OpenDACT(Delta Automatic Calibration Tool) - For Repetier

[RollieRowland]

Hi'
thanks for the update. jusk got myselt the dial indicator. should be set to 0 when the hotend touch the center of the palte with the paper? and then start the calibration process?

thanks

Yes, the center should be zeroed, however, the hotend does not have to be the same level as the dial gauges, so the dial gauges can extend beneath the tip of the hotend.

I should mention this as well. If your dial gauges cannot be zeroed, no negative values can be measured (negative in your case, so the plate is physically lower). A fix for this would be to offset your horizontal radius so ALL values around the outside of the plate are able to be measured. This new horizontal radius will need to be entered into the program. You may also have to change the horizontal radius after each iteration, but this is a work around.

[duvdev]

Thanks. I will try and report.

by the way what kind of Z probe are you thinking to get and do you plane to wire it to the Rambo board?

[RollieRowland]

Thanks. I will try and report.

by the way what kind of Z probe are you thinking to get and do you plane to wire it to the Rambo board?

I am just starting to look into Z-probes, and am seeing how much work it will be to integrate it. Once I find a suitable one, most likely mechanical due to glass beds, then I will link it here.

[ramai]

How do you tag someone, we gotta get 626Pilot in on this thread.

Edit: Here is a link to his diy z-probe
http://www.thingiverse.com/thing:161753" onclick="window.open(this.href);return false;

[RollieRowland]

How do you tag someone, we gotta get 626Pilot in on this thread.

Edit: Here is a link to his diy z-probe
http://www.thingiverse.com/thing:161753" onclick="window.open(this.href);return false;

Funny you linked that, I already had it up! And yes, that would be a big help!

I am trying to figure which method for auto-calibration would be best, in firmware or in a desktop application. I would prefer a desktop application simply because there would then be support for other delta printers. Also, the only information needed from the user would be for the radius of the print bed (for sending gcode that wouldn't break their printer). All of the other data could be pulled from the printers EEPROM and a Z-probe.

[geneb]

...and the desktop solution can do the heavy lifting (mathematically) that many controllers don't have the chops for.

g.

[RollieRowland]

...and the desktop solution can do the heavy lifting (mathematically) that many controllers don't have the chops for.

g.

That is also a good point! If you or Ramai or anyone else, know who can help with this application development, then send them this way! I will be back on later to message a few people to see if they can help or point me in the right direction.

[bot]

While we're all here and talking about this neat form of calibration, how about a two-pronged approach: calibrating x/y positioning errors by putting a laser on the hot end (that we know is point straight down -- no easy task) Combined with the dial gauge, and a piece of graph paper on the bed, we can "zero" the graph paper when the effector is centered, then see how far off the x/y coordinate we are with the laser dot. The laser has to be totaly perpendicular to the effector plane, tho.

[Jimustanguitar]

Just mount a 'real' dial indicator sideways and measure against a printed pillar. If we're accurate to the hundredth of a millimeter in Z, might as well have the same accuracy in XY.

[bot]

Well, the point would be to measure absolute inaccuracies across the build plate. When a tower endstop is not set correctly, the error distorts the x/y plane as a curve, so we could detect this with the laser.

[geneb]

The problem would be the diameter of the dot. Many off the shelf laser pointers have a huge dot, even close in. You'd almost need to build a custom aperature(sp) for it, or use something like this: http://www.lasercenteredgefinder.com/" onclick="window.open(this.href);return false;. (I have one and it's AWESOME.)

g.

[bot]

Yup... I destroyed two $1.50 laser pointers experimenting. The dot would definitely be big. It would probably give you some hint at least, as long as it was true to the axis that it is meant to be pointing in.

How much are those nice center/edge finders? I was looking at gun bore lasers but they are apparently not that parallel to their housing.

[RollieRowland]

Ok, well there may also be a simpler way to measure inaccuracies in the XY plane. I am assuming with this that the inaccuracies will be in part from issues with the tower, so I drew up an stl real quick. From the center to the outer tip of each point, the measurement should equal 124mm. I am not sure how well this will work, if at all. But I figured it'd be enough to ponder on for now.

