mesh auto leveling

Code hacking talk

mesh auto leveling

Postby zfrenchy » 2016-Oct-Thu-17-Oct

Hello,
I have a play with Y upgrade, now the printing surface is 98x209 (X/Y)

The auto leveling at the beginning of the print is only three point, Xmin/Ymin, Xmin/Ymax, Xmax/Ymax.

I think it is not enough for this long build surface, can I change something to pin point more than three ?
  • 0

User avatar
zfrenchy
Waiting to connect to printer...
 
Posts: 5
Joined: 2016-Oct-Tue-19-Oct
Reputation: 0

mesh auto leveling

Sponsor

Sponsor
 

Re: mesh auto leveling

Postby RetireeJay » 2016-Oct-Thu-19-Oct

The firmware that handles compensation for bed alignment can ONLY compensate for a FLAT (i.e. PLANAR) surface that is tilted with respect to the plane of the X, Y axes. Therefore, there is no possibility of compensating for a bed that is warped or in any way non-planar. And in geometry class, we learn that "three points determine a plane."

There have been extensive debates in this forum about the utility of using more than 3 points. The best that can be said for using more points is that you can calculate a better "fit" to the actual bed. But if the bed is truly flat, then 3 points is sufficient. And if the bed is not flat, then you will still have some problems with the nozzle being either too close or too far away from the bed on the first layer.

If you really want to do it, yes, there are firmware packages out there that will probe at 9 points.
  • 0

Printrbot Plus operational January 2013
Brass threaded rods (5/16" X 18) & nuts for Z axis
GT2 belts & pulleys
Cable chain to reduce probability of fatigue failure in wires
E3D V5 Hot End, 0.4mm nozzle, also 0.8 and 0.25 in use occasionally
PB fan mount + 40mm fan -- using printed mount adapter, not the E3D supplied fan
Injection molded extruder gears
Optical Z "endstop" (custom designed and built)
Have used many pounds of T-Glase filament. Now also doing some work with Ninjaflex SemiFlex
Print on glass with Scotch Craft Stick or other glue stick
User avatar
RetireeJay
My next printer is...
 
Posts: 4797
Joined: 2013-Jan-Wed-13-Jan
Location: Greenville, SC
Reputation: 477

Re: mesh auto leveling

Postby Mooselake » 2016-Oct-Thu-20-Oct

To echo RJ, if your bed is so far from being aligned properly then you need to fix the actual problem. 9 or 900 point leveling won't make the problem go away. Extra probing is best left to aliens and not becoming a time waster on your printer. Auto-leveling is really a way to cut down on production costs by somewhat compensating for improperly adjusted (or designed) printers.

Leveling is not a very good name for what we need to do, and what the rest of the CNC world calls tramming. You are trying to make Z0 be the same distance from the bed over the entire surface (you want a tiny gap, not actually touching), and it doesn't matter what it's relationship to the surface of the earth (or your resident planet) is. You can tilt it 45 or 90 (or even turn it upside down, some non-printrbot's are demoed upside down) and it's still "level" for 3D printing purposes.

To answer your original question, no you don't need extra probe points for an expanded bed, especially something as small as a Play, expanded or otherwise. I do 1 point probing on my 200mm square plywood plus, with the same type of microswitch as the other axes.

Kirk
  • 0

Modified KickStarter Classic Plus 7/2012
KS Thingybot Delta Pro 10/31/16
User avatar
Mooselake
My next printer is...
 
Posts: 3509
Joined: 2011-Dec-Tue-23-Dec
Location: Moose Swamp, Upper Michigan, USA
Reputation: 170

Re: mesh auto leveling

Postby gz1 » 2016-Dec-Sat-23-Dec

I'm sorry guys, but these responses are just not helpful. It's sad because it pops up in the search results for people looking for answers.

