Hot Bed Leveling and 3-Point Mounting

by MakeItMakeItMakeIt Feb 9, 2019
Download All Files

Thing Apps Enabled

Please Login to Comment

Three points leveling is an obvious thing, just basic geometry and it always surprised me it is not standard. But you need a flat bed to start with. May be some mfg use 4 points in order to bend and correct somewhat the bed. This is cheap and unprofessional.
Same with auto bed leveling. Useless on a properly built Cartesian machine (but a must for DELTA for ex.)

Note that you didn't account for thermal expansion, can be a few tenth from ambient to 80° (Aluminum coef exp 23E-6 /°C)
You need a kinematic mount, many example can be found in the net.

Definitely agree with you here. I've currently got my Ender 3 at 4-point but had it at 3 point once before and the results were wonderful. I've always wondered why this isn't industry standard, though it does make sense that four points could be used to correct the bed to a certain extent. Still, on higher end machines, it's puzzling as to why this is not implemented.

In the machine shop that I retired from our vertical lathes had 4 jaw chucks but we used jacks at three points to level parts, works very well quick and easy.

Sure, 4 jaws for part retention. In that instance, I agree completely.

However, your Hot Bed is a plane, and a plane only needs 3 points, as with your jacks.

I do check my Hot Bed periodically.

That being said, it's still level at 3+ months and 300+ parts. :-)

Maybe I did not make my point clear the jaws just hold the part the three points are for leveling for years they use the 4 jaws and shims to level a part. We came up with the idea to use 3 jacks for leveling.
We first level the part with the jacks then tighten the jaws to center and hold the part.

Three point leveling makes the job faster and easier than using four points

Hmmm can I post an unbiased opinion... yes I can. So I spent so time researching bed leveling and what not prior to this document existing (yea i know when it was created opposed to when it was post, I have insider info).

First let me say this is a well thought out and complete guide, probably with more info than some will need or understand (should have no issue understanding it though IMO). I have employed this method for about the last 3 weeks and I have leveled my bed once, but only because I added new thumbwheels to my bed (on 2/8/2019).

Prior to the 9th of this month I had not releveled the bed at all, even after a thumbwheel lost its hold on a nyloc nut I had installed. (see my dates and the date of the post don't line up.... think about it). So as of 2/8/19 I have not touched the bed to level it again. I have printed about 10-15 prints and everytime the first layer goes down perfectly.

When done properly you will be able to start a print and walk away from the printer before it starts and come back a few layers later to observe, and at which point they print should go flawlessly. No more printing, waiting for the first layer to go down, and or a few more layers afterwards. YOU CAN HAVE A PRINTER THAT DOES WHAT IT IS SUPPOSED TO BE AND THATS PRINT PROPERLY AND EFFECTIVELY.

No was my journey to level perfection easy.... by no means was it at all. I have spent hours aligning, calibrating, checking every bolt, screw, belt, rod, carriage endstop etc. My biggest issue was an endstop that was not triggering properly. The metal clip on the stop was lose changing the z home position everytime it homed, that took me 3 weeks to realize (never again will that happen just removed the clip and moved on).

The process to ensure your printer is capable and able can be daunting at first. But so is building a car, you get a car and expect it to work without fail. But printers are machines as well, but a machine you built (at least you should have to better understand how it is put together and should be put together). Once your printer is square x to y, front to back, and left to right, and why not top to bottom for the z rods smooth and threaded. basically look into the guide and follow it all.

Now all I do is level the x axis (I REFUSE TO CALL IT Z CALIBRATION, AS YOU ARE NOT CALIBRATING ANYTHING) I don't do it everyprint, but I do it every few prints. For larger prints I would definitely recommend doing an x level afterwards. The OP has spacers intended for this purpose, check them out. I have designed, just not uploaded, a spacer for the Anet A8 2018 version (has the v1.5 board).


