i changed the S7 to S8.
M666
Endstop adjustments X-0.54 Y-2.28 Z2.82, tilt X0.62% Y-2.94%
M665
Diagonal 375.057, delta radius 205.677, homed height 215.573, bed radius 130.0, X -0.835°, Y 0.026°, Z 0.000°
G32
Calibrated 8 factors using 10 points, deviation before 0.073 after 0.017 -
Now what makes it a good result? is it the deviation being as low as possible or is it to do with with the m665 results being close as possible to the M666?
-
Getting better time to try a test print eh
-
As excited I as I am to do a test print I have to get ready for job number 2
-
Now what makes it a good result? is it the deviation being as low as possible or is it to do with with the m665 results being close as possible to the M666?
The deviation is a measure of the statistical variance over the probed points compared to what the algorithm predicts them to be. You have a deviation over the bed of just 0.017mm between all of your probed points. Mine's running at 0.02 which I was very pleased with (with Doug's lovely aluminium effector) I had it this low before when I used to have the k800 effector and magnets. The metal carriages and effector (and linear rails obviously) really do reduce play.
No M665 is just some of your delta parameters and M666 are the endstops. Some firmware roll them all into one command (like Rich Cattels marlin), its arbitrary.
Whats important now as Doug said leave it on 8 factors.
Measure your rods as accurately as you can (from centre of rotation at each end) and enter that rod value in M665 L parameter (in config_override.g, not config.g). Do autocalibration. Then print a test cube. If its 10% too big in X and Y lower your rod length by 10%,autocalibrate, and print another (in reality its more likely to be 1%). If the cube is too small, increase rod length and print again. If the cube is wrong in Z (and its not due to squashed first layer, so print one at 20mm tall and one at 40mm tall see if the error remains constant - first layer squash, or doubles - steps/mm are wrong change x,y and z steps/mm). It transpires that unless your belts are really tight a gt2 belt's pitch is a fraction less than 2mm. Hence some people use 80.5 steps/mm (assuming 1/16 stepping and 20t pulleys).
-
But at least we have got the OP somewhere close now!!
-
Thank you guys so much for your help I was pretty much ready to just bin it and give up. Also the information you provided has helped full most of the black hole in my understanding of what to do next.
I'm running gt2 belt and pulleys that are 16 teeth with 1/16th stepping.
-
So 100 steps /mm then, see how your test cube comes out. If it's perfect then enjoy printing some accurate stuff. And turn on interpolation in config.g (M350 X16 Y16 Z16 E16 I1) you'll be pleased that you did.
-
Thank you so much guys I really appreciate all the help. I'll keep this all in mind when I can finally get to do the test prints on Friday.
-
Ok sadly this thing has decided to act up once again. I managed to get some time to sit down and try the cube test. So I figured I'd double check my z height and do another calibration preprint.
With the paper test the height still had 40mm to go before it hit zero. So I 're adjusted the height within the config file and double checked it.
It was paper perfect, so I thought ok time to auto calibrate and it came back with a devation of before 5.002 after 1.236. Then I recalibrated untill I managed to get it down to 0.2xx I can't remember exactly. Then it climbed all the way up to 9.xxx.
I restarted the printer to get back to the 0.2xx that was saved And tested the z height and once agian it wanted to go through the bed.
I spent about an hour going back and fourth with this before I lost my temper in pure frustration I turned everything off and walked away. This is what it was doing before the rebuild but now I have no idea what to do because it feels like it just doesn't want to work.
What do you guys think is the issue?
-
Could it be that you have the acceleration set too high, so that the amount that each carriage moves is not consistent? Or motor current much too low?
Another possibility is that you have at some stage saved incorrect M665 and M666 parameters into config-override.g using the M500 command, and the parameters are being reset to those parameters every time you restart. The config-override.g file behaves like EEPROM does in Marlin. You can open config-override.g in the System Files Editor to check what's in it. It is only used if you have a M501 command in config,g (or M98 Pconfig-override.g).
I suggest you start by getting the printer to the point that you can home it, then run G1 Z50 and it will move to 50mm above the centre of the bed every time - even after a restart. Use the M665 H parameter to adjust the height.
When you have done that, measure the trigger height of the IR sensor as described at https://duet3d.com/wiki/Calibrating_a_delta_printer#Measuring_the_trigger_height and put the trigger height in the Z parameter of the G32 command in config.g. Again, be aware that it can be overridden by an entry in the config.g file.
After that, set the H parameter in your M558 command to 30 (so that probing starts 30mm above where the bed ought to be) and run auto calibration several times.
If you are using a config-override,g file, you can then run M500 to save those values. If you are not, then run M665 and M666 to read off the values found by auto calibration, and manually edit the M665 and M666 commands in config,g to match.
HTH David
-
Thank you David, I did try that last night following your guide and it didn't help but this printer seems to be like night and day, and today is a new day so I'll try again when I get home.
Thank you for the help.
-
If you still have problems when you try again, let me know which step you get stuck at. You mentioned "night and day" - are you sure that the IR sensor is giving a consistent trigger height regardless of the lighting conditions?
-
If you still have problems when you try again, let me know which step you get stuck at. You mentioned "night and day" - are you sure that the IR sensor is giving a consistent trigger height regardless of the lighting conditions?
Yeah I followed your guide on checking for consistent triggers and it aligns to the guide. I finaly have a few hours to work on it so ill start fresh with endstop adjustments set to zero and see if it will play ball today.
Thank you David
I just want to print something haha
-
Ok I tested the trigger height 10 times now and 5 times it came up with Z1.363 and 5 times it came up with Z1.403
I was going to just test it 5 times to make sure it was consistent but on the 5th time it came up with Z1.403.
So I did one more test as piece of mind and it went back to Z1.363. then from the 7th time to the tenth time it produced Z1.403So ill use 1.403 as the trigger height since it was the last set that showed up but is this something that I should be worried about?
-
You have a trigger height variation of 0.04mm, which is higher than ideal. If you see only those 2 values, this suggests that your Z resolution is only 0.04mm, which is unusually high. The steps/mm in your config.g file is 80, which would give 0.0125mm. Perhaps you have binding in one of your carriages, or play in the joints.
However, this should not stop you printing.
As I said before, do make sure that settings in config-override.g are not overriding changes you make. If in doubt, clear out or delete the config.override.g file.
-
I have changed my Step/mm to 100 as DjDemonD suggested.
I have a Feeling its the linear rail on Z now that you mention it when I was installing it it didn"t seem quite as smooth as the others so i washed the bearings down with brake cleaner and re-lubed them with some lithium grease rebuilt the carriage which seemed to help but i guess it wasn't enough. luckily I have some Hiwin Linear rails coming to bad they aren't black but function over form
-
If your steps/mm is 100 (which implies you are using 1.8deg motors and 16-tooth pulleys) then your Z resolution is 0.01mm. So I would expect the trigger height measured by repeated probing at the same point to vary by +/- 0.01mm if your mechanics is working really well.
-
I think it's the z linear rail as after I do a calibration I home the printer then go to check the z height and it's set too high.
