Rostock Max Geometry Error?

[wonderemporium]

I hate to keep dumping weird problems on here but this one is stumping me. I have been having constant problems with getting the heated bed level and keeping it level. I added 3/8 glass on top and verified that it is absolutely flat to less than 0.001". I made some acrylic clips to hold the glass top down on to my Onyx bed and it is all rock solid.

I have the acrylic machine and the whole thing creaks a little as it heats up and I've been wondering if it moves at all as the temperatures change. I followed the manual and got the head to exactly identical measurements to the surface at each of the 3 axis towers. I can run it up and down and move to all the reference locations and they are all perfect at that point.

Now I move to the center and the head is 0.4mm too high. This has been repeated more times than I care to admit but I always get identical results. My brain is hurting from trying to figure out the delta arm math to see what could cause this geometry error. My only thought is the vertical aluminum extrusions not being perfectly plum? I don't know how to measure them accurately enough to prove that out. Anyone else ever see something like this or know what could theoretically cause it?

[v8tony]

I had the same problem and it drove me crazy... I had trouble getting my first layer to stick on parts that were larger than about 25mm in diameter (because the motion of the platform was essentially concave)...

Here is what I have done (and it took me waaaayyy to many hours to get to this point)...

There is a variable in the firmware that is:

#define DELTA_RADIUS (DELTA_SMOOTH_ROD_OFFSET-DELTA_EFFECTOR_OFFSET-DELTA_CARRIAGE_OFFSET)

Here is what mine looks like now:

#define DELTA_RADIUS (DELTA_SMOOTH_ROD_OFFSET-DELTA_EFFECTOR_OFFSET-DELTA_CARRIAGE_OFFSET-1.15)

The 1.15mm subtraction changes how concave (or convex) your motion is away from center by changing the effective radius of the delta arms. I moved it in 0.1mm increments but, my machine needed this much offset. I didn't go straight to this point... I've taken my machine apart about 3 times now and checked that everything is true and straight... it is...

In the picture attached the left circle represents the best I could get my machine before tonight and the right it where it ended up. There were 2 steps to get here...

1 - take my arms off and refit them. I spent about 2 hours fitting everything when i built my machine, but, I obviously left a little too much friction. I noticed that if I moved the arms around a bit after I had stopped the platform to look at the hight I could get the platform to "settle" in one spot by sqeezine the arms together and letting go. I also noticed that i would get inconsistent measurements depending on which direction I was moving to that position from... I think refitting the arms helps on the opposite side from the towers because of the angle on the joints at those times...

2 - I went through trial and error leveling the 3 towers and the center with a dial gage mounted to my platform to get the motion as level as possible. I was still left with the center being "lower" than the 3 towers no matter what I did... Thats when I started going through the math on how the motion works for the delta arms (I had never done that before)... I found the delta arm math in the main marlin file in the firmware. You can look at what makes up each component, I'm not sure why mine doesn't work or if they will all be that way, but, this sure does have my first layer working better than ever... From looking at your post you might need to go the other way and "add" to the radius to flatten the center out (if I understand right)...

Now, I still see issues with "motion" where it will pick up about 0.012" of an inch during travel in one direction and not the other. I think it still might be related to friction somewhere in the system, or maybe belt stretch... Mine moves towards the towers dead level but comes back to center with a "hump" where it picks up and come back down... doesn't seem to change with speed between 100 mm/min and 12000 mm/min... I guess the other option is that it might be software... I'm open to any suggestions or observations if other people bot's work the same way...

Sorry for the crazy long response...

Tony
Tony

[wonderemporium]

Tony,

Wow, that's what I call a thorough and informative reply! You may have just saved my sanity. I will play with that variable this morning. It makes perfect sense and my measurements are very close to yours, just inverted I think. I'll post my results here after the change.

Thanks again!

Patrick

[wonderemporium]

This appears to have gotten me really close, the measurement errors I have remaining now are due to damaged belts I think.

I set the DELTA_RADIUS to the following...
(DELTA_SMOOTH_ROD_OFFSET-DELTA_EFFECTOR_OFFSET-DELTA_CARRIAGE_OFFSET+0.9)

So my additional 0.9 seems to get me extremely close to flat, +/- 0.005"

I will wait for a final tweak until I replace my damaged belts but I'm definitely on the right track now.

[v8tony]

Cool, glad it worked out for you. I'm very happy with how things are working out since I added my offset... I'm able to get the first layer on pretty large objects to stick much easier now...

Tony

[kirbymills]

I have the exact same problems, Zero in front of towers & middle, z lifts -.018 between the towers on the outside of the radius. I have been trying to square the towers to the bed in both directions ( left-right & in-out ) having some success, I'm down to -.008 . The towers don't hold their position very well (NOT STURDY) so I'm going to make small turn buckles at 45 deg angles at the bottom of each tower. I am also going to make a measuring device to measure between towers to insure parallelism ( This will also allow me to check tower location) I don't know what else to check, it's got to be DELTA GEOMETRY. I have never seen anyone print 11" diameter, and I think this is why, Rather than bitch about it, I'm gonna fix it. If all goes well I'm going to sell a kit to calibrate the Delta Geometry. I might of found it, although my towers are parallel, they seem to be on a helix ( slight lean counter-clockwise) which I think will cause z lift outside the Delta triangle. Either way it shouldn't be there. I'll have my guy supports done in the next couple days.

[kbob]

although my towers are parallel, they seem to be on a helix ( slight lean counter-clockwise)

Interesting! In all the alignment procedures I've seen here, none have tested whether the towers spiral. It seems like you could detect it simply by using a carpenter's square in more than one direction.

To correct it (if that's really what you have), are you going to have to reassemble the base and verify that the -- uh, the boards the towers are screwed onto -- are square to the top and bottom?