Yes, sometimes the bed is not perfectly flat. From factory. Or it gets damaged in some way. Not in a major way, but in a way that skews the amount of stick for the 1st layer by just a little bit too much. Who cares what "the rest of the world calls it"?

You can glibly condescend and say, "You need to fix the actual problem" and roll your eyes, but the fact is for an otherwise perfectly functioning printer, it's far easier to just have the software compensate for slight bed imperfections than it is to go through whatever non-articulated procedure you had in mind for fixing the bed when you rolled your eyes.

My Printrbot Play has a minor dip in the print surface. It's not a major one, but it's clear it's bowed in near the middle-right side, exactly where the 3-point probe *doesn't* look.

How should I "fix the actual problem" in this case?

Because it seems to me the simplest fix would be to account for it in software.
  • 0

gz1
Waiting to connect to printer...
 
Posts: 1
Joined: 2016-Dec-Sat-23-Dec
Reputation: 0

Re: mesh auto leveling

Postby RetireeJay » 2016-Dec-Sun-06-Dec

I'm sorry if Mooselake and I appear to you to be "rolling our eyes" but the simple fact is that no existing software that we know of can compensate for a bed with a dip in it, regardless of how many points you probe.

We could have mentioned that the problem of non-planar beds has been around for a long time, and there is another solution besides fixing the bed mechanically. It's called "raft" and it's a function of the slicer (not Marlin). Basically the slicer will create a virtual plastic bed that is flat on top of your warped or non-planar bed. In other words, there will be a layer of plastic that you peel away from your model and discard, but the top of that extra layer of plastic will be parallel to the X-Y plane. If it sticks well enough to the bed.

By the way, modern sheet glass (such as is used in windows and mirrors) is extremely flat. Both of us are using glass as the top surface on our beds. If you don't apply too much force to the glass where you clamp it to the underlying structure, it is planar within microns. Then if you prefer to compensate for mis-alignment with the X-Y plane and do your leveling in software instead of hardware, that's up to you. But three points for leveling is all you need.

If you have a planar bed, then using more than three points for "leveling" (actually, "tramming") with the existing software has diminishing returns because in the end, the algorithm in Marlin is only going to rotate the model to be perpendicular to the plane of the bed. In fact, besides printing a "raft" it is theoretically impossible to perfectly compensate for a bed with a dip in it. Would the printed model carry this dip all the way to the top of the model? How would you eliminate it?

However, if you prefer a software solution, you are perfectly welcome to go ahead and write your own software that compensates for a bed with a dip. No one else has done it yet.
  • 1

Printrbot Plus operational January 2013
Brass threaded rods (5/16" X 18) & nuts for Z axis
GT2 belts & pulleys
Cable chain to reduce probability of fatigue failure in wires
E3D V5 Hot End, 0.4mm nozzle, also 0.8 and 0.25 in use occasionally
PB fan mount + 40mm fan -- using printed mount adapter, not the E3D supplied fan
Injection molded extruder gears
Optical Z "endstop" (custom designed and built)
Have used many pounds of T-Glase filament. Now also doing some work with Ninjaflex SemiFlex
Print on glass with Scotch Craft Stick or other glue stick
User avatar
RetireeJay
My next printer is...
 
Posts: 4797
Joined: 2013-Jan-Wed-13-Jan
Location: Greenville, SC
Reputation: 477

Re: mesh auto leveling

Postby Mooselake » 2016-Dec-Sun-09-Dec

I remember that somebody has actually done firmware (or maybe slicing software) to compensate for a warped bed, but I don't remember who or how. The most likely way would be to vary the thickness of the first layer, and you still get an object that won't sit flat.

You could take an aluminum bed to a local machine shop and have it milled or ground flat. Better yet, use that as an excuse to buy a Bridgeport or surface grinder. Or perhaps get a piece of thicker flat glass (or a surface plate), an assortment of wet/dry sandpaper, and flatten it yourself. It's common for over clockers to flatten CPU heatsinks (better conductivity so lower temps and higher speeds), so online directions abound.

