Auto_Bed_Leveling: Enhanced G29 command

[Roxy]

Hi I just tried again and its downloaded fine this time, weird. The new copy i've made seems to be working better as it compiles and uploads and now when I run a g29 it doesn't do the annoying servo retract every move

Yes, that is intentional. But you can get it to E-ngage and retract the probe for each measurement if you want with the 'E' option. G29 E will give you back the annoying behavior.

but at the end the nozzle stays above the bed and cant be moved down to touch it, it actually raises the nozzle a fraction and shows as being at 0. Im using the probe as a z endstop so do I need to change M206 offset?

I suspect the reason you can't lower the nozzle is you have:

#define min_software_endstops true

It is trying to protect you. The Z-Axis height is computed as a result of all the probing and some points are measured higher than others. Where ever the zero point gets calculated, that #define keeps you from going lower. I don't use the software end stops. But before you turn this off, you might want to consider and try changing this line:

#define Z_MIN_POSITION -5 // (or how much you dare to allow it to violate the 0.00 mark on the Z-Axis)

I don't know about or use the M206 stuff. I doubt you need to mess with it. Let me know if these two things don't fix things for you. It is easy enough to add a couple lines of code to push the nozzle down some amount.

The last code I tried using added the two together so if both had an offset it would double the distance.
The other thing is the n parameter doesn't seem to have any effect, it always does 3x3

The n parameter doesn't have any effect??? Try G 29 n 99 and see if you get an error message. If not, we have some serious problems. But if so, it means the code is recognizing you gave it a number for one axis of the grid. If that gives you an error message, then G29 n 4 should do a 4 x 4 grid. Or G29 n 2 should give you a 2 x 2 grid.

although it moves around in a different pattern.

The order of the points is dependent on whether it is going to print a Bed Height Topology graph. If you set the V-erbosness level high enough or specify a T option, you will trigger the alternate path. The alternate path takes slightly longer (1 or 2 seconds) to accomplish, but it reduces the code complexity and code size dramatically.

Ive copied the output from the bed level matrix and topology map for you if that is of any help? Might be best to just leave the old copy alone as its a bit of a mash-up of different code. Sorry if i've wasted any of your time.

22:07:11.001 : Bed Level Correction Matrix:
22:07:11.009 : 0.999997 0.000000 0.002244
22:07:11.009 : 0.000000 1.000000 0.001029
22:07:11.009 : -0.002244 -0.001029 0.999997
22:07:11.312 : echo:endstops hit: Z:-0.40
22:08:14.532 : N180 G28 X0 *114
22:08:15.940 : N182 G28 Y0 *113
22:08:20.974 : N184 G28 Z0 *116
22:08:34.858 : N190 G29T0 *94
22:09:58.962 : Eqn coefficients: a: 0.00 b: 0.00 d: -1.02
22:09:58.962 :
22:09:58.966 : Bed Height Topography:
22:09:58.966 : --0.08199 +0.11606 +0.31254
22:09:58.970 : --0.27027 --0.05777 +0.11606
22:09:58.970 : --0.14293 --0.02691 +0.03520
22:09:58.970 :
22:09:58.974 : planeNormal x: -0.00 y: -0.00 z: 1.00
22:09:58.974 :
22:09:58.974 : Bed Level Correction Matrix:
22:09:58.978 : 0.999997 0.000000 0.002283
22:09:58.978 : 0.000000 1.000000 0.001003
22:09:58.982 : -0.002283 -0.001003 0.999997
22:09:59.285 : echo:endstops hit: Z:-0.40
22:10:43.899 : N213 M81 *58

I don't now what all that extra Nxxx stuff is. My suggestion is to just work off of the one I posted. Without the same source code, it is difficult to guess what is going on. But you have .5mm of difference between your lower left and upper right corner. The Auto_Bed_Leveling will correct for that, but you really ought to get that leveled out better. That is hugely excessive!

p.s Is there anyway to have the nozzle lift before the servo retracts at the end of G28 and G29? A few other people wanted the same thing and i'd tried using some of their code which might have caused me these problems in the first place. Thank you again for all the help.

