Bed Stability: 3-point leveling should be in fact 2-point leveling and a fixed one

Hi guys, I am sure somebody else already thought about it and implemented it so I am asking those guys to post their solutions in this thread I am starting.
My reasoning is that if one of the 3-points is fixed at a certain pre-determined height the other two points can be thumbscrews that adjust the plane relative to that pre-determined height.

This way the complexity of the bed assembly decreases a bit and the adjusting can only be done from the front instead of reaching to the back of the printer where the third thumbwheel would be positioned.

Edited 1 time(s). Last edit at 06/01/2016 07:00AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Yes, that's the usual way to do it. However i would only use a three point system if the build plate is either not moving (Delta) or only moving along z. For Y movement i would always use four points.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Realthor is 100% correct. There may be a parallel universe somewhere where the rules of geometry are different, but in our universe, 3 points define a plane and 4 points define a saddle (a potato chip). That doesn't change with the direction of bed motion. The common 4 point "leveling" systems bend the bed and the undercarriage, they don't level it. How 4 point bed bending has managed to masquerade as bed leveling for so long is one of the great mysteries of 3D printing.

Two adjustments and one reference are exactly all that is needed and exactly what I did in my printer and have been advocating for the last three years.



The screw at the back (hard to reach) is the "reference" and the other two have (had) thumbwheels. I found that in my printer the mechanism was so stable that I didn't really need to adjust it at all, so a couple months ago while doing other work on the Y axis, I removed the springs and thumbwheels and just bolted the bed in place, still using the three screws that stand it off the bearing blocks.

More details of the original design here: [www.instructables.com]

I prepared a document that illustrates the differences between 3 point leveling and 4 point bed bending. You can see it here: [mark.rehorst.com]

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

@d_d: what about the shock_absorbing function of the spring part? Are you betting on the hotend never loosing position and never crushing into the bed?

I wonder if I make one spring-thumbscrew a fixed one, if the head ever crashes in that area, there will be nothing there to cushion it and one might damage the nozzle...

Edited 2 time(s). Last edit at 06/01/2016 08:35AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

If the bed moves in the Y axis, as mine does, each time the bed reverses direction the bed will bounce unless the springs are very stiff. The bouncing will affect print quality (I had this problem in a much earlier version of my printer when I used a glass bed plate). In my adjustable implementation as seen in the CAD drawing previously posted, the springs were very strong and would not provide any sort of safety for the nozzle in the event of a crash into the bed. Fortunately, it hasn't been an issue.

I did something dumb about a year ago and made the nozzle crash into a print that was stuck to the bed. It bent the heat break which I was able to straighten and continue using, but I ordered a spare for just-in-case. So far I haven't had to use it.

There are a lot of ways a printer can damage/destroy itself. Most people don't use endstops at the maximum ends of the axes. There's nothing to stop the carriage or bed from slamming into the far end of the axis if there's a malfunction, but most accept it as a manageable risk. Few add thermal safety devices to their bed or extruder heaters. You weigh an assumed probability of damage/self-destruction vs the effort and cost of mitigation/repair vs the effect on print quality and make your decision. I can live with mine.

If you wanted absolute protection against a crash for the hot-end, you could install screws to act as hard mechanical stops for the Z axis that would mechanically prevent the nozzle from ever actually touching the bed. You'd have to carefully level and zero the bed first, then adjust the stop screws to prevent the nozzle from ever touching the bed, which means adjusting them within 50-100 um or so. How many printers are so stable that the leveling and zeroing and the position of those stop screws would remain accurate for any length of time/number of prints? As solid as my printer is, I'm not sure I could make it work.

Edited 2 time(s). Last edit at 06/01/2016 09:19AM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Agreed.
Also keep in mind that the mass of the Y axis changes as the print goes on. I had a 6.5 hour print the other day that was unsettling seeing a tall part in ABS move so fast while printing infill at about 60mm/sec.
Personally I love the XY mechanics and just move the bed in the Z axis. Much easier on the print IMO.

