OpenDACT(Delta Automatic Calibration Tool) - For Repetier

[Jrjones]

I figured I would upload a picture of the results.

0821150032.jpg

-Default mattercontrol settings (except nozzle size, which is 0.4mm - I have an e3d installed at the moment)
-Pressed print right after calibration, no z-height or any other modification. (Same calibration as the log from the previous post.)
-No bed heat or any form of adhesion - the true test.

Also, sorry I forgot to oversize it to fit the 11 inches, I just used the default size. There is an area in the center where the plastic didn't stick to the bed, I am assuming it is mostly due to the cumulative dust/dirt of 2 weeks - very bad, else just because of no form of adhesion.

Edit:
Thickness measurements of the print
XTower: inner 0.3 outer 0.31
YTower: inner 0.29 outer 0.3
ZTower: inner 0.3 outer 0.32

One more thing, I may need data from different printers for setting correct XY scaling. It is a very tedious task, and requires a lot of patience. However, it will help improve the programs accuracy in the XY dimensions. If you want to help, then let me know. You will need some things: calipers(accurate to 1/100th of a mm), FSRs, several sheets of paper, graphite, and maybe an hour of your time. If anyone is interested, I will write a reply describing how to properly get the data that I need.

I'm going to try the feeler gauge as a switch tomorrow, if it works than I would gladly help you out. Unless you need FSRs and think that the feeler gauge won't work.

Edit: I can't find where you mention it, but all you need to change in the firmware for the auto calibration is enabling the Z-probe correct?

[RollieRowland]

I figured I would upload a picture of the results.

0821150032.jpg

-Default mattercontrol settings (except nozzle size, which is 0.4mm - I have an e3d installed at the moment)
-Pressed print right after calibration, no z-height or any other modification. (Same calibration as the log from the previous post.)
-No bed heat or any form of adhesion - the true test.

Also, sorry I forgot to oversize it to fit the 11 inches, I just used the default size. There is an area in the center where the plastic didn't stick to the bed, I am assuming it is mostly due to the cumulative dust/dirt of 2 weeks - very bad, else just because of no form of adhesion.

Edit:
Thickness measurements of the print
XTower: inner 0.3 outer 0.31
YTower: inner 0.29 outer 0.3
ZTower: inner 0.3 outer 0.32

One more thing, I may need data from different printers for setting correct XY scaling. It is a very tedious task, and requires a lot of patience. However, it will help improve the programs accuracy in the XY dimensions. If you want to help, then let me know. You will need some things: calipers(accurate to 1/100th of a mm), FSRs, several sheets of paper, graphite, and maybe an hour of your time. If anyone is interested, I will write a reply describing how to properly get the data that I need.

I'm going to try the feeler gauge as a switch tomorrow, if it works than I would gladly help you out. Unless you need FSRs and think that the feeler gauge won't work.

Edit: I can't find where you mention it, but all you need to change in the firmware for the auto calibration is enabling the Z-probe correct?

Yes, you need to enable the Z-Probe and set the pin.
Here is my settings(using Rambo with the Z-Minimum pin as the probe):
#define FEATURE_Z_PROBE true
#define Z_PROBE_PIN 10
#define Z_PROBE_PULLUP true
#define Z_PROBE_ON_HIGH true
#define Z_PROBE_X_OFFSET 0
#define Z_PROBE_Y_OFFSET 0
#define Z_PROBE_WAIT_BEFORE_TEST false
#define Z_PROBE_SPEED 30
#define Z_PROBE_XY_SPEED 150
#define Z_PROBE_HEIGHT 39.91
#define Z_PROBE_START_SCRIPT ""
#define Z_PROBE_FINISHED_SCRIPT ""

Well I know this is most likely possible with methods other than FSRs, it is just that the FSRs are going to be the easiest. What I need is the measurements in the XY dimensions, how I measured was by putting paper on the plate, then scraping graphite onto my nozzle while probing in the center and near the tower. I don't absolutely need the heights, it would help but it's not neccessary. Without height measurements, the probe could be dropped just by the z-height on the printers display or in the host software(just so you don't break your bed by moving below the plate). I need about 49 data points from the printer for consistency. Then when I have at least 3 or 4 measurements from different printers, I can average them. This will give the best results for everyone.