Yes, that is not too hard to do. And in fact RegPye wanted the same thing. Let's get the first part working right... And I'll get a couple lines of code added to handle this request.

[brainscan]

The Nxxx is just the check sum/line number Repetier host uses, it can just be toggled on and off. G29 n99 just went through a normal 3x3 grid, no error message. Tried changing MIN_Z_POSITION to -8 (slightly more than my probe z offset) seems to have made the problem worse, now rather than moving up a fraction and reading 0 it moves up over 8mm and reads 0!!!! What the hell have i done wrong? You're right though, get this problem sorted first before wishing for more.

As soon as I get home I will try switching the max software endstop to false and see if that helps. I must have done something wrong if the n parameter isn't working. I haven't tried the e parameter as that's the main thing I wanted to stop, v does as you said and t works as it should, seems strange it's only the n parameter that's being ignored, plus it's a new copy of marlin so scratching my head at the moment.

I haven't tried to level my bed manually since last removal so it made me chuckle that .5mm difference from corner to corner is hugely excessive. Reminded me of a joke about accuracy. Cabinet makers work to the closest 1/16 inch, carpenters the closest 1/8 inch and joiners work to the closest pub. In this case you're the cabinet maker and I just need a drink .
I will try and level it out more each time I run a test.

[Roxy]

Are those 3 .ZIP files up above what your are using? Because the extra (incorrect) code at the end of the G29 command is back in there????

As part of the communication protocol PronterFace uses capital N's and those get lost from the command line. I'm wondering if Repetier eats all n's ??? I looked at the code and it has to produce an error message if it gets the 'n 99'. Here is something to try:

At the start of the G29, you will see this code:

if ( code_seen('n') ) {
n_points = code_value();
if (n_points<2 || n_points>AUTO_BED_LEVELING_GRID_POINTS ) {
SERIAL_PROTOCOLPGM("?Number of probed points not plausable.\n");
break;
}
}

Change the:

if ( code_seen('n') ) {

to

if ( code_seen('u') ) {

and give it a G29 u 99 command. With that change, even if Repetier is eating all the n's, you will be able to specify more points (and have it produce an error message). You can also use PronterFace and see if that allows you to specify the number of grid points and that would keep you from having to make the suggested firmware change and reflashing the electronics. If that is what is happening... It will probably make sense to update the code so people using Repetier don't hit this problem in the future.

[brainscan]

The 3 zip files are the old firmware I couldn't get to compile (so not using at all now) you've already seen the marlin main and config from it, that's why the duff code is still in there.
I started with a brand new version as previously I had just been copying my configuration, pins and thermistor table files over so one or all of those may have been old/corrupt and no longer compatible. I thought if I follow your instruction as though it's a fresh build with fresh copy of marlin we'd at least know the starting point is the same, does that make sense?
Once I get home I will try pronterface first and see what happens, if it works I will try the change and see if that helps Repetier or not. Hopefully it will fix the other problems. If not I will zip my new version that I'm using for you.
Thank you again, seriously you don't know how much I appreciate it.

[sprocket1597]

As part of the communication protocol PronterFace uses capital N's and those get lost from the command line. I'm wondering if Repetier eats all n's ???

Repetier does seem to eat n's.. It changes "n" to "N" I couldnt figure out how to turn off check sum, but changing the code to use "u" does the trick.

[Roxy]

So is everything working?

[DarkAlchemist]

So is everything working?

Is it? I am unsure what the next step is supposed to be. It compiled and uploaded just fine.

[DarkAlchemist]

Repetier does seem to eat n's.. It changes "n" to "N" I couldnt figure out how to turn off check sum, but changing the code to use "u" does the trick.

*HIGH-FIVE* as I stuck a U/u there as well because Pronterface eats the n and N as well.

[Roxy]

*HIGH-FIVE* as I stuck a U/u there as well because Pronterface eats the n and N as well.

This is going to have to be changed. N is a nice symbol to use for the number of probe points, but it is causing too much pain.

[DarkAlchemist]

This is going to have to be changed. N is a nice symbol to use for the number of probe points, but it is causing too much pain.

I agree but I also find it alarming that these host programs are eating anything.