Unsolved Delta calibration failed

Hi all,
I have this delta printer.
I have replaced the defective Azteeg X5 mini by the DUET3D 2 WiFi.
After calibrated the printer I'm only able to print small parts, since the calibration is not good on the edge of the bed.I made a video to clearly explain the situation. The resolution of the micrometer is 0.01mm. Which means that the difference between the max and min height after calibration is about 0.6mm on the edge of the bed.
https://dai.ly/k3UsOmQYkj40dtsYV26
Here is my beg.g
M561 ; clear any bed transform, otherwise homing may be at the wrong height G31 X0 Y0 ; don't want any probe offset for this G28 ; home the printer ;*** Remove the following line if your Z probe does not need to be deployed ;M98 Pdeployprobe.g ; deploy the mechanical Z probe ; The first time the mechanical probe is used after deployment, it gives slightly different results. ; So do an extra dummy probe here. The value stored gets overwritten later. You can remove this if you use an IR probe. ;G30 P0 X0 Y0 Z99999 ; Probe the bed and do 6 or 7factor auto calibration G30 P0 X86.60 Y50.00 Z99999 ;H0.04 ; Pos 1 X tower G30 P1 X0.00 Y100.00 Z99999 ;H0.054 ; Pos 2 between XY towers G30 P2 X86.60 Y50.00 Z99999 ;H0.055 ; Pos 3 Y tower G30 P3 X86.60 Y50.00 Z99999 ;H.023 ; Pos 4 between YZ towers G30 P4 X0.00 Y100.00 Z99999 ;H0.019 ; Pos 5 Z tower G30 P5 X86.60 Y50.00 Z99999 ;H0.058 ; Pos 6 between ZX towers G30 P6 X43.30 Y25.00 Z99999 ;H0.008 ; Pos 7 X tower G30 P7 X0.00 Y50.00 Z99999 ;H0.027 ; Pos 8 between XY towers G30 P8 X43.30 Y25.00 Z99999 ;H0.018 ; Pos 9 Y tower G30 P9 X43.30 Y25.00 Z99999 ;H0.023 ; Pos 10 between YZ towers G30 P10 X0.00 Y50.00 Z99999 ;H0.044 ; Pos 11 Z tower G30 P11 X43.30 Y25.00 Z99999 ;H0.053 ; Pos 12 between ZX towers G30 P12 X0 Y0 Z99999 S9 ; center and autocalibrate 6 factors ;*** Remove the following line if your Z probe does not need to be retracted ;M98 Pretractprobe.g ; retract the mechanical Z probe G1 X0 Y0 Z150 F15000 ; get the head out of the way of the bed
And my config.g
; General preferences G90 ; Send absolute coordinates... M82 ; and absolute extruder moves ;M83 ; ...but relative extruder moves ;M555 P1 ; Set firmware compatibility to look like RepRapFirmare M555 P2 ; Set firmware compatibility to look like Marlin ;*** The homed height is deliberately set too high in the following  you will adjust it during calibration. ;M665 R105.6 L215 B85 H230 ; Set delta radius, diagonal rod length, printable radius and homed height ;M665 L290.810 R171.507 H227.139 B114.0 X0.045 Y0.138 Z0.000 ;M666 X0.184 Y0.468 