My delta refuse to get closer to the bed then 1mm

SlowFoot

Hi all!

First post and with a strange problem.

I am trying to get my delta up and running and it is getting close but not quite close enough.
I run RAMPS on Repetier 1.0dev. Auto-calibration works, it is a  limitswitch on a servo.
But it refuses to get the nozzle closer to the bed then 1mm. If I jog it, it simply stops moving at Z=1 and if I run a print it will pause at first move to the bed. If I unpause the print it will extrude a mm over the bed. I use a 3mm block to check Z height, and it is ok.
I guess some setting, somewhere, is off a mm.
Any suggestions ?

(Previusly posted at RepRap forum without any response)

Cheers!

Repetier

I hope you have set z-min endstop to not defined, otherwise will the z-probe work as min endstop when being triggered. I see this frequently but the thought is not correct at least for a delta. Ther eit is just a z-probe connected to z-min endstop pin. No reason to enable z min endstop.

Otherwise z length is 1mm to short. Increase it in eeprom and it will go deeper. If z length was computed with autoleveling yout z probe height would then be 1 mm wrong.

Last thing possible is that after hitting top endstops you do not go down a bit. So every tiny sid emove might git endstops causing wrong positions.

SlowFoot

Thanks for the answer!
I'm not with the printer right now so I can not check the Z endstops min, I changed the probe switch and that changed the triggerpoint but not height it stopped.  
Z height is correct, checked that numerus times in different ways. If I increase Z height the nozzle can touch the bed.
I don't go down after hitting the stops, will try that tomorrow.
Cheers!

SlowFoot

 At it again
All MIN_HARDWARE_ENDSTOP are "false"
All min_software_endstop are "false"
ENDSTOP_Z_BACK_MOVE is 2, enough disengage the switches.  
Z height is still correct, G33 did not adjust Z height, G32 did but did not account for endstop back move so nozzle ended up high.  

And the nozzle still stop at  Z=1.0 

Roko

Hi Slowfoot, I don't have a Delta anymore but I just checked  my configuration.h for it (I save everything) and I see that my min_software endstops are set to "true" I used a different pin for my autolevel probe than Z-min  so I guess you would leave that false.
Also did you try disabling the autolevel with M321 - deactivate  autolevel temporarily, and then set your  Z=0?. If I recall you can do it through the LCD menu very precisely. Then test it by homing and then going to Z=5 and moving down in manuall to Z=0? It could be a setting for your autolevel probe that is skewing the results.

SlowFoot

Hi Roko! With autolevell off (via M321) and jogging with the LCD control it will still not go below Z=1. When I go below it will turn and go up some.

I changed the probe to a different offset and that did not changed the way the printer behaves, also Z=0 is at bed level, so probably not the prob.

   

MartinH

Hi Slowfoot,.It might help if you could post(link) your config.h so we can look thru it.

Repetier

"When I go below it will turn and go up some."
Sure? Don't see why it would go up. I understand stopping under some condition but don't know a part in firmware that goes back except homing/probing of course.

What does M119 show? If you see z-min in answer then you have the error I told you in config.

SlowFoot

It is not mechanical, I added about 1.2 mm to the bed (a stack of paper sheets) and ran a M32. The error moved with the surface.
Hope to be able to shared configuration.h shortly.
   

SlowFoot

Hi MartinH and all others!
New to Google Drive too, don't know if I got this right. Config.h 
I'll make a video tomorrow if I can.

Cherrs!

RAyWB

this might be the issue

#define UI_BED_COATING 1

set value to 0  , think it will go down to 0 then

btw the configuration is 0.92 , not dev

SlowFoot

Hi RAyWB.

The UI bed coating did not help. unfortunately.
Wrong version might explain a few things. I was about to start over anyway.

Cheers!
 

SlowFoot

The previously promised video 
I am sort of leaning towards that it is a bug. As Repetier pointed out this is not what it should do, ever.

MartinH

.92 is so out of date now,update firnware to the dev version, you wont regret it

Repetier

#define Z_PROBE_Z_OFFSET 0
is what should be 0 RAyWB mixed these. But in any case you should also check it in eeprom (bed coating it is named there). If you have a inductive endstop this can in deed make it stop 1mm before bottom.

Otherwise increasing zlength by 1mm always goes down 1mm more when going to z=0.

And of couse my latest states are still not answered so they are also potentially valid.

SlowFoot

M119 show "L" for the Z-probe when I try to move in to the Z<1-zon. Status for the z min endstop also show in the video.

I made a new firmware, 1.0.0dev this time, but it will not compile correctly on my Arduino 1.6.5 r5 and my Windows 10 refuse to run newer versions of Arduino. So for the moment (the local "helpdesk" is on vacation) I'm stuck with the 0.92.

While running the 1.0.0dev setup I could not find where "EXT0_STEPS_PER_MM" is set and I noticed that  "PRINTER_RADIUS" is set to same value as "ROD_RADIUS" (I think a delta with that setup would be a bit limited).
I'm also still a bit confused why Z height have to compensated by hand for the ENDSTOP_Z_BACK_MOVE.

Thank you all for your interest, keep the suggestions coming and Cheers!

Repetier

Ok just an idea - try 
G33 R0
to delete the distortion map.

Maybe first check values with 
G33 L0

at least if you are using dev version. Default config for distorion map is to start at 1mm and if you have big corrections stored at that point it will start to correct.

Just an idea what it could be.

SlowFoot

I think you might be on to something there.  
"G33 X0.00 Y0.00 Z2.28"

And after a "G33 R0" it works beautifully. :fireworks:

How do I proceed from here?
My guess is that I have to calibrate the delta radius by hand befor I auto calibrate, Z is dropping towards the towers.
Lowest Z in the correction table is 1.462, how do I bring that down?

Thanks and Cheers!

Repetier

diagonal length / horizontal radius ratio mismatch is the main reason for a big bump/dome with center as apex. So change one of them to fix it. Normally you can measure diagonal length quite exactly so I would change horizontal radius. You can also try searching "escher delta calibration" and use the method with fixed diagonal. If you do not fix diagonal it will also fix z but you get geometric errors in xy so better keep that fixed.

Only when you have calibrated as good as possible you could think about using G33 again. Then values should all be < 0.5mm. With good calibration you get below <0.2 with calibration allone in most cases.

SlowFoot

A little report.
I put in lot of time trying to get a good outcome from the G33 but no luck so far.
Tonight, out of frustration, I thought "what the .." I know my measurements are fairly good, I know the endstops are good, I calibrated the delta radius and the print bed is not that crooked, so do I really need that autocalibration to print? I put a thicker paper (0.2-ish) on the bed, lower the nozzle till it touched and started to jogged the head around the bed. It tugged the paper a little in the +X +Y area, but did not rip it. I printed on worse. 
I will return to autocalibration when I need it, for the time I'm doing OK without.
Cheers! 

Repetier

It's never the bed that is so bumpy. It is the dependency from 8 geometric variables in nonlinear fashion that makes the extruder go up/down. Even the angles between towers have big influence. 0.1° error and it makes a bump. And it is nearly impossible to build a delta with absolute precision.

SlowFoot

Well, in the choice between an autocalibrating printer that don't work and one without autocalibration that do work, I pick the later. At the moment I can not do 0.1 mm first layers (unless I use a raft) and that is about as far as the drawbacks goes.