Welcome to the SeeMeCNC Community

I've had this problem since I first built my machine. For those that don't know its a custom build, frame is all birch plywood, and the extrusions I got from McMaster in an 8ft length, which I divided into thirds. Ever since I first finished it and calibrated it I haven't been able to get it to print entirely flat across the whole build plate. I've spent around 30 hours working on leveling it, not counting the hours reading on the internet on what could be the culprit. I've followed Gene's build guide, the minow blog, Pilot's posts on this forum, several google groups posts, MSURunner's calibration video (which got me the furthest along), and I've tried both Repetier and Marlin auto leveling systems.

I know the build is good, Ive checked 10s of times with a large framing square to see if the extrusions were perpendicular to the bed (just checked now even), everything is nice and tight. I've got the TL CF arms, so they are very accurate.

Part of MSURunner's calibration routine involved printing a tower offset calibration, which I can't do because of the Kraken's nozzles not being in the center. Ive tried screwing with the tower rotation tweaks, which helped a little, but it was still around +/- 0.1-.2mm.

I don't exactly know how adjusting the tower rotation/offsets affect the motion of the effector in relation to the bed, so I'm hesitant to mess with these values any more than I have. The only things I can think of that might help are the tower rotations/offsets, or possibly the birch panels being slightly warped in some way, causing nonsense. I'd prefer to not have to change to melamine, more for the sake that my printer is one of a kind in its current state than the part of rebuilding it (the birch looks reaaaal nice), but if I really have to to get better effector motion I can deal with doing without the eye candy.

Can I ask, not because I think this is the culprit per se, but because I've been doing a lot of experimenting. What is the Horizontal radius that you find is best for you? Mine is at 130.5.

Can you describe more the extent of your problems? What is the size and shape of the area that does provide good adhesion?

Didn't pilot 626 write a flatness correction addition to smoothie ware that solved it? You did read the 17+ page thread on z0. http://forum.seemecnc.com/viewtopic.php?f=37&t=4868

And as I seem to understand, jimustanguitar solved his by shimming the snowflake to get it flat (enough)

I have a bit of the same problem, and am going to try the Jim method first, then after that I eventually will switch to smoothie ware .

I think the radius is around 129.6 or .7, don't have it plugged in to check atm. The shape is sort of hard to describe. I can get each tower and the center at the same height, but the opposite corner might dip or move up a bit. Ive gotten that fixed sort of with rotation correction (not currently fixed at the moment though), but in the spaces between the towers and the opposites of the towers it tends to have little dives or places where it "flies" up a bit. At best, I've gotten a 6 pointed star of sorts with about 50-80mm circle in the center of fairly consistent height with some points coming off of that (I could manage a print that covered most of my built plate, but I had to squish the first layer a bit). The spaces between the points either are also good or they vary quite a bit in a short distance.

Adjusting the angle doesn't seem to have any adverse effects that I have been able to measure. I ended up having the 3 dips between towers at -.004-.005" so I set zero then moved down another .002" and called that zero to have it somewhat averaged out.

The main issues is I don't know what adjusting the rotation on each tower does. I don't know how that affects the delta geometry/calculations, and I doubt I've taken enough trig/calc to understand the delta math in the first place. I know how the 3 main variables, arm length, horizontal radius, and steps per mm, affect things, but rotation and offsets are completely foreign to me.

Playing with tower rotation has helped me, but having a dial indicator mount is a must. If a point between to towers is higher than zero, you adjust the distance the two towers (by degrees) to be greater. the firmware then compensates and you get closer to zero. I only play with the X and Y tower, I leave the Z tower at 90. The final adjustment is shimming the bed. Which I had to do opposite the Z tower, but at least now I can get a 9 inch diameter base to stick.

Bed shimming is out of the question. FSRs mean my bed has to be free floating, I only have a few long M3 bolts running through the empty holes to keep the bed from falling off during transit. I have a dial indicator now, just bought one today. Ive read most of that thread in the troubleshooting subforum, at least the relevant info for us Arduino based board users.