Suggestion for a simpler, less error prone, way of bed levelling

deckingman

@dc42 said in Suggestion for a simpler, less error prone, way of bed levelling:

Do you find the present scheme unsatisfactory, and if so, in what way?

Not at all. As I said in my OP, it matters not one jot to me because I don't have a need to use it.

Your title suggests that the current method is error-prone: in what way?

I was going to bugger off but seeing as you asked, a quick gander at these forums will reveal a number of threads related to bed levelling in it's various guises where people seem to have misinterpreted the instructions or misunderstood how it works. Another quick gander at M671 shows a list of pre-requisites that people might get wrong. For example the number and position of lead screw/attachment points/adjustment screw points. All of which have to be with respect to an origin which people might get wrong. As long as one probes in a line parallel with a line passing through the attachment points (and why would one do otherwise?) then none of this matters - the only parameter that matters is the distance between attachment points in order to calculate the length of the line and therefore it's slope.

I would also suggest that, the recommendation to probe near the lead screw positions is incorrect. For reasons that I detailed in this video tat I made over a year ago, https://www.youtube.com/watch?v=H9O1r46Izn8 with a 3 point system, a slope in the Y direction will affect the readings taken in the X direction. Whereas probing the X axis along a line that runs through the centre of the Y axis will negate that error. Likewise probing along the Y axis should be done on a line on the centre of the X axis.

That video also explains how one can end up with a level bed, but it's not necessarily level with the XY gantry because there might be a some twist or misalignment of the frame. You could, if you felt like, incorporate a check whereby after levelling the bed across the centre, you then check all 4 corners. This will show up any misalignment although it could be due to either a twisted build plate, or (more likely) a misaligned or twisted frame. But I appreciate that most people will choose to use mesh compensation rather than correct the mechanical error.

But hey, it was just a suggestion - wish I hadn't wasted my time..........

appjaws

@deckingman you have not wasted your time, I am sure a large number of users will have appreciated your detailed explanation.

mrehorstdmd

@fcwilt Mesh compensation can correct if the bed plate/print surface bows when heated, and/or the Y axis guide rails are not coplanar, and/or the print surface is not uniform thickness.

I ensured coplanar Y axis rails by mounting them on tooling plate (two pieces cut from one piece, green in the image below) that is in turn mounted on a rectangular frame of 4040 t-slot.

alt text

I use 0.7 mm PEI with a layer of adhesive for the print surface. I have no idea what the thickness specs are for either the PEI or the adhesive, but it has been good enough in the 4 printers I have used it on.

A Former User

This post is deleted!

oliof

@deckingman nice idea! I agree that people get too many of the moving parts of bed leveling wrong (like order of probe points having to match up order of reference points, which can cause any number of weird issues, or forgetting to probe Z after tramming the bed, etc.). Even the fact that you need a G30 to establish Z=0 instead of a standard homing move for an end stop seems to cause issues again and again.

Could be a fun exercise to implement this in a macro and see how it goes. Shouldn't be more than 30 lines of code I think.

sinned6915

@oliof more often than not, people's brains shut off and expect that using a probe and chunk of toolplate automatically gets them NASA/JPL accuracy and precision.

deckingman

@oliof said in Suggestion for a simpler, less error prone, way of bed levelling:

...........................
Could be a fun exercise to implement this in a macro and see how it goes. Shouldn't be more than 30 lines of code I think.

That's the best idea I've heard all day. I'd do it myself but as I've explained numerous times, I don't have a need myself for any type of firmware assisted levelling as I prefer the "mechanical approach" whereby my bed is flat, level and stays that way. I only have to check it if I do any "major surgery" on the bed (which almost never). If you do decide to write a macro, let me know how you get on.

dc42

@deckingman said in Suggestion for a simpler, less error prone, way of bed levelling:

I would also suggest that, the recommendation to probe near the lead screw positions is incorrect. For reasons that I detailed in this video tat I made over a year ago, https://www.youtube.com/watch?v=H9O1r46Izn8 with a 3 point system, a slope in the Y direction will affect the readings taken in the X direction. Whereas probing the X axis along a line that runs through the centre of the Y axis will negate that error. Likewise probing along the Y axis should be done on a line on the centre of the X axis.

Perhaps it would be better to probe 4 points, at the mid point of each edge of the reachable area of the bed?

deckingman

@dc42 said in Suggestion for a simpler, less error prone, way of bed levelling:

Perhaps it would be better to probe 4 points, at the mid point of each edge of the reachable area of the bed?

Exactly so.