One of the inherent flaws of the Y axis bed motion type design is that tall prints with small footprints on the bed will tend to wobble as the bed reverses direction, reducing the print quality. You can mitigate it somewhat by reducing print speed and acceleration, but who wants to do that? It usually isn't a problem for prints with large bases.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
I did something dumb about a year ago and made the nozzle crash into a print that was stuck to the bed. It bent the heat break which I was able to straighten and continue using, but I ordered a spare for just-in-case. So far I haven't had to use it.

Shouldn't motors skip steps before that happens? What monsters motors do you use?

Quote
the_digital_dentist
If you wanted absolute protection against a crash for the hot-end, you could install screws to act as hard mechanical stops for the Z axis that would mechanically prevent the nozzle from ever actually touching the bed. You'd have to carefully level and zero the bed first, then adjust the stop screws to prevent the nozzle from ever touching the bed, which means adjusting them within 50-100 um or so. How many printers are so stable that the leveling and zeroing and the position of those stop screws would remain accurate for any length of time/number of prints? As solid as my printer is, I'm not sure I could make it work.

That is actually pretty clever but difficult to do if the screws are on the bed. But what if the screws are sitting just around the hotend? Like 3 screws that you adjust when you do the manual leveling. Take a thickness gauge and "calibrate" the nozzle to 0.1 or 0.2 and then repeat calibration for each of the 3 bolts that go around the hotend using the 0.05mm gauge just as if they were nozzles themselves.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

The X and Y axis are driven by NEMA-23 motors in my printer. They won't skip steps until the bed or extruder carriage are driven to the ends of their axes.

You can put the protective screws on the extruder carriage because they can't protect the nozzle from hitting the bed plate. They might protect it from a side impact...

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Yeah, that was stupid of me indeed to fail to see that.
Wouldn't it be a better way to have 2 endstops for Z (in case of a corexy for example)? A parallel circuit (using both Z min and Z max for preventing a head crash and get rid of the flimsy spring-based shock absorbers?

I have a design that would prevent sideways movement but funny enough I haven't yet designed a way to clamp the bed, just floating it on screws with silicone caps. It's basically a hex coupler in a hex hole and it can slide up and down should the bolt's cap be pushed down. But in what situation would the bed move side to side?



Edit: and yeah, the clamps are plastic so maybe useless anyway...

Edited 2 time(s). Last edit at 06/01/2016 12:19PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

You can put in all the switches you want, but if the electronics fails to recognize the switches, bad things may happen. The only way to really protect the extruder from a head crash is to mechanically prevent the bed from moving high enough to hit the nozzle or prevent the nozzle moving low enough to hit the bed.

The same is true for the extruder and bed heaters. You can put in temperature sensors but if the controller malfunctions and doesn't read the sensor or ignores it, you could have excessive heating. The only real protection for the heaters is to use a bimetallic switch or thermal fuse mounted on the heated part that interrupts the current to the heater. Its hard to find such parts that are rated for hot-end type temperatures and hard to find parts rated for bed heater level currents, so most everyone does without that type of protection.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
You can put in all the switches you want, but if the electronics fails to recognize the switches, bad things may happen. The only way to really protect the extruder from a head crash is to mechanically prevent the bed from moving high enough to hit the nozzle or prevent the nozzle moving low enough to hit the bed.

The same is true for the extruder and bed heaters. You can put in temperature sensors but if the controller malfunctions and doesn't read the sensor or ignores it, you could have excessive heating. The only real protection for the heaters is to use a bimetallic switch or thermal fuse mounted on the heated part that interrupts the current to the heater. Its hard to find such parts that are rated for hot-end type temperatures and hard to find parts rated for bed heater level currents, so most everyone does without that type of protection.

I would have thought it more likely that one of the wires to the endstop gets loose or something like that then the electronics failing to see the endstop. And even if it does, would it fail to see all endstops? My reasoning about using Zmin and Zmax was that even if it fails to see one switch it would see the other. On that train of thought why would the electronics fail to see the switch? And if it does, are all switches in the same boat?

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

A failure is a failure, whether it's a loose or broken wire, or malfunctioning code. I don't know every possible reason (stray beta particles from radioactive decay in the uCs plastic packaging?), but controller boards occasionally behave contrary to their programming. Noise on the power or logic inputs can cause all sorts of odd behavior, and with a minimum of 4 motors in a printer, there are plenty of noise sources available. How many times have you seen problems caused by ground loops via USB between a printer and a host computer?

