Level Bed Graph

I took apart my Y axis and bed to try and improve how level the bed it by default
I measured all spacers, and washers to try and mix and match them so that its as close to even as i could get.
not using any springs.. all spacers all around. so the bed sits solid.
Finally got it reassembled.

Did a before and after measurement using a Dial caliper to measure bed distance from X carriage in 20 spots
I compared this to after I did a ABL Routine

These below charts are exagerated to make it clear

White is Flat
Red is default bed with no auto levelling
Green is after Auto Levelling

On the red the deepest valley was -.52mm and the highest hill was .92mm
the middle of the bed was 0 on all three. also the very left side was not measured thats why it all comes together

Not sure what is causing it to raise up in the back right corner.

http://3dprintboard.com/attachment.p...tid=5421&stc=1http://3dprintboard.com/attachment.p...tid=5422&stc=1http://3dprintboard.com/attachment.p...tid=5423&stc=1

also I have ABL set to probe 3 points per X &Y so it checks 9 places.

my i3v 12in y bed is heat warped with out question i can see it in plan sight
and v wheels are ok at best i think i like LM8UUs or igus dry glides more
less of a pain

note over tension on v wheels will stretch the bed lifting the right side so keep an eye on that wheels to tight on the 12in will mess with bed leveling

so what can I do to make ABL more accurate? so the bed is very level. it does well but I want it to do better
im working on switching to Geared Z Motors. and I already have lead screws

I found this thread, its very interesting, I hope something like this is done
https://github.com/MarlinFirmware/Marlin/pull/1619

Its in already Adam, but they are working out the printing in air issue right now so I would wait but you can use m851 to adjust your offset so it will print anyway

but its not part of the normal ABL routine?
M851 just offsets the Z probe

I think you miss my meaning... The mesh abl has been added to the main trunk of marlin. I don't flash it daily but I do keep up with it. Right now there is a bug that has been introduced with the raise after probing but before retracting code where the zoffset isn't correct... They are working that out ATM. You can still print by dynamically adjusting your offset using m851 without having to reflash your firmware. Most that have done this are having to make their offset between 2.3 and 2.7 mm more negative.

the variations from flatness and levelness you're talking about are huge - did you leave out a zero to make the graph easier to see? from what you're saying I'm guessing you're measuring the plywood y bed support piece? I would suggest that that is a mechanical part to help set levelness and receive the y motion input. the flatness should come from a build plate you put on top of that. plywood will never be flat but it really doesn't have to be. Agree with Techmasterjoe the 1/4" ply is not stiff enough to resist bending if the wheels are as tight as you can make them with the eccentrics.

If this is a real issue you could consider one of the aluminum bed plates that's out there.

graph was exagerated
deepest valley was -.52mm and the highest hill was .92mm
so a max of 1.5mm range

anyone know where i can get an alum bed for a i3v 12"