Repeat for each tower:
1. Measure distance from the center of the plate to the tower being tested with horizontal radius, diagonal rod, and alpha rotation values at their defaults - XYZ offset does not need changed as it doesn't affect XY scaling.
2. Change delta radius X(dependent on the tower tested, Y, Z for the other towers) to 1, then measure the XY distance. Repeat for delta radius being set to 2,3,-1,-2,-3. Reset back to 0 after test.
3. Set Alpha Rotation A(specific to tower B,C for the others) to 211(specific to tower), then measure the XY distance. Repeat for 212, 213, 209, 208,207. Reset to 210 after test.
4. Set Horizontal Radius to 131(1 plus your default) and measure at the X, Y, and Z towers - all can be done at the same time. Repeat for 132, 133, 129, 128, 127. Reset back to 130.
5. Set Diagonal Rod to 270 (1 plus your default) and measure the X, Y, and Z towers - all at the same time. Repeat for 271, 272, 268, 267, 266.

The 4 points that are needed to measure(from tower to center for each test) for the Rostock Max V2:
Center - G1 Z0 X0 Y0
X Tower - G1 Z0 X-112.58 Y-65
Y Tower - G1 Z0 X112.58 Y-65
Z Tower - G1 Z0 Y130

For other printers replace the gcode with the appropriate sizes for your printer:
//Large = Your build diameter * 0.482;
//Medium = Your build diameter * 0.417;
//Small = Your build diameter * 0.241;
Center - G1 Z0 X0 Y0
X Tower - G1 Z0 X-Medium Y-Small
Y Tower - G1 Z0 XMedium Y-Small
Z Tower - G1 Z0 YLarge

With these measurements - when your EEProm has the defaults horizontal radius, diagonal rod, and alpha rotation - your first measurement should be about 130.

If this wasn't clear enough, then let me know, and I will help.

[stonewater]

Bump

[RollieRowland]

Progress Update:
I have mostly finished the new method which involves iterating step-by-step through each section of the calibration. I need to test, and will not be able to do any testing til next Friday.

For now, I am beginning to implement the correction for XY scaling, using the small amount of data that I have already collected from my printer - barely enough. However, I should be able to get a fair amount of it completed.

Also, I have corrected the UI so it has feedback from the advanced and normal calibration in the more panel. Still need to correct the log.

I may update again later this week on the progress, if much is made. Otherwise, check back next weekend.

[stonewater]

first off this is awesome! a question?

Rollie, the code you have in the above post for enabling the Z min pin in EEprom, does that go into the pins.h or somewhere in config.h? I assume that this will not mess with the printer in the sense that it would no longer look for the calibrated Z zero and instead look for the Z min to trigger to set its Z zero at the beginning of a print.

I just ordered some of the 3 pin connectors from ultimaker so I can use the z min endstop on the Rambo.

Tom C

[RollieRowland]

first off this is awesome! a question?

Rollie, the code you have in the above post for enabling the Z min pin in EEprom, does that go into the pins.h or somewhere in config.h? I assume that this will not mess with the printer in the sense that it would no longer look for the calibrated Z zero and instead look for the Z min to trigger to set its Z zero at the beginning of a print.

I just ordered some of the 3 pin connectors from ultimaker so I can use the z min endstop on the Rambo.

Tom C

That code I provided can be found in your config.h file. Changing this will not affect your Z zero and you can still set it with the method provided in the manual.

Enabling the Z-Probe only allows the printer to utilize the G30 script, nothing else is changed.

[RollieRowland]

Progress Update:
I have concluded that 3 steps of the calibration Do Not affect the XY scaling, this being said the 2 steps that did affect scaling - horizontal radius and diagonal rod calibration - cancel each other out in the changes they make to scaling. However, there is a slight error though in this cancellation - if a tower is off by 5 degrees there will be a maximum change of 0.1mm across the plate (at the zero height) and will significantly get worse as the angle expands. That is the maximum change that this program will affect the printer in the X and Y dimensions and that is only if your towers are not square.

One other feature that I have added is visualizing the error in the towers - angled inward or outward. This was done through a chain of relations that I've found. I'll over-simplify: use the height at the tower and the opposite of the tower to find the angle of the bed in relation to the towers, and use the angles found there to find the difference in the printers radius at the bed of the printer and the top of the printer. To find this relationship your printer must be set to its default settings. Otherwise the angles will be inaccurate. There's one more thing I should mention; this analysis is done via the first iteration of the program, so it does not take any longer to complete.

Since the XY scaling is no longer an issue that I am able to truly fix at all heights, I do not need anyone to test anymore. If someone still reports there data to me though, it will still be very useful.

I still will not be able to test these changes until next Friday, so this is most likely my last update til then.

[Jrjones]

Good to know. I was away for the weekend but just tried the autocalibration with a feeler gauge and my dial indicator (so there was some safety in case of a crash) and it works.

I am going to wait for my angle gauge before I move on.

[RollieRowland]

Good to know. I was away for the weekend but just tried the autocalibration with a feeler gauge and my dial indicator (so there was some safety in case of a crash) and it works.