Z0.284 A0.00 B0.00 ;M665 L292.355 R172.023 H227.087 B114.0 X0.217 Y0.195 Z0.000 ;M666 X0.113 Y0.816 Z0.703 A0.22 B0.24 M665 L290.542 R171.422 H226.851 B114.0 X0.242 Y0.211 Z0.000 M666 X0.276 Y0.097 Z0.373 A0.25 B0.14 ;CPU temperature correction M912 P0 S1.5 ; Network M550 K250VS ; Set machine name M551 xk250vs ; Set password M552 S1 ; Enable network ;M586 P0 S1 ; Enable HTTP M586 P1 S0 ; Disable FTP M586 P2 S0 ; Disable Telnet ; Drives M569 P0 S1 ; Drive 0 goes forwards M569 P1 S1 ; Drive 1 goes forwards M569 P2 S1 ; Drive 2 goes forwards M569 P3 S0 ; Drive 3 goes backwards M350 X16 Y16 Z16 E16 I1 ; Configure microstepping with interpolation M92 X200 Y200 Z200 E850 ; Set steps per mm M566 X1200 Y1200 Z1200 E1200 ; Set maximum instantaneous speed changes (mm/min) M203 X18000 Y18000 Z18000 E1500 ; Set maximum speeds (mm/min) M201 X1000 Y1000 Z1000 E500 ; Set accelerations (mm/s^2) M906 X1000 Y1000 Z1000 E500 I40 ; Set motor currents (mA) and motor idle factor in per cent M84 S30 ; Set idle timeout ; Axis Limits M208 Z0 S1 ; Set minimum Z ; Endstops M574 X2 Y2 Z2 S1 ; Set active high endstops ; ZProbe ;M558 P4 H5 F120 T6000 ; Set Z probe type to switch and the dive height + speeds M558 P5 X0 Y0 Z0 H.5 I1 T15000 F50 ;G31 P500 X0 Y0 Z.189 ; Set Z probe trigger value, offset and trigger height G31 P500 X0 Y0 Z0.25 ; Heaters M305 P0 T100000 B4138 C0 R4700 ; Set thermistor + ADC parameters for heater 0 M143 H0 S120 ; Set temperature limit for heater 0 to 120C M305 P1 T100000 B4138 C0 R4700 ; Set thermistor + ADC parameters for heater 1 M143 H1 S280 ; Set temperature limit for heater 1 to 280C M307 H0 A152.3 C416.0 D2.0 S1.00 V23.7 B0 M307 H1 A393.8 C209.3 D1.7 S0.80 V24.0 B0 ; Fans M106 P0 S0 F500 H1 ; Set fan 0 value, PWM signal inversion and frequency. Thermostatic control is turned off M106 P1 S1 F500 H1 T40 ; Set fan 1 value, PWM signal inversion and frequency. Thermostatic control is turned on ;M106 P2 S1 I0 F500 H1 T45 ; Set fan 2 value, PWM signal inversion and frequency. Thermostatic control is turned on ; Tools M563 P0 D0 H1 ; Define tool 0 G10 P0 X0 Y0 Z0 ; Set tool 0 axis offsets G10 P0 R0 S0 ; Set initial tool 0 active and standby temperatures to 0C ; Automatic saving after power loss is not enabled ; Custom settings are not configured ; Miscellaneous ;T0 ; Select first tool ; Mesh Grid M557 R120 S30 ;Scale Correction ;M579 X1.020 Y1.012 Z1 M579 X1.013 Y1.01
I guess that defective bed leveling comes from the glass but I did the same test after rotating the glass by 180°, and the issue is still the same.
Any help will be appreciated.
Thank you.