Most of the potential problems are dismissed as unlikely, and we pretend they don't exist, and most of the time it's OK, or we do things like twist motor leads to reduce the noise coupled to other lines. But if you really want your machine protected against every possible failure, you have to design the mechanism so it can't hurt itself under any mechanical, electrical, or thermal conditions. That's pretty hard to do.

In theory, there won't be any forces trying to move the bed sideways in a CoreXY printer. But vibration can do odd things. Your design has silicone "springs" on the screw heads supporting the bed plate. If you turn the leveling screws, there is friction between the silicone and the bed that will essentially wind up that spring which will take every opportunity to unwind. If the plate vibrates and those little springs unwind, will the plate remain in the same lateral position? I think it's a safer bet to prevent lateral motion by having the plate constrained at its corners or by holding it down with magnets, etc.

Never forget the dumb human factor. If you're moving the machine and turn it on its side without remembering to remove or secure the plate first, where is it going to end up? The only way I can transport my printer is to lay it on its side in my car, so I have to be careful to remove anything that might come loose (side panels, filament spool and holder, stray tools, etc.) when I turn it sideways.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
A failure is a failure, whether it's a loose or broken wire, or malfunctioning code. I don't know every possible reason (stray beta particles from radioactive decay in the uCs plastic packaging?), but controller boards occasionally behave contrary to their programming. Noise on the power or logic inputs can cause all sorts of odd behavior, and with a minimum of 4 motors in a printer, there are plenty of noise sources available. How many times have you seen problems caused by ground loops via USB between a printer and a host computer?

Most of the potential problems are dismissed as unlikely, and we pretend they don't exist, and most of the time it's OK, or we do things like twist motor leads to reduce the noise coupled to other lines. But if you really want your machine protected against every possible failure, you have to design the mechanism so it can't hurt itself under any mechanical, electrical, or thermal conditions. That's pretty hard to do.

In theory, there won't be any forces trying to move the bed sideways in a CoreXY printer. But vibration can do odd things. Your design has silicone "springs" on the screw heads supporting the bed plate. If you turn the leveling screws, there is friction between the silicone and the bed that will essentially wind up that spring which will take every opportunity to unwind. If the plate vibrates and those little springs unwind, will the plate remain in the same lateral position? I think it's a safer bet to prevent lateral motion by having the plate constrained at its corners or by holding it down with magnets, etc.

Never forget the dumb human factor. If you're moving the machine and turn it on its side without remembering to remove or secure the plate first, where is it going to end up? The only way I can transport my printer is to lay it on its side in my car, so I have to be careful to remove anything that might come loose (side panels, filament spool and holder, stray tools, etc.) when I turn it sideways.

Yes, I understand and agree. I was wary about the silicone springiness but couldn't think of something else especially that silicone can be high temp and is also somewhat non-slippery. In a previous iteration I did have constraints at the corners but those are not possible anymore now after I have changed several things ... I am still scratching my head for this as I am not completely happy with the bed support solution I have.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

The RepRapPro Ormerod has an interesting defence against head crashes. The head pivots about the X axis and is held in position by gravity. If you lower it too far, it just pivots and avoids damage to the bed. If it hits a blob or curl up on the print, it usually rides over it.

Another possibility I have considered is to use a spring loaded head, which would again allow the head to rise when contacting the bed, or ride up over a small obstruction.

Edited 1 time(s). Last edit at 06/02/2016 01:03PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I am seriously thinking of implementing some sort of renishaw_probe-style hotend, so that any head-crash would just cut the power to the motors or something like that...

@dc42: ... can you post a pic or something (link?) about that? I am very curious...

Edited 1 time(s). Last edit at 06/02/2016 09:46AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Quote
realthor
I am seriously thinking of implementing some sort of renishaw_probe-style hotend, so that any head-crash would just cut the power to the motors or something like that...

@dc42: ... can you post a pic or something (link?) about that? I am very curious...

