Hello! I wanted to ask you if you remember any web page, tutorial or video tutorial for the delta calibration. I have found only one nice. I had built a Smartrap, and now I am upgrading it to Rostock Mini Pro, this week would arrive the Traxxas joints and would be moving 
Thank you in advanced
@Whosa_whatsis is doing a lot of work in that area.
You found a good site. Also check out Jay Coutour’s videos e.g. Manual Delta Calibration using M665 and M666: http://youtu.be/tDLbqLve128
Or just use @Brad_Hopper version of Rich Catell’s autocalibration code https://github.com/braddo99 and let the printer do the work for you (a z-probe would be needed)
Most people say, and I agree, to manually calibrate using those resources. Then fine tune using auto calibration. The reason I like Rich’s autocal is that it compensates for tower position errors which AFAIK there is no way to manually fix. My version of that code adds very little but its useful if your probe has some variability. The original expects extremely repeatable probing which isn’t always realistic.
Thank you all! and what must i need to add the printer to run the autocal? any servo?
@Javier_Prieto z-min probe (it may be just an endstop).
@Miguel_Sanchez the z min endstop pushed by the z carriage?
@Javier_Prieto the z-min will sense the bed just before the tip of the hotend would touch it. It sits in the effector and moves with it. There are different ways of deploying the probe: https://www.youtube.com/watch?v=1eNz1l56H5E
@Miguel_Sanchez ahh, I understand, it was the same I was thinking hahaha And it is only connected to the z min endstop in the ramps, and with the G29 it does all alone? Wuau Also, the main Marlin branch does have it implemented, doesn’t it?
Autocalibration is G30A actually. G29 is auto bed leveling. Firmware link was provided above.
You don’t have to use a probe, you can use FSR’s as well. Either work just fine.
G29 is all I use, not G30 as that is only in Cattell’s code.
The most important thing is calibrating/presetting flatness with DELTA_RADIUS and X and Y dimensional accuracy with DELTA_DIAGONAL_ROD.
@Andrew_Hodel I have an endstop spare, so I will use that, but I add to the wishlist the FSR.
So, first calibrate DELTA_RADIUS and DELTA_DIAGONAL_ROD and then… auto?!
Here are my notes:
steps per mm and Z calibration
G28
G29
G1 Z0 X0 Y0 // ensure 0 is at bed
G1 Z100 // measure distance from bed to nozzle tip, should be exactly 100
If it’s not 100, then add or subtract the ratio’d difference to steps per mm against what is calculated for the belt/pulley
On a delta you must calibrate the Z axis first, as X and Y are functions of Z. Each tower must move exactly 1mm for a Z change of exactly 1mm.
When testing don’t ever use G1 Z0 because you might slam into the bed, instead use a 50mm or 100mm object and test with G1 Z50
Once you’ve adjusted the steps using the ratio of the measured difference, test with a 100mm or higher object as errors will be more pronounced at higher distances. Adjust XYZ_STEPS slightly using very small (.3) increments until things are perfect.
calibrating flatness with DELTA_RADIUS
Calibrate X and Y, as they are functions of Z on a delta they are dependant on the calculate_delta() function in Marlin_main.cpp which uses only the DELTA_DIAGONAL_ROD and DELTA_TOWER positions from Configuration.h. DELTA_TOWER is calculated from all of the DELTA meausurements however only the DELTA_RADIUS is used in the math. All of the other measurements exist just to simplify measuring (I guess).
Set DELTA_RADIUS to the measured distance (easiest using a square and ruler). There is a diagram image showing what exactly that is attached.
When moving the nozzle across the bed, unless DELTA_RADIUS is perfect it will not move parallel to the print surface. It will either be a bowl or an inverted bowl.
If the nozzle is higher in the center, DELTA_RADIUS should be decreased. If the nozzle is higher on the edges of the print bed, DELTA_RADIUS should be increased.
calibrating X and Y dimensional accuracy with DELTA_DIAGONAL_ROD
The easiest way I found to perfectly calibrate X and Y is to draw a line that is parallel to the X axis on the print surface then mark each 10mm from the center.
Flash the avr and do an autolevel, then G1 Z2 which should put the nozzle tip directly on the center mark. Now G1 X65 (or whatever will bring it to the farthest mark) and note the offset of the nozzle and 65mm.
Lowering DELTA_DIAGONAL_ROD will increase the distance travelled and increasing DELTA_DIAGONAL_ROD will shrink the difference.
Once you dial it in perfectly it should be perfectly calibrated in X, Y and Z.
@Andrew_Hodel hohohohoho I think I have just fallen in love. Thank you very much! I apreciate the help provided. If I could, I would +1 your last comment about one hundred times
You guys are making the muscle in my face twitch from Delta calibration flashbacks. #twitch
FWIW, all of the auto-measuring math and adjustment just masks errors in the physical design. I didn’t TRULY stat to get dependable prints until requesting 10 mm above each tower REALLY RESULTED in the nozzle being 10mm above each tower…then 10mm above 0,0 (This required lowering the Y-tower by 2mm) and even then, I’d have to send M666 X .6 Y.2
Once I got the plate to stop rotating, those were the break-through steps that got me close to dependable prints.
I then pulled my hair out trying to get holes to align for NEMA 17 brackets and could only get close to doing so by scaling by .975…I never DID get it dead-on.
My first cartesian (after all the heartburn of calibrating the Delta) was dead on, second print, day one.
I agree completely about the auto measuring stuff, Cattell’s code really doesn’t work all that well unless things are nearly perfect anyway.
Mechanically it’s easy to get things correct if the person who builds it just sets it up right. Then the software can be tweaked using the instructions I gave earlier very easily.
Auto level is the exception, as it’s not doing anything fancy just plotting out the bed distance over a matrix which will change due to endstop mechanical tolerances (this could be fixed with opto etc), blue tape/surface coatings and the sockets for the ball joints wearing out.
All in all for 3d printing the Delta is a better platform imo, more than anything I like the ability to cleanly Z hop (meaning 3 motors are moving already not having to turn a ballscrew on a new motor) which greatly helps with printing new layers which have a slightly greater x,y than the previous. The edges tend to curl up and the Z hop lets the nozzle approach from above instead of from the same XY plane.
The tradeoff is the bowden tube, but all in all you are going to get faster prints and better acceleration with less weight being moved around and that’s going to be mechanically less destructive on the printer as a whole over long periods of time.
I agree with you both, my first printer was a Smartrap, and the design was awful, but having in front the Rostock Mini, I think it has a very good design. Will look for a laser cut, for the plates, as my cuts have been precise but would prefer a profesional ending. Thanks all of you