Problem with dimension of object

Hello,

I am a french designer and i have design my own Delta 3D printer and i use a duet 0.6.

I have a problem with the dimension of my printed object.

On an edge of 80x80 mm the printed version fact 79.5 on the X axis and 79.2 on the Y axis.
The Z dimension is pretty good with 0.05mm of precision, the quality of print is very good.

I have check my delta Arms is accurate at 0.05mm.
After five auto-calibration i have: M665 R138.1 L312.9 B120 H456.65,
the R theorical is 138mm and for 313mm for my arms.

Why i have this dimension error ?

The design of my 3D printer is available at https://grabcad.com/mylibrarymodels

Thanks in advance.

Quentin

Bonjour,

Pour ta derniere auto calibration tu devrais la faire a factor 7, 10 point , cette calibration va aussi corriger tes positions de "tower"

Pour l'erreur de dimension il faut que tu imprime une pieces 100X100 ( je me suis fait par exemple un axe en forme de + sur cad pour cela, car sa prend moin de plastic qu'un carrer)

Ensuite tu mesure tes dimensions , si exemple X est de 101.10, tu fait 100 / 101.10 = 0.9891, tu fait la meme chose pour Y et tu fait une tower de 100 mm pour le Z et fait la meme equation. Ensuite dans ton config.g tu enter M579 X0.9891 Y(ton resulta) Z(ton resulta).

M579 est la commande pour corriger les dimentions cartesian des axes.

Je recommande de calibrer ton extruder et faire les test de infill et solid infill ( regler ton flow) avant de faire la calibration de la dimension.


Hello,
For your last calibration I recommend a factor 7 / 10 point , this calibration will also fix the tower position.

For your dimension error, you have to print a part 100X100 ( I did a axis in + shape in cad for that, it take less plastic then a square)

Then you have to measure your dimensions, if for example your X is 101.10, you do 100/101.10 = 0.9891, you do the same for Y and you can print a tower of 100mm for the Z and do the same equation. Then in your config.g you enter M579 X0.9891 Y(your result) Z(your result).

M579 is the command to fix the catesian dimension of the axis.

I recommend to do the extruder calibration first and do infill and solid infill test ( set your flow) before doing the calibration of the dimensions.

1/ What is the firmware version ?

2/ Have you done a 4, 6 or 7 points calibration ? I don't see X and Y values in your M665
Avoid 7 point calibration, mesure accurately your arm length and enter them in M665 (center to center)

3/ What kind of Z sensor ? is there any side offset on the sensor ?

4/ Did you checked the offset on all probe points after calibration (with the paper sheet method)

5/ What is your bed material, did you checked the planeity ?

6/ Grabcad is not accessible without registration, so what kind of structure did you use and are the tower angles accurate by construction ?

7/ depending answer to 6/, print a calibration part and measure angles. If you want accurate parts, you may adjust the X and Y angles in M665 to have accurate angles on printed part, even if these angles are different from what the calibration routine tells.

Please don't take offense, but in French, even if this was written for CNC and not printers, you may have a read of this page:
[cncloisirs.com]
I may translate and adapt this page for printing one day.

Edited 1 time(s). Last edit at 04/26/2016 09:08AM by PRZ.

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]

You might be having problems with effector tilt. Check the distance between ball joints at both top and bottom of each pair of arms.

Hi all, thanks for the reply

Même si je n'ai pas un super anglais je vais répondre avec cette langue se sera plus simple.

M665 is X:-0.6 Y:0 Z:-0.2

The autocalibration found a deviation of 0.2deg for X tower and 0.11 for z tower, the 2020 extrusion are guide in a edge milled.
The delta is make with 3 plate of polycarbonate milled with my hobby CNC, the other parts are printed with my sold 3D printer, arms is carbon fiber with balls at end, on effector tilt a molded round form with magnet.

Please test this link for grabcad normaly is possible to view with no inscription : https://grabcad.com/lanoe.quentin-1/public

I will post a pictures of my design.

I have make five 7 points calibration for have a good arms length and after five 6 points calibration.

My sensor is a mechanical temporary mount on my hot end, i have check the x,y offset with a caliper and for the Z make a 0 with my end on the bed mount the sensor and search when the sensor value change.

I have not checked after calibration, i'll test it soon.

My bed is a aluminium heated bed from seecnc with 260mm glass mounted on

I prefere to fix my mechanical error rather than use the M579


Have a good day

Quentin

M579 is not only to fix mechanical problem , for me before the M579 X came at 100.04 and Y 99.97 , that in the ballpark, I guess is fixing plastic shrink and very minor mechanical position that is not possible to get right at 0.001 " accuracy.

Thats another story if your 100mm came at like 95mm but for very minor length different there nothing wrong about using M579, anyway what do you think the 7 or 6 factor calibration do ?

Yes you right its fix mechanical position problem, so in a way you already using software to trick your printer into giving you a flatter bed by moving tower position around and playing with your arm length and radius, there just no way you will be able to get a good result with accurate part without cheating on the software side with a delta build, there so many parts and measurement we just cant get right with a decent accuracy and its more than true when you using plastic parts and a " kossel " kind of build there not a lot of adjustment you can make to fix thing mechanical side. You just do your best and use software trick to fill the rest.

0.2 deg tangent is 0.0035, so for a radius 200mm, that makes a side error of 0.7mm.
As you have machined your plate, I assume your precision is better than that, so I think you may not trust much the angle given by calibration routines, which means you have other faults in your machine (the routine try to adapt the angles to get the lowest calibration errors, whatever the real type of errors, as it cannot correct a parameter which is not in its input).
While a glass bed may be relatively flat, it is always good to check with a ruler.
It is important to know that a non flat bed is what trouble the most the calibration routine.

[edit] Why having in M665 a X and Y different from what is found by the calibration routine ? especially a high value of 0.6, unrealistic for a machined structure ?

So, I think staying with angles 0 in M665 is a good bet, and you shall do a 4 point calibration (so, not letting the routine to modify the tower angles).
This 4 point calibration will gives you larger calibration errors, but it may drive to better dimensional accuracy while printing. If you test that way, could you please report the result ?
Also, have a look (M122) to the errors, to see if there is one point which may be far from the mean position.

Edited 2 time(s). Last edit at 04/26/2016 09:33AM by PRZ.

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]

Are you sure of the values given by M665 ? as in principle Z is always 0, as you only need to make correction on two angles.

Again, what is the firmware version ?

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]

I have corrected the Z sensor height and remake an autocalibration in 7 points and after an 4 points and my configuration is now:

M665 Diagonal 312.00, delta radius 138.02, homed height 457.44, bed radius 120.0, X 0.00°, Y 0.00°, Z 0.00
M666 Endstop adjustments X0.69 Y-0.36 Z-0.33

My firmware is the last 1.11

The machined part is only the 3 plates in polycarbonate, all other part are 3D printed.

I launch an edge 100mm x 100mm now and i check the change.

If you make an autocalibration in 7 parameters, then in 4 parameters without reinitialising the calibration matrix or the machine, the extra parameters calculated by the 7 parameters (say, arm length and tower angles) will stay during the 4 parameters calibration.
This can be checked by running M665 with no parameters after each calibration.

Edited 1 time(s). Last edit at 04/26/2016 05:05PM by PRZ.

---

Pierre

- Safety [reprap.org]
- Embedded help system for Duet and RepRap Firmware [forums.reprap.org]
- Enclosed delta printers Lily [rouzeau.net] and Lily Big [rouzeau.net]
- OpenScad delta printer simulator [github.com]
- 3D printing on my site [www.rouzeau.net]