The simplest fix is a warranty replacement of the defective bed.

Kirk
  • 0

Modified KickStarter Classic Plus 7/2012
KS Thingybot Delta Pro 10/31/16
User avatar
Mooselake
My next printer is...
 
Posts: 3509
Joined: 2011-Dec-Tue-23-Dec
Location: Moose Swamp, Upper Michigan, USA
Reputation: 170

Re: mesh auto leveling

Postby Mooselake » 2016-Dec-Sun-09-Dec

You care what the rest of the world calls it. The poor description has made more than one person use a spirit level to adjust their print surface. Redefining commonly used terms understandably confuses people, and there is already a reasonable term for the process in the rest of the CNC world

Kirk
  • 0

Modified KickStarter Classic Plus 7/2012
KS Thingybot Delta Pro 10/31/16
User avatar
Mooselake
My next printer is...
 
Posts: 3509
Joined: 2011-Dec-Tue-23-Dec
Location: Moose Swamp, Upper Michigan, USA
Reputation: 170

Re: mesh auto leveling

Postby ChefScott » 2016-Dec-Sun-10-Dec

gz1 wrote:....My Printrbot Play has a minor dip in the print surface. It's not a major one, but it's clear it's bowed in near the middle-right side, exactly where the 3-point probe *doesn't* look.

How should I "fix the actual problem" in this case?....


http://www.printrbottalk.com/forum/viewtopic.php?f=15&t=7161#p48525
  • 0

User avatar
ChefScott
Waiting for printbed temp...
 
Posts: 142
Joined: 2014-Apr-Fri-21-Apr
Reputation: 18

Re: mesh auto leveling

Postby RetireeJay » 2016-Dec-Sun-11-Dec

gz1 wrote:My Printrbot Play has a minor dip in the print surface. It's not a major one, but it's clear it's bowed in near the middle-right side, exactly where the 3-point probe *doesn't* look.

How should I "fix the actual problem" in this case?

Because it seems to me the simplest fix would be to account for it in software.

Upon reflection, I can say that there may be something else you can do, if the "minor dip" is minor enough. Basically, you would tweak your first-layer thickness to be thick -- maybe even as thick as your slicer will allow (normally the limit is the diameter of your nozzle) -- and then you would also tweak the Z zero offset so that the trace you lay down does not go to zero thickness at the highest point on your bed while still having some "squish" at the lowest point on your bed. Depending on the actual elevation map of your bed, you may find it beneficial to tweak the locations where the 3-point probing takes place so you get an overall "best fit" to the actual shape of your bed. I'll agree that in some cases, the 9-point probing might do a better job of finding the "best fit" plane to use with a warped bed. But even with that help, you'll still need to do the first-layer thickness and Z zero adjustments to get a good result, because getting good adhesion on the first layer is critical to building a good model. And if the warp is too severe, you'll never get really great adhesion everywhere.
  • 0

Printrbot Plus operational January 2013
Brass threaded rods (5/16" X 18) & nuts for Z axis
GT2 belts & pulleys
Cable chain to reduce probability of fatigue failure in wires
E3D V5 Hot End, 0.4mm nozzle, also 0.8 and 0.25 in use occasionally
PB fan mount + 40mm fan -- using printed mount adapter, not the E3D supplied fan
Injection molded extruder gears
Optical Z "endstop" (custom designed and built)
Have used many pounds of T-Glase filament. Now also doing some work with Ninjaflex SemiFlex
Print on glass with Scotch Craft Stick or other glue stick
User avatar
RetireeJay
My next printer is...
 
