Calibrating a large delta

paboman

Hi,
I am trying to calibrate a large delta printer with a printable radius of 500mm and a height of 1500mm
doing the autocalibration at factor 6 I get : Calibrated 6 factors using 16 points, deviation before 0.016 after 0.013 which i think is good enough but when I run Mesh Grid compensation I get a very deformed mesh

The bed.g file was created with the Escher wizard the I added the H value for each probe point, as you can see the deviation in H is quite large, I think this is due to tilted bed.

; bed.g
; called to perform automatic delta calibration via G32

M561 ; clear any bed transform
; bed.g file for RepRapFirmware, generated by Escher3D calculator
; 16 points, 6 factors, probing radius: 500, probe offset (0, 0)
;The value of each H parameter should be (probe trigger height at that XY position) - (probe trigger height at centre of bed).

G28
G30 P0 X0.00 Y500.00 Z-99999 H3.298
G30 P1 X321.39 Y383.02 Z-99999 H2.815
G30 P2 X492.40 Y86.82 Z-99999 H2.305
G30 P3 X433.01 Y-250.00 Z-99999 H1.957
G30 P4 X171.01 Y-469.85 Z-99999 H0.956
G30 P5 X-171.01 Y-469.85 Z-99999 H0.997
G30 P6 X-433.01 Y-250.00 Z-99999 H2.687
G30 P7 X-492.40 Y86.82 Z-99999 H3.444
G30 P8 X-321.39 Y383.02 Z-99999 H3.585
G30 P9 X0.00 Y250.00 Z-99999 H1.266
G30 P10 X216.51 Y125.00 Z-99999 H0.899
G30 P11 X216.51 Y-125.00 Z-99999 H0.254
G30 P12 X0.00 Y-250.00 Z-99999 H-0.204
G30 P13 X-216.51 Y-125.00 Z-99999 H0.565
G30 P14 X-216.51 Y125.00 Z-99999 H1.423
G30 P15 X0 Y0 Z-99999 S6

The build plate is quite planar as I have checked it with a calibrated 1m steel straight edge tool and a filler gauge
For sure the bed is tilted in some direction because my zprobe triggering height is not the same across the build plate while the repeatability is something like this :

Stopped at height 12.008 mm
Stopped at height 12.005 mm
Stopped at height 12.009 mm
Stopped at height 12.008 mm
Stopped at height 12.007 mm
Stopped at height 12.018 mm
Stopped at height 12.011 mm
Stopped at height 12.028 mm

Bed rest on 6 points which can be regulated, I have tuned them to get it coplanar with the straight edge tool in 3 directions each parallel to X, Y and Z towers.

The probe I am using is a very simple design, when triggered it close the signal to ground as the nozzle touch a metal piece under it of a know height ( 12mm = trigger height )

Why my probed mesh is so deformed ? Am I missing something?

denke

From the looks of your bed mesh I'd say that you need to decrease the "delta radius".
You should also check the steps/mm and diagonal rod length.

The delta radius is the horizontal span of the delta rod from joint center to joint center when the carridge is centered.

dc42

With the probe at the same XY coordinates as the nozzle, the trigger height should be the same all over the bed, so you should not need any H parameters.

paboman

@dc42
I understand, do you agree that my in probing height difference is due to tilted bed ?

dc42

@paboman said in Calibrating a large delta:

@dc42
I understand, do you agree that my in probing height difference is due to tilted bed ?

Not necessarily, it could also be due to the endstop adjustments being out. The trigger height of the probe arrangement in your photo should not be sensitive to effector tilt.

paboman

endstop are fixed on the top, optical type, but yes they can be triggered not at equal height

dc42

@paboman said in Calibrating a large delta:

endstop are fixed on the top, optical type, but yes they can be triggered not at equal height

That doesn't matter, calibration will correct the offsets. The main issue with the height map that you published is that one or more of the M665 rod length, the M665 delta radius and the M92 tower steps/mm is out by a large amount.

paboman

I will be able to measure steps/mm with a dial gauge but getting the rods length is quite problematic as they are theoretically long 1155mm, I measured them with a steel ruler...but I was unable to find a caliper that big

Do you think changing the calibration factor to 7 can get the rod length correct ?

paboman

I have corrected the R Radius manually to get the nozzle at the same height from the bed in the center as on the circumference of the printable area, then I run the Mesh Grid and get better result then my previous:

Why the probed mesh is lower than the reference plane ?

Now I will measure if steps/mm are dialed in precisely

thanks again

dc42

That looks a lot better!

The probed mesh may be lower than the reference plane if your H parameter is not correct in the M665 command, or if you home the printer when mesh bed compensation is already applied (this is fixed in firmware 2.0RC5).

You should run delta calibration and mesh bed compensation at the same bed and hot end temperatures, in case thermal expansion has an effect.

BenDiesel

If like to see bigger pic of the whole thing

paboman

@dc42

I have upgraded to the last RC5
Corrected my 640 step/mm as I was getting a real Z travel shorter by 3%
Done the automatic delta calibration and copied the values in the config.g, then run the Grid Mesh and now I am getting this mesh....something is wrong i will try to calibrate the radius by hand because before the result was more comfortable.
what do you think?

dc42

Are you sure that you have the correct rod length, measured between bearing centres, in the M665 L parameter?

paboman

@dc42 I am pretty sure is right within half a millimeter
I am using magnetic rods articulated on 10mm steel spheres.

From one end of the sphere to the other end of the sphere is 1155mm - 10mm = 1145 (I am subtracting 5mm from each sphere as the centre of rotation is given by the radius of the 10mm sphere which is 5)

I am getting better result, or maybe its just an illusion, with the Radius manually adjusted instead of using the R values proposed by the automatic probing, why ?

dc42

Automatic calibration minimises the sums of the squares of the height errors. So for good results the probe points should cover all areas of the bed. How many probe points did you use, and how were they distributed?

paboman

the probing points are generated by the wizard using radius 500mm as input (for 1000mm printable area)

Danal

@paboman said in Calibrating a large delta:

I am using magnetic rods articulated on 10mm steel spheres.

From one end of the sphere to the other end of the sphere is 1155mm - 10mm = 1145 (I am subtracting 5mm from each sphere as the centre of rotation is given by the radius of the 10mm sphere which is 5)

1145 sounds correct if YOU MEASURED 1155 to the "outer" part of the spheres.

If you obtained a measurement from the manufacturer of your rods, they are NORMALLY labeled "joint center to joint center" and the subtraction would be incorrect.

paboman

@danal they are manufactured by myself and yes 1155 mm is the distance between the outer part of the spheres

Danal

@paboman said in Calibrating a large delta:

@danal they are manufactured by myself and yes 1155 mm is the distance between the outer part of the spheres

Excellent!!

When it is convenient, it would be really cool to see some pictures. There's a few pics of my 600 Dia x 600 Z Delta if you scroll down on this blog:

http://danalspub.com/

paboman

@dc42 To use the saved mesh grid I understand that I need to add G92 S1 in the config.g, is that right ?