Auto_Bed_Leveling: Z-Probe Repeatability code

[Trike]

Here's my result:
>>>M48 x 100 y 48 V4
SENDING:M48 x 100 y 48 V4
M48 Z-Probe Repeatability test. Version 1.85
Positioning probe for the test.
1 of 10 z: 4.321000 mean: 4.321000 sigma: 0.000000
2 of 10 z: 4.327250 mean: 4.324125 sigma: 0.003125
3 of 10 z: 4.324750 mean: 4.324334 sigma: 0.002568
4 of 10 z: 4.328750 mean: 4.325438 sigma: 0.002934
5 of 10 z: 4.328500 mean: 4.326050 sigma: 0.002896
6 of 10 z: 4.325250 mean: 4.325917 sigma: 0.002660
7 of 10 z: 4.324750 mean: 4.325750 sigma: 0.002496
8 of 10 z: 4.327000 mean: 4.325906 sigma: 0.002371
9 of 10 z: 4.323750 mean: 4.325666 sigma: 0.002336
10 of 10 z: 4.326250 mean: 4.325725 sigma: 0.002223
Mean: 4.325725
Standard Deviation: 0.002223
echo:endstops hit: Z:4.33

[Roxy]

Here's my result:
...
Standard Deviation: 0.002223

I'm Jealous!!! That is amazing! You must have very good equipment!

You can try:

M48 x 100 y 48 V4 E L 8

To beat on the hardware a little bit harder. That will engage and retract the Z-Probe for each measurement and move the X & Y axis around between measurements. Probably, you will get similar results, but not as good as the one you just posted.

[Trike]

I'll give it a try. I'm just using a cheap kw12 switch and a hitec HS65HB servo. This is my first printer (prusa i3 rework), so i'm still dialing it in and getting a feel for it. The bed is con-vexed in the center once it warms up. I was getting pretty inconsistent prints. Once I enabled auto bed leveling and used 25 probe points, everything magically started printing good.

I like your code, it'll at least give me a baseline to work off of to see if the switch might be failing.

[Roxy]

I'll give it a try. I'm just using a cheap kw12 switch and a hitec HS65HB servo. This is my first printer (prusa i3 rework), so i'm still dialing it in and getting a feel for it. The bed is con-vexed in the center once it warms up. I was getting pretty inconsistent prints. Once I enabled auto bed leveling and used 25 probe points, everything magically started printing good.

I like your code, it'll at least give me a baseline to work off of to see if the switch might be failing.

A lot of people want to use a 3x3 grid. Probably because in the Configuration.h file it says there shouldn't be a need to go higher than that. But I know 4x4 works much better than 3x3 and 5x5 works a little bit better than 4x4.

How do you know there is a high spot in the center of your bed? (I'm guessing that is what you meant by 'convexed') Did you see that with the T option on the G29 with a 5x5 grid?

[Trike]

Yeah, I did. I tried shimming one corner to see if it'll help, but one washer was too much. It was worse before the shim. You can see the values increase around the center of the bed. Here's my current topography after the attempted shim in the back right corner.

Bed Height Topography:
--0.32057 --0.14357 +0.05043 +0.16118 +0.18893
--0.35782 --0.15632 +0.02168 +0.14243 +0.16393
--0.33832 --0.11982 +0.06268 +0.19568 +0.23068
--0.31157 --0.08457 +0.11868 +0.24968 +0.29593
--0.35957 --0.14382 +0.05868 +0.17768 +0.21768


planeNormal x: -0.00 y: 0.00 z: 1.00


Bed Level Correction Matrix:
1.00 0.00 0.00
0.00 1.00 -0.00
-0.00 0.00 1.00

[Roxy]

It is easier to see the imperfections in the 'flatness' of the bed when you have the bed closer to level. Here is the topography for a print I'm doing right now:

Mean of sampled points: 4.498123

Bed Height Topography:
--0.10869 +0.02526 +0.07123 --0.07925
--0.13349 +0.01203 +0.03584 +0.05469
--0.14705 --0.00054 +0.06495 +0.10530
--0.12225 +0.00277 +0.12745 +0.09174

Notice the first row starts negative, goes positive, and then goes back negative. And notice the second column. It starts positive, goes negative and comes back to positive.

Right now, your bed is tilted pretty hard from left to right. If you can get it more level, the imperfections will jump out at you. But with that said, looking down your 3rd, 4th and 5th column, the 4th row jumps up in value and then back down when it goes to the 5th row. But the 1st and 2nd column is the opposite. On the 4th row it jump down in value and then back up.

[AbuMaia]

I'm guessing this is pretty good, then?

>>>M48 x 100 y 100 n 15
SENDING:M48 x 100 y 100 n 15
M48 Z-Probe Repeatability test. Version 1.85
Mean: 10.933267
Standard Deviation: 0.000403
echo:endstops hit: Z:10.93

I'll try again with the E included.

Edit:
>>>M48 x 100 y 100 n 15 e
SENDING:M48 x 100 y 100 n 15 e
M48 Z-Probe Repeatability test. Version 1.85
Mean: 10.933835
Standard Deviation: 0.006625
echo:endstops hit: Z:10.94

[Roxy]

Yes... That first number is so good I doubt you can get it that good every time you run the test! Your mechanics must be very precise!

[AbuMaia]

Just printed out a new Z probe arm that's shorter and sturdier than my old one. http://www.thingiverse.com/thing:201406

I adjusted the parameters so that it places the switch just 1.5mm below the nozzle.

SENDING:M48 n 20 x 99 y 99
M48 Z-Probe Repeatability test. Version 2.00
Full support at: http://3dprintboard.com/forum.php
Mean: 1.497163
Standard Deviation: 0.001727
echo:endstops hit: Z:1.50

SENDING:M48 n20 x99 y99 E
M48 Z-Probe Repeatability test. Version 2.00
Full support at: http://3dprintboard.com/forum.php
Mean: 1.509900
Standard Deviation: 0.003012
echo:endstops hit: Z:1.51

I'm quite happy with these numbers.

[sniffle]

@roxy

didn't you manage to get this merged into the main branch of marlin?

I vaguely remember seeing this get merged