I am going to wait for my angle gauge before I move on.

Which version of the program were you using?

[Jrjones]

v2.01b (in your About button it says version 3.00). I didn't see the height map update at all though.

I got sick of moving the feeler gauge around so I stretched some aluminum foil over the build plate, I know its not the best for flatness but since I'm waiting I wanted to try it out.

I added the output from the printer console after clicking the "advanced calibration"

[RollieRowland]

v2.01b (in your About button it says version 3.00). I didn't see the height map update at all though.

I got sick of moving the feeler gauge around so I stretched some aluminum foil over the build plate, I know its not the best for flatness but since I'm waiting I wanted to try it out.

I added the output from the printer console after clicking the "advanced calibration"

The version info in the program must've been off, and yes, the height map was not finished in that version. I would have removed it for the time being but it would have been extra work that was unnecessary. This will be fixed in the next version that will be uploaded most likely this weekend.

That is definitely more efficient! And thank you for attaching the log, I'll look at it in a bit.

[Jrjones]

Received my digital angle gauge today.
Relative to the bed +/- 0.1°
X Tower = 90°
Y Tower = 89.9°
Z Tower = 89.8°

And the other way they are 89.9-90° as well.

[RollieRowland]

Received my digital angle gauge today.
Relative to the bed +/- 0.1°
X Tower = 90°
Y Tower = 89.9°
Z Tower = 89.8°

And the other way they are 89.9-90° as well.

You shouldn't see much of an error then. You should still check your delta radii for each tower, then adjust to compensate for the error. After adjusting your delta radii you will have to re-calibrate, as it will offset the z heights all around the plate.

Once I get my towers aligned, this is what I plan on doing.

[Jrjones]

You shouldn't see much of an error then. You should still check your delta radii for each tower, then adjust to compensate for the error. After adjusting your delta radii you will have to re-calibrate, as it will offset the z heights all around the plate.

Once I get my towers aligned, this is what I plan on doing.

How do you check the delta radii? I know what it is but am not sure how to measure it accurately. Maybe using a plumb bob in line with the arm axles?

[RollieRowland]

You shouldn't see much of an error then. You should still check your delta radii for each tower, then adjust to compensate for the error. After adjusting your delta radii you will have to re-calibrate, as it will offset the z heights all around the plate.

Once I get my towers aligned, this is what I plan on doing.

How do you check the delta radii? I know what it is but am not sure how to measure it accurately. Maybe using a plumb bob in line with the arm axles?

All that I did when checking data was rubbing graphite on the nozzle and lowering the nozzle onto paper. You can use the coordinates for the manual calibration on the first page - except replace 113 with 112.583 and you should get exactly 130 for measurements to the X and Y towers (and of course 130 with Z).

This will not give you the delta radius, but it will give you the offset (measured distance - 130) - which is what you need for setting the delta radius in Repetier. And I am sure you could find a way to use a plumb bob.

[RollieRowland]

I am improving upon the delta geometry analysis, I want to make sure the math behind it is sound.

Another correction that I am working on is setting the XY scaling - need to acquire a fair bit of data first. This could correct the scaling for printers that have (almost) no error in the angle of each tower(< 0.5 degrees). However, in printers with offset angles in each tower, the scaling would only be accurate on the first layer, then would gradually degrade in accuracy as the print gets higher. There is no non-mechanical fix for this at the moment, this would involve calculating the angle of the tower then modifying the firmware to make live changes to the delta radii - heh.

I've included a screenshot so that it is simpler to show what all I have deemed possible to calculate so far (this information is only coming from the EEProm and height-map values). Having these values opens up other possibilities as well, but for now it will just let you know if your scaling will be off and by approximately how far, when you reach the top of your build volume(in comparison to the bottom). I will add a feature soon that will let you enter a height and figure the scaling offset at that height - not much work, only factor is finding time to do so.

I will be testing and correcting any errors tonight and Saturday.

[U.S. Water Rockets]

This software looks really impressive. It seems like Rollie is not to far away from a simple automatic calibration that anyone can use, even the non technical person. Nice work.

I saw a sensor in another thread and it looks like a simpler solution than the FSR. Is there any chance this could be made to work with the software: http://forum.seemecnc.com/viewtopic.php?f=35&t=8876 if so, then cool!

[KAS]

This software looks really impressive. It seems like Rollie is not to far away from a simple automatic calibration that anyone can use, even the non technical person. Nice work.

I saw a sensor in another thread and it looks like a simpler solution than the FSR. Is there any chance this could be made to work with the software: http://forum.seemecnc.com/viewtopic.php?f=35&t=8876 if so, then cool!

