I’ll admit, from the picture, I still don’t quite understand how this is working. But hope it can be figured out soon enough for my Z stage, seeing as auto bed leveling with software method isn’t compatible with a quad extruder.
I suspect the bed is pushed upwards, the nozzle pushes down to the level it wants, and the disk rotates around, locking each point in place
here’s his video:
Where to source that kind of gimbal pin is the question now. I’m still looking …
This is a very cool solution, but doesn’t appear practical for bots with moving beds. Maybe just applying the idea of locking the posts in place manually would work? Just sleeves and grub screws might be sufficient.
@John_Davis The Kühling&Kühling machine already works that way, with a system they call “semi-automatic bed leveling”. https://www.youtube.com/watch?v=8hDSFuC7HkQ
This is the same technique, but with a motorized cam lock instead of the manual setscrews they use. A hobby servo would seem like the ideal actuator for this type of system.
+scottbee
12:17 PM (5 hours ago)
Thanks!
I don’t haven’t formulated any long-term plans for the design. Right now I am concentrating on building and testing them for my printer farm. I have also release a unit “out into the wild” for external testing and perhaps some refinement of the software implementation.
Of course, virtually any system will deflect when you apply a load to it, and the R2X is no exception. When you use the nozzle to press down on the build surface the gantry will deflect “up” and you’ll also get some elastic deflection of the Z-stage springboard. This can be reduced through the proper selection of the “levitation springs”, but you’ll never eliminate the “error”. The good news is that the deflections have been very repeatable on my machines, and therefore you can easily characterize them out via the gcode. Although I was concerned about this in the beginning, it has mostly proven to be a non-issue.
The “gimbaling pins” is only important if you want to talk theory. In practice the M3 studs and stainless HBP baseplate allow for enough angular deflection to eliminate the need for some true “gimbal” mechanism. The pin clamps (printed in ABS) have a fairly small contact patch on the pins (which are just SST tubes installed over the M3 studs) and they allow a bit of angular deflection also. I was worried about a shimmy-bind… but haven’t really seen that either.
Scottbee
5:27 PM (28 minutes ago)
Randy,
I bugged-out of the office early today to ensure that my good Uncle Sam and I stay on speaking terms. I will have access to my library of images and 3D CAD models when I get back to the office tomorrow.
The mechanism is probably a lot less complex than you think. The three colored ABS “lever arms” are actually the clamps. They ride on a cam profile on the underside of the worm gear (that creates lateral movement)… and they have pivots and “living hinges” close to the centerlines of the three leveling pins (somewhat hidden by the MBI thumbnuts). So there is a lot of mechanical advantage. The actual clamping is done with two half-moon jaws printed right into the arms. They’re one-piece units.
The “gimbaling pins” are just stainless tubes that are slid over the factory M3 leveling studs, and the tubes are locked in place by the MBI thumbnuts. The thumbnuts are no longer used for any kind of adjustment. The half-moon jaws on the ABS arms ride on the OD of the stainless tubes, and when the cam moves the lever the jaws squeeze the tube, creating the “locking” friction. It’s not a “hard lock”, and if a nozzle crashes into the build surface the plate will move before the nozzle becomes damaged.
One of the advantages of using the stainless tubes as “liners” around the M3 studs is that there is an annular gap between the stud and the ID of the tube. This can be used to compensate for center-to-center variations between the M3 studs and the corresponding clamp locations in the span plate. You allow the thing to find its own “happy place”, then lock the tubes into place with the MBI thumbnuts.
I would love for the DC motor and cam to have feedback to the controller (or have the cam motor be a stepper), but the architecture of the existing MightyBoard just doesn’t support that (on dual extruder machines). So, instead, there is a microswitch under the worm gear/cam that you can’t see. After the cam has rotated about 320 degrees the microswitch will kick in and drive the cam to the 0/360 position. It’s a “poor man’s auto-homing”, and it helps to ensure that the cam starts from the same position every time that you initiate the tramming routine. It doesn’t artificially create accuracy, but it does help to eliminate cumulative errors.
- show quoted text -
I just put auto leveling on my printer a few days ago, the only problem I see with the z-probe solution is extra wear and tear on the threaded rods but it works great so far. This solution looks like it may need some hacking to work on other printers but cool idea.
You can also use 3 servos to lock the bed
@William_Frick says it’s just stainless tubes around the m3 bed screws with the old leveling nuts for end caps. The locks are printed and he claims that in the event of head crash in the z they will give before damaging the hot end
@Nuker_Bot_NukerBot_3 true but that increases cost and complexity. This is one motor and the locks and the gear triggers a unlock mode and when it passes on it relocks. Simplicity in itself.
I’m not sure I like dragging my nozzle across the print surface that much. If the build plate is prepped already with PVA or ABS juice, painters tape etc… Some kind of roller point probe that retracts instead of the hotend. Again it adds more ‘stuff’ !
I’m not sure you’d have to drag the nozzle across the bed. Couldn’t you lift, move to new location, then lower to Z=0 again?
This system relies on the head ‘tramming’ and pushing against the bed springs to ‘level’ it.
@William_Frick Yes, but the point is that you don’t drag the nozzle between those points, you lift it to travel between them and then press straight down on top of each spring.
Something to try for sure on my next printer. The travelling bed of a Mendel90 isn’t suitable for this approach though.