Delta bed levelling

I have an interesting one here I cannot seem to find online anywhere.

Apparently, the most needed adjustment to level a delta is offsetting the delta radius to compensate for a concave or convex bow in the z-height, from outer points at the towers to the center of the bed. Ok, so I'm all set with that; my z-height is very even at points "C" "X" "Y" and "Z". So given that, everything should be great, right? wrong.

at points "F" "L" and "R" I am evenly, consistantly, 2mm too high. :-/

how can this be? Something with the math is not adding up. How can I be level at center and outer poles, but be off at the outer extremes between poles?

What value can affect this?

Everything I can check and test tells me my values are set correctly, even after my manual calibration of this machine and z-height, my values perfectly match the seemecnc firmware posted just days ago.

So what am I missing here?

Thanks!
-Dave

Edited 1 time(s). Last edit at 04/19/2013 11:30PM by xclusive585.

I think it's likely you have a geometry problem.
The only things that matter in the firmware are setting th plane so it's level and removing any curvature from that plane.
If you're seeing errors at FLR, the odds are the towers aren't evenly positioned or they aren't parallel.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

Dave, did you find a solution? I have exactly the same problem; triple checked the geometry . . cant find any issues.

What firmware are you using, Repetier at least has a bug in it, the steps/mm must be set correctly both in configuration.h AND in the EEPROM settings.
If they do not match you get weird motion, the issue is that the actual delta calculation uses constants derived from the configuration.h settings, but the motor motion is based on the EEPROM values.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

Wow, I wish I had known that sooner...

But my eeprom should contain all the values found in config.h as last time I compiled I used a <>0 besides one, which forces eeprom to be overridden


And to answer, no I havent found a solution, I'll pull values from eeprom and check that they match config.h.

I actually just got back on this today after giving up a few weeks ago.

Will update l8r


Update, yea no luck. EEPROM contains same steps/mm that are calculated out in config.h

So, nope, same damn issue.




I'm going to give in, and try software levelling, but I doubt that's going to make a difference unless it can be told that points F,L,R need to be lower.

Another issue I am having is software stops just dont work with Repetier firmware for me. You tell the bot to go to a size way outside of its area, and it will try to go... :-(

I really may just try Marlin Delta. I've had too many complaints about repetier firmware after having such good luck with marlin on my cartesian.

Edited 2 time(s). Last edit at 05/19/2013 05:06PM by xclusive585.

Hopefully you do not have skewed the towers. That changes things and what I'll write does not apply.

So lets assume that your towers are all parallel/equidistant with each other (if you drilled both top and bottom plate at once this should be easy to achieve).

Your tower locations and diagonal rod length can be wrong. You can go on and write down all the queations for the model in the firmware and for the real situation (i.e. this second set will contain error terms for tower locations and diagonal rod). Then measure distances from heatbed at your locations and solve the equation set to find all the error terms. Then fix the parameters in firmware based on the error terms.

If you are lazy to do that then you can wait till nicholas.seward does it for his printer and adapt them to the common rostock.

Or here are some tips to go on manually. This is an iterative process, so after each modification you will need to start from the beggining. The changes will influence each other.

* First level X,Y and Z points with endstops.

* X,Y,Z are leveled. If the head is too high/low in C then you need to lengthen/shorten diagonal rod length a bit (about as much as the difference).

* X,Y,Z,C are leveled. In the next section fix the worst error first (if they are not all the same).


* if points L,F,R are all too high/low by the same amount, then your delta radius needs to be lengthened/shortened (by somewhat smaller amount than the difference); this can happen because you have entered wrong tower/effector/carriage offset (any of them)

* if for example point R is high/low and L,R are leveled, then you need to move coordinates of X tower further/nearer to C (i.e. you will modify DELTA_TOWER1_{X,Y}

* if your towers are placed very badly then you may need to move tower locations in firmware clockwise/counterclockwise depending on where the leveling is worst; the rough idea is that wrong tower location most screws up the leveling just oposite of it

Have a luck.

If you happen to write all the equations and solve them, then please post the math.

Are You quite sure your glass/plate/heatbed is totally level at all points? This would be the obvious culprit? no warping?

FinnRap

You need one of these...

Auto Bed Leveling

---

Protoneer.co.nz - Blog - [blog.Protoneer.co.nz]
Protoneer.co.nz - Shop - [shop.Protoneer.co.nz]