delta printer levelling..... big problems

RhinoRap

I'm working on my new Delta - bed is as mechanically level as i can achieve so I wanted to use te distortion matrix to correct for minor irritating deviations i have.

I've set the probe height to the trigger height as per the Documentation ( which could do with updating).
Bed is 200mm wide, so I have set the matrix radius to 80mm to take in to account the screw holes and a safety margin.
Probe is offset to the z tower by appx 15mm, which is set in the FW.
Z0 is kissing the bed nicely in the center.
Test probe reads and confirms functionality.

I'm using G33 to activate the routine.... and this is where it all goes wrong..... :-/

The grid starts to probe inside the bed radius, but it is offset to the y tower and drifts over the extent of the bed towards the front of the printer.... and the probe misses the bed. ( interpolate corners is enabled).
If I reduce the radius to 60mm the probing stays on the bed ( not ideal for a sample area).

The distortion map doesn't seem to be applying to the prints even though it's saved to EEprom, and activated. I say this because one area of the bed I know to be sagging a little has gaps between the extruded material passes - because the nozzle gap is too much - so the height maps can't be applying properly.

Also, the Bed map tool in Host is behaving VERY oddly - it'll do one probe then crap out and stagger multiple probes in an upwards motion towards Zmax until i press the kill button....

This is becoming very frustrating as I prefer this FW over Marlin due to the functionality - the documentation lets Repetier down but I can sort of live with that as the forum is littered with posts about inconsistencies which generally clear things up.

I'm sure I'm missing something somewhere... is G33 the only levelling routine I need to run for a Delta ?

Repetier

G33 is the last you would run. You need to have run G32 S2 before you continue with G33. It is meant for small bumps and correction reduces over z so only at Z=0 you normally get full correction and depending on your configuration it reduces if going upwards. It is just a last resort to fix geometry errors you could not fix by calibrating dimensional data.

Be aware that G32 measurement area and G33 area is configured independently.

Are you using latest dev version or 0.92.9? Especially for z probing I still find special cases and smaller bugs so I would always use dev version.

RhinoRap

gotcha - i'll give G32 another run and check I've set it properly.
Correction gain for G33 at present is set to .5 and 1 respectively - so I'm not looking for huge correction - however as you have pointed out my glaring cock up i'll go run them both in sequence and see what happens.

I'm using 92.9 at present as it seemed to be working great for everything else - if I don't get a satisfactory result today with more tinkering i'll pull down the DEV and see what happens - thanks @Repetier ;-)

RhinoRap

Hmmm - so i've been "fiddling" and the z correction on G32 S2 still does not seem to be compensating for the bed irregeularities.

I decided to try the Dev branch but this is reporting an error in itself (i'm using the same Z probe offests as the 92.9 branch) The probe normally reads a z height of appx 12mm....

11:03:53.065 : fatal:Could not activate z-probe offset due to coordinate constraints - result is inprecise! Printer stopped and heaters disabled due to this error. Fix error and restart with M999.
11:03:57.673 : Z-probe:22.119 X:-60.00 Y:-55.00
11:04:00.184 : Error:z-probe did not untrigger on repetitive measurement - maybe you need to increase distance!
11:04:00.196 : Error:Probing had returned errors - autoleveling canceled.

RhinoRap

Quick update, @repetier....

Ive tickled the Z probe height to compensate for a bit of z probe error and it seems to be "ok" with the rotation correction now (on 92.9).

Dev branch still throws the error above, so i've given up with it for now.
Bed map function in Host still behaves very odd - I personally can live with it not working as I've never used it before, but it's a feature that isn't working ( in my case at least) which I feel you need to be aware of.

Keep up the good work ;-)