So now that we have some very nice drawings to look at - back to the calibration. I made a comment about starting with a flat print and never explained why. Here's what I'm thinking:
1) Make a simple stick and use it to get all the lead screws spaced the same. Do this on the top and bottom plates.
2) Check your other parts to make sure they all are the same.
3) Check your support pipes to be sure they all are the same length and are square.
4) At this point you don't know any real distances, but you should have symmetry. Hopefully it’s good to some pretty small number (< 100 microns).
5) Get each axis zeroed as best you can. No 100 micron guarantee on this.
6) There are three straight lines you can lay down on the build plate at this point. Each goes through the center and one of the lead screws. Since you aren’t fully calibrated they don’t quite hit those points, but they should be pretty straight in X and Y. They will not be straight in Z.
7) Either print a long thin part or put a dial indicator in place of the hot end. Trace out each of the lines, collecting Z data for each estimated line location. Collect 10 or 20 points per line. You will get an error curve that tells you what you need to know.
8) Plot out the data and ignore the bed level for now. The wiggly line that’s left has a signature in it for arm length, shoulder offset and hub offset. When they are all dialed in correctly you get a straight line in Z. You need to be sure the offsets are correct as a percentage of the distance between the lead screws. Fiddle / analyze / model each until you get a straight line. If they don’t all converge, one of your supposedly equal distances may not be (or......).
9) Once the wiggle lines are fairly straight from your math fitting, use the bed level information to first order correct that.
You are not anywhere near done with the calibration, but you have the crazy part taken care of. You still need to play with bed level and the home locations on each axis. As you get them worked out, you may need to repeat the offset check. You may be able to play the same trick (use a stick) on the diagonals to shortcut the level stuff. At the end of it all you need to print a cube to measure and calibrate the real distances between everything.
Since this is a lead screw machine. I’m guessing that the steps / mm will be pretty close simply by how it’s built. If not, the cube will show you the error.
Yes this assumes you can trust the flatness of your build plate.
-----------
Is this the only way to do this? No, of course not. If there is one way, there must be a bunch of ways. Is it the best way? Probably not, there likely are some subtle interactions that might be taken out earlier in the process. All I'm trying to provide is a method that indeed gets the printer to a very accurate state without a bunch of crazy measurement tools. If you have a 36" micrometer, feel free to use it ....
1) Make a simple stick and use it to get all the lead screws spaced the same. Do this on the top and bottom plates.
2) Check your other parts to make sure they all are the same.
3) Check your support pipes to be sure they all are the same length and are square.
4) At this point you don't know any real distances, but you should have symmetry. Hopefully it’s good to some pretty small number (< 100 microns).
5) Get each axis zeroed as best you can. No 100 micron guarantee on this.
6) There are three straight lines you can lay down on the build plate at this point. Each goes through the center and one of the lead screws. Since you aren’t fully calibrated they don’t quite hit those points, but they should be pretty straight in X and Y. They will not be straight in Z.
7) Either print a long thin part or put a dial indicator in place of the hot end. Trace out each of the lines, collecting Z data for each estimated line location. Collect 10 or 20 points per line. You will get an error curve that tells you what you need to know.
8) Plot out the data and ignore the bed level for now. The wiggly line that’s left has a signature in it for arm length, shoulder offset and hub offset. When they are all dialed in correctly you get a straight line in Z. You need to be sure the offsets are correct as a percentage of the distance between the lead screws. Fiddle / analyze / model each until you get a straight line. If they don’t all converge, one of your supposedly equal distances may not be (or......).
9) Once the wiggle lines are fairly straight from your math fitting, use the bed level information to first order correct that.
You are not anywhere near done with the calibration, but you have the crazy part taken care of. You still need to play with bed level and the home locations on each axis. As you get them worked out, you may need to repeat the offset check. You may be able to play the same trick (use a stick) on the diagonals to shortcut the level stuff. At the end of it all you need to print a cube to measure and calibrate the real distances between everything.
Since this is a lead screw machine. I’m guessing that the steps / mm will be pretty close simply by how it’s built. If not, the cube will show you the error.
Yes this assumes you can trust the flatness of your build plate.
-----------
Is this the only way to do this? No, of course not. If there is one way, there must be a bunch of ways. Is it the best way? Probably not, there likely are some subtle interactions that might be taken out earlier in the process. All I'm trying to provide is a method that indeed gets the printer to a very accurate state without a bunch of crazy measurement tools. If you have a 36" micrometer, feel free to use it ....