I like the idea, but they give a warning for delta users. I would assume you need a fairly squared away build before using that type of sensor.

Note for delta printer owners: It is important that the effector does not change its tilt relative to the horizontal as it moves in the XY plane. Otherwise, the height difference between the tip of the nozzle and the bottom of the sensor board will vary with XY position, giving rise to calibration errors when you use the sensor to auto-calibrate your printer. If you are using Duet electronics then you can compensate for effector tilt in the bed.g file, but it is better to avoid it in the first place.

[bvandiepenbos]

I think FSR is the way to go... you are using the hot end tip as the probe, it can't get much better than that. Close 2nd choice would be a simple feeler guage wired to endstop. One wire on hot end the other on feeler guage... when nozzle pinches gauge to bed it completes the circuit as a switch.

[geneb]

Close 2nd choice would be a simple feeler guage wired to endstop. One wire on hot end the other on feeler guage... when nozzle pinches gauge to bed it completes the circuit as a switch.

This is the method I'm going to try for the first run out the gate. I suspect Blue MAX is going to be the scratch monkey for this particular test.

g.

[crocky]

I am improving upon the delta geometry analysis, I want to make sure the math behind it is sound.

I will be testing and correcting any errors tonight and Saturday.

I think I will wait until Rollie finishes his test and puts another version up. I have got the FSR's working (not sure what happens when they triggered) but seem to be having some problems with the Z height not being where it should be. I have done the code changes for the z-probe to be used as per Rollie...... But?

[RollieRowland]

I am still testing, but figured I would let everyone know the current standing.

I have made a few modifications; one being a correction or seven to the geometry analysis, a fix to actually calibrate the delta radii - allowing for accurate scaling, and I am currently trying to speed up the process/accuracy of the XYZ offset calibration. (There were a ton of minor changes that aren't worth mentioning)

I have included a picture of the two height-maps while the delta radii offsets were being calibrated. As you can tell by the picture, the towers and the opposites of the towers are their exact opposites in the current height-map, while (X and Y towers are on) the Z tower was offset by nearly a whole millimeter. This doesn't affect the outcome of the bed leveling, however, it does correct the XY scaling. After this test my scaling accuracy (at Z=0) in the XY dimensions went from 129.5-130.35 to 129.95-130.05 - about as accurate as I could measure.

Crocky, for using the FSRs you want you zprobe height to be 0. And what do you mean the Z-height is not where it should be? Do you mean that it is lesser than normal? If so it is most likely because the mounts you added for the FSRs raised your build plate. Mine shifted by around 2cm.

I will have a version up soon hopefully, I just need to adjust the XYZ offset calibration.

[crocky]

Hi Rollie,

It came down to about 150cm and started its routines there! It didn't get down far enough to get a FSR. I thought I needed the build dia, baud rate, com port no and then connect. There is probably more to it.... I'll wait till the new one is finished....

I reset the z settings again and did a test print and everything is still working

[RollieRowland]

Version 2.0.3PA - Tested without error(stabile on RMax V2)

I am going to go ahead and upload what I have so far, the XYZ offset is not as efficient(higher # of iterations) as I would like, but it gets the job done. Also, the printer analysis is not complete either. I have the equations written out over a couple sheets of paper, but need to transfer it to code, so at the moment it will not be correct. I have included two screenshots; one without changing the EEPRom settings, and one with setting them to defaults. Both worked.

-Do not run this if you zheight is off by more than 10 millimeters(without changing settings), your nozzle may crash into the plate.
-Make sure you change the Zmin type before running.
-Use G31 to check your probe.
-The advanced calibration does not differ much from the basic calibration at the moment, use either.

I will be changing the basic calibration eventually, but for now it uses the advanced calibration code without checking the offsets. It works the same, if not better, but it will take way longer.

How to:
-Run setup for first time use, run .exe for 2nd+ use.
-enter build diameter
-enter baud rate
-select COM port
-check settings
-press calibrate
-when log says "Calibration Complete", disconnect and close application
-print

This program has not been vigorously tested, there may be some errors, so keep close to the kill-switch. If this causes errors, then copy the console and send it to my email with a description of what you think may have happened.

I plan on eventually compiling my notes/diagrams into a readable format to upload so everyone can understand what all is happening in the calculation - most likely in .PDF format.

Download(with Source and .exe):
https://drive.google.com/file/d/0B06WCA ... sp=sharing

[geneb]

I really appreciate your work on this! Does the probing happen at a rate that's slow enough for someone to be able to successfully keep up with the nozzle if using a redneck z-probe? (alligator clip on the nozzle, other wire to a feeler gauge)

tnx!

g.