Your dial gauge probe is offset from the nozzle, which means that if you have a geometric error that causes effector tilt (very common in delta printers) then it won't give you an accurate reading of nozzle height. My suggestions:

Measure the FSR trigger height at various points on the bed, not by using the dial probe but but dropping the nozzle until it just grips a piece of paper, sending G92 Z0, raising the nozzle a few mm and using G30 S1 to measure the trigger height. This is to check that the FSRs are giving you consistent trigger heights.

If the FSR trigger height is consistent, after calibration run G29 to generate a height map. Then the difference between the dial gauge reading and the height map is caused by effector tilt. See https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer#Section_Make_sure_your_build_is_accurate for common causes.


Thank you for quick help.
I already do such calibration for FSRs.
As you can see in my bed.g, I have commented all the H parameters to do new tests. because even with H parameters I have bad results.
G30 P0 X86.60 Y50.00 Z99999 ;H0.04 ; Pos 1 X tower G30 P1 X0.00 Y100.00 Z99999 ;H0.054 ; Pos 2 between XY towers G30 P2 X86.60 Y50.00 Z99999 ;H0.055 ; Pos 3 Y tower G30 P3 X86.60 Y50.00 Z99999 ;H.023 ; Pos 4 between YZ towers G30 P4 X0.00 Y100.00 Z99999 ;H0.019 ; Pos 5 Z tower G30 P5 X86.60 Y50.00 Z99999 ;H0.058 ; Pos 6 between ZX towers G30 P6 X43.30 Y25.00 Z99999 ;H0.008 ; Pos 7 X tower G30 P7 X0.00 Y50.00 Z99999 ;H0.027 ; Pos 8 between XY towers G30 P8 X43.30 Y25.00 Z99999 ;H0.018 ; Pos 9 Y tower G30 P9 X43.30 Y25.00 Z99999 ;H0.023 ; Pos 10 between YZ towers G30 P10 X0.00 Y50.00 Z99999 ;H0.044 ; Pos 11 Z tower G30 P11 X43.30 Y25.00 Z99999 ;H0.053 ; Pos 12 between ZX towers G30 P12 X0 Y0 Z99999 S9 ; center and autocalibrate 6 factors
I followed as well all the calibration documentation including the possibility of mesh correction.
Here is the height map result:
As I already said on my first post, rotating le glass by 180° gives quite the same height map. So I guest that this comes from the frame and not the bed...Even with G29 S1 in the config.g, I still not be able to print big part, this is why I decided to replace the duct by a dial gauge probe to try to understand what's wrong.
Just for fun, I think there is a bug in the mesh process. Changing the mesh grid to 9 points gives me this strange result.
Thanks

Looks like you have 6 points around the periphery, 6 points half way, and 1 in the centre. You may get better results with more peripheral points. I use 9 around the periphery, 3 half way, and the centre.
Yes, that height map looks odd.

Ok, I will give a try with this settings.
Thanks 
What is the maximum number of points we can set for probing ?
Does it make sense if I use the maximum ? 
Maximum points in bed.g is 32. Using more points helps if your height map shows substantial errors in places well away from your existing probe points.

Hello !
Using this awesome tool http://escher3d.com/pages/wizards/wizardbed.php , I set 10 points on edge, 3 half way and 1 in the centre.G30 P0 X0.00 Y110.00 Z99999 H0 G30 P1 X70.71 Y84.26 Z99999 H0 G30 P2 X108.33 Y19.10 Z99999 H0 G30 P3 X95.26 Y55.00 Z99999 H0 G30 P4 X37.62 Y103.37 Z99999 H0 G30 P5 X37.62 Y103.37 Z99999 H0 G30 P6 X95.26 Y55.00 Z99999 H0 G30 P7 X108.33 Y19.10 Z99999 H0 G30 P8 X70.71 Y84.26 Z99999 H0 G30 P9 X0.00 Y55.00 Z99999 H0 G30 P10 X47.63 Y27.50 Z99999 H0 G30 P11 X47.63 Y27.50 Z99999 H0 G30 P12 X0 Y0 Z99999 S9
Autocalibration result:
G32 Calibrated 9 factors using 13 points, deviation before 0.075 after 0.067
Height map result
This gives different result, but that still sounds not very good

@viald I notice that you are using S9 on the last line of your G30's in the bed.g file I think you would do better if you changed that to S6 or S8 S7 and S9 will auto adjust the arm lengths for a best fit and we know that isn't always optimum as I am sure DC will explain.
Just my thoughts.
Doug

Thank you Doug.
Here is the result with S6Unfortunately, almost the same.

@viald that's not actually auto calibration tho that is the result of mesh bed compensation!!

