In any case, nothing will change, everything will work correctly, Even if you do G28,
In the end, everything will be calculated correctly.
In order to understand whether the table map from memory matches the physical glass.
Previously, the mechanism would reset all the settings, now it does not reset and corrects, after all, this is what these mechanisms were made for :))
Here I am now starting a reprint ( Rare Case ).
I made homing, now I check where my table “lives” SCREWS_TILT_CALCULATE,
Here is what it gives:
rear left screw : x=43.0, y=235.0, z=-0.01792 : adjust CCW 00:02
rear right screw : x=303.0, y=235.0, z=0.02250 : adjust CW 00:02
front right screw : x=303.0, y=10.0, z=0.00708 : adjust CW 00:01
front left screw : x=43.0, y=10.0, z=0.02750 : adjust CW 00:02
I think that with such a deviation the table map from memory will match the real glass well.
What do you think? 
I can’t set the table more precisely, because bltouch gives a spread of readings of 0.05, which is very disappointing, but there is no way out, “1-2 minutes” can be considered an “ideal table”
As Sineos described earlier, the first “Screw” is the Base. I find it very useful by defining the first screw at bed center. using the same coords as the z home “virtual endstop”. This sets the “base” to the home (incl. z_offset) height, and allows the app to show recommended turns for each corner based on that height, rather than to an arbitrary corner.
This a recent run on my Sovol SV03 with a 350x350 glass bed.
[screws_tilt_adjust]
screw1: 208.5, 184.5
screw1_name: center (base)
screw2: 65.5, 44.5
screw2_name: front left screw
screw3: 65.5,335.5
screw3_name: rear left screw
screw4: 355.5,335.5
screw4_name: rear right screw
screw5: 355.5,44.5
screw5_name: front right screw
horizontal_move_z: 10.0
speed: 60.0
screw_thread: CW-M4
I also use a bltouch and get very good results using 5 probe samples and median, to get very good sampling accuracy. It takes a bit longer overall but I find the benifit worth it.
to show bltouch accuracy this pic is of a recent z_tilt probing run on a new project of mine, a Cylindrical Y axis, or.. Rotary Print Bed.
1 Like
Hello 
!
Immediately the first clarification - do you use ** “Standard Use Case?” **
The five -point system works only if your table surface is even even to count it “perfectly even” (just your case 
). Because on one plane is an infinite number of points.
You can add 10 more points to your 5 points to [screWs_tilt_adjust], and all of them will be in the same plane … (it already smells of surface map :))
Therefore, I do not understand what you do not understand, and what exactly did you want to say? - What is “everything works well in the old way”? 
Now imagine that you have a table in the form of a “waves” or “steel basin”, how are you going to pull the “end of the basin” to its “center? 
Therefore, I would not recommend the 5 -point scheme of work [screws_tilt_adjust]. This only complicates the work of the tool. The table card will do this business!
5 points do not make sense and are counterproductive in essence. (except for smooth tables, like yours).
The point of “house” can exist anywhere in the table, it is impossible to equate the “crooked tables”.
Friends, I just discovered that I didn’t attach the corrected checked files in the first post. 
Here they are: *.py and *.diff 
fix_screws_tilt_adjust.zip (2.8 KB)
And no one asked: “Dude, where are the files?” 
I didn’t notice before, but now I’ve discovered that they’re not there. I’ll send my friends a link, they’ll want to try it, but there’s nothing here. 
1 Like
@SlinkySlonkyWoffle,
Hi , look here, maybe this is the solution to your problem