Posts: 4797
Joined: 2013-Jan-Wed-13-Jan
Location: Greenville, SC
Reputation: 477

Re: mesh auto leveling

Postby KD6HQ » 2016-Dec-Wed-14-Dec

For what it is worth:

The term that most folks use is "Auto Leveling", that really is misleading. What it really dose is to compensate
for a bed that in on an angle and not parallel to the horizon. The firmware does not compensate for a bed that
has dips in it. That is why I say it is misleading to call it "Leveling". If the bed is tilted say 5 degrees out of
being parallel to the horizon then it will compensate for that.

So it is left up to the owner of the printer to make the bed as flat as possible. In the case of aluminum this
can be done in a minimum of 2 ways. You can take the bed to a milling shop and have it milled flat.
Purchase a new bed that has already been milled or try flattening the aluminum your self by sanding it.
Hard work but it can be done.

I eventually changed to printing on glass as has been mentioned.
Printing on glass in my case met changing both the bed makeup and the "leveling sensor" that I was using.

I now use a heated bed, with a glass bed topped by a product called PEI. PEI is a material used in the
place of "blue painters tape" or other methods to help PLA stick to the bed. After cool down of the bed
it also provides a easy release of the object you have just printed. You can remove the object while
the bed is still hot but you stand a chance of warping the object if you do that. The bed I use is now very flat.
When I run the "auto leveling" routine I get difference of 0.00xxxx readings. Yes that is correct 0.00xxxx.
I have also changed the auto leveling routine to check 9 points during it's check.

You can check here

viewtopic.php?f=80&t=9667&start=100

to see the makeup of my current bed. You will note that the bottom layer is cardboard.
It is used as a thermal layer break and works very well. My heated bed is 8.4" x 12.6".
Using the combination that I show in the link the bed heats up to 60-70 C in 3-4 minutes from a cold start.
  • 0

Printrbot Simple Makers Kit (1405) modified.
Core XYZ

It's all in the details!
KD6HQ
Layer 300 of 1234
 
Posts: 329
Joined: 2014-Jun-Mon-12-Jun
Reputation: 5

Re: mesh auto leveling

Postby KD6HQ » 2016-Dec-Wed-15-Dec

For what it is worth:

The term that most folks use is "Auto Leveling", that really is misleading. What it really dose is to compensate
for a bed that in on an angle and not parallel to the horizon. The firmware does not compensate for a bed that
has dips in it. That is why I say it is misleading to call it "Leveling". If the bed is tilted say 5 degrees out of
being parallel to the horizon then it will compensate for that.

So it is left up to the owner of the printer to make the bed as flat as possible. In the case of aluminum this
can be done in a minimum of 2 ways. You can take the bed to a milling shop and have it milled flat.
Purchase a new bed that has already been milled or try flattening the aluminum your self by sanding it.
Hard work but it can be done.

I eventually changed to printing on glass as has been mentioned.
Printing on glass in my case met changing both the bed makeup and the "leveling sensor" that I was using.

I now use a heated bed, with a glass bed topped by a product called PEI. PEI is a material used in the
place of "blue painters tape" or other methods to help PLA stick to the bed. After cool down of the bed
it also provides a easy release of the object you have just printed. You can remove the object while
the bed is still hot but you stand a chance of warping the object if you do that. The bed I use is now very flat.
When I run the "auto leveling" routine I get difference of 0.00xxxx readings. Yes that is correct 0.00xxxx.
I have also changed the auto leveling routine to check 9 points during it's check.

You can check here

viewtopic.php?f=80&t=9667&start=100

to see the makeup of my current bed. You will note that the bottom layer is cardboard.
It is used as a thermal layer break and works very well. My heated bed is 8.4" x 12.6".
Using the combination that I show in the link the bed heats up to 60-70 C in 3-4 minutes from a cold start.
  • 0

Printrbot Simple Makers Kit (1405) modified.
Core XYZ

It's all in the details!
KD6HQ
Layer 300 of 1234
 
Posts: 329
Joined: 2014-Jun-Mon-12-Jun
Reputation: 5


Return to Hack talk

Who is online

Users browsing this forum: No registered users and 2 guests

cron