If the value is smaller, then increase the delta radius for that tower... maybe? I haven't quite delved into the XY domain yet. Let me know if this method won't work.

calibrationTest.stl

(11.8 KiB) Downloaded 389 times

[KAS]

You can also tape a sheet of paper down and run a script to touch each point. Then measure from center dot to each location. A .4mm nozzle leaves a perfect circle to take a measurement with a digital caliper. Might be quicker than reprinting after each change. Ultimately you want equal values for each tower to center.

[RollieRowland]

You can also tape a sheet of paper down and run a script to touch each point. Then measure from center dot to each location. A .4mm nozzle leaves a perfect circle to take a measurement with a digital caliper. Might be quicker than reprinting after each change. Ultimately you want equal values for each tower to center.

20150721_212719.jpg

That is definitely a faster method! Thanks!

[teoman]

You could put a pen on the effector and make it draw lines, rectangles and circles on paper paced on to the bed.

[bot]

This is the curvature that i was speaking about:

[img]http://boim.com/img/towerShiftDistortion.svg[/img]

from http://wp.boim.com/?p=67" onclick="window.open(this.href);return false;

[RollieRowland]

This is the curvature that i was speaking about:

[img]http://boim.com/img/towerShiftDistortion.svg[/img]

from http://wp.boim.com/?p=67" onclick="window.open(this.href);return false;

Well.. Does this distortion only occur when the towers are off? Or does this also occur when they are calibrated properly?

[Jimustanguitar]

Has anybody with Smoothie tried this yet? Does that firmware have the same capacity to adjust angle and radius for each tower?

[KAS]

Has anybody with Smoothie tried this yet? Does that firmware have the same capacity to adjust angle and radius for each tower?

I would think yes, although I haven't messed with the settings myself.


Advanced settings

The following settings can be set in the config, or via M665. They are used to compensate for a frame that is not perfectly aligned, it is recommended the frame be fixed rather than using these to compensate. Generally speaking adjusting any of these to anything other than 0.0 will fix one thing but throw another thing off.

delta_tower1_offset
delta_tower2_offset
delta_tower3_offset
delta_tower1_angle
delta_tower2_angle
delta_tower3_angle



alternatively the values can be changed live (and saved with M500):-

M665 Axxx Bxxx Cxxx Dxxx Exxx Fxxx

where A is delta_tower1_offset, B is delta_tower2_offset, C is delta_tower3_offset
where D is delta_tower1_angle, E is delta_tower2_angle, F is delta_tower3_angle

[enggmaug]

It does not seem to work on my computer (Win7 + Chrome).

Calculation never ends... or not within usable time. How long is it supposed to last ? less than a second ? seconds ? minutes ? hours ?

I am french and tried using data following both french decimal coma, or american decimal point, as I figured it could be a source of problems.

I'll look into that when I have more time.

[bot]

This is the curvature that i was speaking about:

[img]http://boim.com/img/towerShiftDistortion.svg[/img]

from http://wp.boim.com/?p=67" onclick="window.open(this.href);return false;

Well.. Does this distortion only occur when the towers are off? Or does this also occur when they are calibrated properly?

According to that website, that specific example was because the z tower (I believe) was 10mm off. So, that is quite exaggerated.

[techstorage]

Thanks for all the work on this Rollie, and a video to boot!

[Jrjones]

It does not seem to work on my computer (Win7 + Chrome).

Calculation never ends... or not within usable time. How long is it supposed to last ? less than a second ? seconds ? minutes ? hours ?

I am french and tried using data following both french decimal coma, or american decimal point, as I figured it could be a source of problems.

I'll look into that when I have more time.

Make sure you have all the fields filled out, I believe Rollie is still working on doing a check in the program to make sure that all the data fields are filled, but right now if you forget one or more then the program doesn't complete.
If you have everything filled out it takes less than 0.5 second for the new values to be calculated.

[RollieRowland]

It does not seem to work on my computer (Win7 + Chrome).

Calculation never ends... or not within usable time. How long is it supposed to last ? less than a second ? seconds ? minutes ? hours ?

I am french and tried using data following both french decimal coma, or american decimal point, as I figured it could be a source of problems.

I'll look into that when I have more time.

Yeah that is a problem that I still need to fix. It is set to use the american decimal point. Also, it is best to click enter default settings because it enters defaults into the expert settings. Without this, the code may run forever.

Edit: ... I had just noticed that Jrjones had already explained this.. Anyway, thanks Jrjones!