IK IK IK its been way too long since I posted this. Finally had the time to design, redesign, redesign some more, print, redesign, print and so on so forth. But finally my anet a8 x axis levelers to go with this guide are down, tested(limitedly) and posted. CHeck them out here > https://www.thingiverse.com/thing:3554155


No worries Paul!

It's really ALL part of the topic.

I DO feel they ACED it there, but like so many other manufacturers, they kinda dropped the ball with 4 corner screws versus 3 triangled.
Yes, that puzzles me TOO!

Thanks Paul!

Hi Paul,

I believe you answered the question in and of itself.

Based on what I've researched, the Trigorilla board does use a parallel Z output (it's advertised in the illustrations on sites where you can buy it), but Anycubic is not using it in that manner, based on your report that the 2nd Z is empty and E1 is populated with the 2nd Z.

In THAT configuration, I completely agree and can see the how's and why's of your ability to utilize 2 separate Z end stops, 1 per motor.

I find that to be BRILLIANT, in that it ELIMINATES the need for manual X leveling for every print, as it self-levels the X axis at the start all by itself. Just BRILLIANT!

Your most recent offering about that clears up everything for me about your machine.

Your testing and procedure was never in question by me: I'd just never seen a machine configured in that manner before.

Paul, please don't mind me regarding YOUR printer, just keep printing. :-)

As you and I both said, there's ALWAYS something new to learn, and I consider myself to have been schooled! :-)


Ah, thank you! Now the penny has dropped! I am happy to accept that the two Z ports are commoned as you stated. That explains exactly why Anycubic elected to use E1 instead of the second Z. In so doing they achieve independence for left and right Z in the manner I described. Good thinking Anycubic. You've cleared that one up for me - no need for me to pursue that further. Consider me schooled too!!

Apologies for pulling this off topic. I do hope many take note of your pdf and start to think 3 is good, 4 is bad (and understand why that is)! Thanks once again for taking the time to put this all together.

Thank you PaulC-00 for your comments.

On your Anycubic, as far as I can tell, it uses a Trigorilla controller board, which employs a parallel Z driver circuit: both Z axes are driven at the same time by the same circuitry.
That indicates to me that separate Z end stops won't do anything special as they are not independent.

That would also indicate to me that it potentially could trigger one Z end stop and halt the motion before the 2nd Z end stop is triggered, resulting in an out of level X axis.

So I must politely disagree with you that dual Z motors are "fine," with respect to your machine.

Synching the Z's with a belt, or driving 2 thread screws with a single Z motor would be much more appropriate and accurate.

Removing one corner screw from your hot bed is a good temporary relief, but re-locating the 3rd screw midway from front to back brings equilateral support across the entire hot bed and makes for more accurate and retainable hot bed leveling.

And yes, pre-heating before X leveling is ALWAYS a good thing. :-)

Thank You, Paul.

I've just noticed that you amended your first comment. Consequently I feel my earlier response requires further clarification.
Yes, my Mega uses a Trigorilla board. No, it DOES NOT employ parallel Z driver circuitry.

I have just performed the following test:

  1. Move Z axis to top of gantry.

  2. Selected 'home Z'.

  3. Both left and right motors run and begin to lower Z.

  4. I manually trip the left Z endstop. Left motor stops and beep is emitted. Right motor continues to lower Z axis. At no time does the right motor stop.

  5. I release left Z endstop. Left motor restarts and resumes lowering Z. Both left and right Z motors are lowing Z at this point.

  6. I manually trip the right Z endstop. Right motor stops and beep is emitted. Left motor continues to lower Z axis. At no time does the left motor stop.

  7. I release right Z endstop. Right motor restarts and resumes lowering Z. Both left and right Z motors are lowing Z at this point.

  8. Once homed, two beeps are heard (as I have obviously put them out of sync). Pressing Z home again results in one beep as they are automatically back in sync.

I don't see as there can be any doubt that my earlier assertion is correct. Left and right are indeed independent, self correcting and (assuming set up accurately) self levelling.

