Finally got automatic bed leveling using force sensitive resistors to work on my homebuilt

3D Printing
,
1

Finally got automatic bed leveling using force sensitive resistors to work on my homebuilt mini kossel!
https://www.youtube.com/watch?v=3LP_8gZNHbs

2

Why does it touch down so many times? It only takes 3 points to mark a plane, more points won’t make it more accurate unless your surface isn’t a plane.

3

@Daniel_ShadowDrakken that is true for a cartesian bot where the axes are linear, but not for a delta though. Deltas are actually pretty terrible at keeping a constant Z-height - you’ll often end up with the first layer being squished in the center and thicker at the edges. So to compensate for this you need many points.

4

I thought the many point was to account for form error of the measurements. I was also wondering… Does the autolevel account for inconsistency in the bed flatness or does it average it into a theoretical plane based on the points.

5

@Eclsnowman it’s not plane-based math. The reason for the multiple points is to account for each of those multiple points in the planner path. If you throw a stack of pennies at one of the measurement points, just that point is raised up, while spanning out from it will slowly make its way to the level of the other probed point. It’s a thing with deltas…you can’t just do a 3-point leveling like you can with cartesians.

6

Mind sending me your configuration.h? I’m in the middle of doing the same thing. How are your FSRs connected to the controller?

7

@Mike_Miller they’re wired in parallel and just hooked up to the Z_MIN endstop. It’s basically the same configuration.h as in Johann Rocholls: https://github.com/jcrocholl/Marlin/tree/fsr

8

With no voltage divider/conversion to binary?

9

Nope, I didn’t need a voltage divider.

10

hmmm…I am not sure if this term is right for it, but I think it might be considered parabolic math. It is essentially a big 3D curve.

11

actually I hate to disagree but if your bed is flat you absolutely can make the head tram to the bed from three points and adjusting the endstops (or use M666) the delta math allows for that. Probing a grid and doing z mapping is what you do for cartesians and IMO is totally the wrong thing to do for deltas as pointed out deltas do not move linearly, and Johanns leveling uses linear interpolation of the z grid. it works well enough but is an approximation at best.

12

@Wolfmanjm 4 points. You must test the center.

13

Yes it’s an approximation, but it’s the best way to get your axes aligned with the bed. Theoretically it can be calculated from just 4 points, but an atmega 2560 is not exactly good at doing that kind of complicated math. It might be doable on the new fancy ARM controllers though.

14

Sorry for following up on such an old thread. Maybe you or someone else can give me a hint.
I am using a Kossel Mini, added FSRs to the heatbed sensors. I basically followed the documentation [1] and [2].

Took me a while, but I the Marlin fsr [3] is building now and installed. When I connect to the printer using pronterface, push on the unheated base plate I see the temperature of the bed “rise” - so the FSRs seem to do something.

I never get z_min triggered though, sending the printer a G29 command results in some stragne movements, but not what I’d expect from autlevelling.

Any clue? Thomas, could you send me your Configuration.h so I can crosscheck?

[1] http://www.mini-kossel.com/FSR-SETUP-GUIDE
[2] http://reprap.org/wiki/FSR
[3] https://github.com/jcrocholl/Marlin/tree/fsr