You are right, this is the height map calculation after the auto calibration I did
What information do you what about the autocalibration ? 
Finally, I set 19 points, 12 on edge, 6 half way and 1 in the center. I also add the trigger height for each point in real condition (head at 210° and bed at 65°)
G30 P0 X0.00 Y110.00 Z99999 H0.092 G30 P1 X55.00 Y95.26 Z99999 H0.024 G30 P2 X95.26 Y55.00 Z99999 H0.005 G30 P3 X110.00 Y0.00 Z99999 H0.073 G30 P4 X95.26 Y55.00 Z99999 H0.03 G30 P5 X55.00 Y95.26 Z99999 H0.011 G30 P6 X0.00 Y110.00 Z99999 H0.026 G30 P7 X55.00 Y95.26 Z99999 H0.029 G30 P8 X95.26 Y55.00 Z99999 H0.015 G30 P9 X110.00 Y0.00 Z99999 H0.039 G30 P10 X95.26 Y55.00 Z99999 H0.011 G30 P11 X55.00 Y95.26 Z99999 H0.072 G30 P12 X0.00 Y55.00 Z99999 H0.05 G30 P13 X47.63 Y27.50 Z99999 H0.015 G30 P14 X47.63 Y27.50 Z99999 H0.009 G30 P15 X0.00 Y55.00 Z99999 H0.014 G30 P16 X47.63 Y27.50 Z99999 H0.010 G30 P17 X47.63 Y27.50 Z99999 H0.025 G30 P18 X0 Y0 Z99999 S6
What do you think about the trigger heights ? Is it necessary to add it ?
Here the autocalibration log
Calibrated 9 factors using 19 points, deviation before 0.067 after 0.062
The same using S6
Calibrated 6 factors using 19 points, deviation before 0.065 after 0.064
I don't know what this information means, what could be the conclusion ?
As Doug said, what is the difference between S6/7/8/9 in G30 command ?Thank you.

It seems to be good

In fact, the first layer is still no so good, so I still try to improve the precision of the Delta.
I have completely disassembled the delta and check each part.
I have rebuild it, calibrated again and again for in fact having the same poor calibration resultG32 Calibrated 6 factors using 19 points, deviation before 0.068 after 0.067
The default seems to be symmetrical and I have absolutely no idea where it comes from.
I switched the connection of the stepper and end switch on the board, X >Y, Y>Z and Z>X, and did another calibration.
I supposed that the height map will also rotate 90°, actually not, the height map has exactly the same symmetrical default in the same directionWhat could it mean ? This Delta makes me crazy

check out http://boim.com/DeltaUtil/CalDoc/Calibration.html for possible reasons for deviation.

Thanks. I already read this interesting information, but none of the effects exposed are precisely mine. It seems that my issue is a combination of multiple error effects.
I think, I should start with the rod length effect. I should precisely manually measure each of them, and check is they are equal.
But they are about 290mm long and my caliper is not large enough

@Viald, have you read https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer#Section_Make_sure_your_build_is_accurate ? A circular (bullseye) spirit level mounted on the effector will show up many types of geometric error that can cause height variations.

Yes I read it several times. The bullseye spirit level is a very good idea, If the result is not correct, I think It won't be easy to interpret the feedback.
If the bubble is fixed during effector mouvements, the issue clearly comes from the bed, at the opposite it could comes from too many things.I'm currently trying to compare rods length with a micrometer. I'm going step by step, because I'm lost.

As my caliper is not large enough to measure the magnetic arms length, I made this tool to compare the length.
I took one arm length as a reference and compare it to others, here is the result:
1/ Ref (0.00)
2/ +0.02
3/ 0.03
4/ 0.04
5/ 0.00
6/ 0.00I guess that this is good enough, and my issue does not come from the arms length.
Arms are good
To check if the autocalibration is correct, I measure as precise as possible the length in 3 parts
The sphere diameter is 9.41, so the arm length is 213.64 + 42.55 + 42.53  9.41 = 289.31
The autocalibration (S7) always find a longer length around 291.786M665 Diagonal 291.786, delta radius 171.632, homed height 229.294, bed radius 114.0, X 0.180°, Y 0.129°, Z 0.000°
So I decided to fixe the real length in config.g and did multiple S6 autocalibration
That's the best I can have:G32 Calibrated 6 factors using 19 points, deviation before 0.078 after 0.077
M665 L289.300 R170.748 H229.244 B114.0 X0.167 Y0.125 Z0.000 M666 X0.135 Y0.152 Z0.287 A0.30 B0.19
@dc42 I bought circular spirit level, I first checked the bed level which is perfect.
Check the video https://dai.ly/k4fXd07BAdjzOztaR39Bed is good
Then I put the level on the effector, adjusted its level at X0 Y0 and move to each tower, below is the result:
The video link https://dai.ly/k50wYDzWkYziSLtb1Nv
What ever the Z position is, I have exactly the same levelling, and moving along the Z axe, the levelling does neither not change.
But, actually the effector is tilting when it's close to each tower. For sure this is the issue, but where does it come from ?
Thanks for your help.