I think the issue here is not all Anycubic i3 Mega printers work in the same way. It follows that information regarding the Mega should not be relied upon unless it can be ascertained which 'variant' is being discussed. On Anycubic's site you can view instructional videos. However, some are out of date! They have a video showing how to set Z - by manually turning one of the lead screws!!!! That only works for common driver versions. Mine would override/correct such adjustment.

I think circuit changes were made when the 'Ultrabase' build platform was introduced. They had to remove the auto levelling feature because it could not work through glass. I suspect that the duel independent Z feature was introduced around this time as some sort of
compensation but, to be honest, that's nothing more than a guess on my part.

Finally, just to muddy the waters further, Anycubic appear to have wired it rather oddly (I haven't checked yet but I will). The (my) Trigorilla has a total of 6 stepper drivers on board. They are Y, X, Z, Z, E0 and E1. The second of the two Z connects is not populated but E1 is!!! I don't have a second feeder. It would appear that (for reasons best known to them) they elected not to use the second Z driver for the second Z motor but use E1 instead. I will investigate further when I next take it apart.

Hopefully, this goes some way to verify how my Mega is constructed but, should you consider my test outlined above is incorrect, inconclusive or just downright wrong please tell me - I am more than willing to be corrected. That's how we learn after all and I'm always happy to learn.

Yes, Z motors are independent. Each has its own endstop and independent 'beep' when homed. If I deliberately set them out of sync
I hear two beeps when homed. If I immediately perform a second 'home' I then hear only one beep. They are back in sync. Of course
it's only as good as the effort you make to precisely set the micro adjusters. If they are set close to correct you would hear a single tone but likely to be two overlapping beeps.

I'm not suggesting one method is better than another. They both have failings of course. I totally agree with you that a single Z driver controlling two motors is poor and should be avoided if at all possible.

Strictly speaking, I believe that the position of the 3 point geometry shouldn't make any difference but that would be in an ideal world. Locating as you suggest would minimise error due to warping and I do want to position the single side to be equilateral for that reason. However, I am a little reluctant to attempt that at the moment (the glass base is stuck fast to the heat bed and they need to be separated to relocate the screw). I will attempt it after I have studied it further. However, it is working so well (I don't even bother to check for level now - it's always correct) that the old adage 'if it ain't broke, don't fix it' springs to mind!

I DO find that to be very interesting, in that the board is advertised as having "parallel" Z ports, which indicates to me that they are "clones" of each other, and not separate/independent.
Learn something new every day!

And yes, I am in complete agreement that "if it ain't broke, don't fix it," with regard to your hot bed setup.
Having to PRY something apart is never a good thing if it doesn't have to be.
Maybe you'll go that route down the road after some investigation.

Thanks Paul.

I've not read this fully yet but, essentially, I know what it says!
Well done.
I am new to 3D printing but, having just got an Anycubic i3 Mega (which I struggled to get level), the penny quickly dropped and I removed the rear left adjuster and spring and left that corner floating. It's dead easy to get level now!
(The mega has a 3 bearing carriage plate and not 4 so very well done to Anycubic for knowing enough to do that correctly but why did they then fit four corner levelling screws? I simply don't know. How many people suffer hot bed and build plate delamination as a result I wonder. I've even seen on the internet that there are some that intend to fit a fourth linear bearing to the Y rail!! Oh dear.)
It's probably worth mentioning that for this to work correctly your build plate must be perfectly flat and the carriage plate (while not needing to be perfectly flat) needs to be rigid. Y rods need to be true and aligned correctly of course.
While I don't disagree with your comments re the Z axis per se, there is absolutely nothing wrong with a dual motor configuration.
The Mega (Ultrabase model) employs two motors together with dual micro adjustable endstops. If set precisely, Z is correctly calibrated whenever 'Home' is actioned. Set it once, no need to touch it again.

Almost forgot, if you run with the bed hot, you need to pre heat before calibration.
Again, thanks for this.