There are some pictures on the assembly instructions page here [reprappro.com]. The X-carriage and head assembly rotates around the smooth rod, and the runner on the back of the X arm stops it going too far. The weakness in the original design is the acrylic X-arm which tends to twist slightly, but many Ormerod owners have replaced it with an aluminium arm.

Delta printers sometimes use use the nozzle itself as the Z probe, with either FSRs under the bed supports or an FSR or microswitch in the head assembly to detect contact with the bed. The same device could be used to cut power to the motors.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Very strange design... why would they go with a single rod and an acrylic plate... Two rods would have been cheaper/less flimsy ...

So it's the general consensus that the springs in the bed leveling bolts should be strong enough to keep the bed in position but will be too strong for an eventual crash for absorbing the shock?
If that's the case why not ditch them completely like d_d did and only have some bolts with a locknut instead of the spring, with a built-in wheel for easier turning.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

You might as well have a switch/contact/opto endstop to detect nozzle to bed contact if you are going to have a hinging carriage.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

The problem with schemes that use a switch to detect a head crash is the same that problem that causes the z=0 switch to be ignored. You can't rely on malfunctioning electronics to behave the way you want.

The problem with mechanical methods, such as allowing the head to tilt or have a spring(s) to allow the extruder/hot-end or bed to move out of the way is the limited range of motion they can provide.

The only guaranteed way to prevent hot-end damage is to physically stop the bed and extruder nozzle from moving close enough for the two to touch. That requires installing finely adjustable, sturdy mechanical stops for the Z axis. If you physically block contact, you're counting on the motor to slip and not tear any teeth off the drive belt if the axis is belt driven.

That prevents damage due to the nozzle crashing into the bed plate, but doesn't help with the nozzle crashing into a print or clip used to hold a glass plate on the bed.

I don't think there is a mechanism that can protect from all possible crashes without severely compromising print quality.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
realthor
Yes, I understand and agree. I was wary about the silicone springiness but couldn't think of something else especially that silicone can be high temp and is also somewhat non-slippery. In a previous iteration I did have constraints at the corners but those are not possible anymore now after I have changed several things ... I am still scratching my head for this as I am not completely happy with the bed support solution I have.

What if you glued disc magnets to the top of the silicone pads and to the underside of the bed plate? The smooth magnetic discs would slip against each other when turning the leveling screws, preventing the silicone "unwinding" problem, and would keep the bed plate in place better than relying solely on gravity to do so.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

How about installing two (inductive) sensors probably the narrower lighter type, one setup to give a z min probe position in the conventional ABL sensor setup, the other backup hardwired sensor setup to trigger at or just below nozzle to bed contact, this could be set to just trigger the kill switch or some other power interrupt to kill the motor.

Yes I know it's still electronic and can fail but it would add significant redundancy for little cost and be far more reliable than a software solution. Physical limits sound nice but they are going to be difficult to set and prone to drifting off.

Edited 1 time(s). Last edit at 06/07/2016 01:42PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
the_digital_dentist

Quote
realthor
Yes, I understand and agree. I was wary about the silicone springiness but couldn't think of something else especially that silicone can be high temp and is also somewhat non-slippery. In a previous iteration I did have constraints at the corners but those are not possible anymore now after I have changed several things ... I am still scratching my head for this as I am not completely happy with the bed support solution I have.

What if you glued disc magnets to the top of the silicone pads and to the underside of the bed plate? The smooth magnetic discs would slip against each other when turning the leveling screws, preventing the silicone "unwinding" problem, and would keep the bed plate in place better than relying solely on gravity to do so.

The only problem with magnets is a heated bed. They loose magnetism pretty quickly with temp. I am thinking more and more that I need either of:
a) a milled recess in the plate that would receive a round bolt-head (allen type) - best would be a half sphere recess to receive a dome-head bolt-
or
b) a bolt that i can file the end to a round and insert a drilled round piece that I would cover with silicone or even some double sided adhesive pad and which will stay fixed while the thumbwheel will rotate the bolt underneath it.

The more I think of it the more feasible the b) option looks .... filing the threads off the end of a bolt is quite easy (put bolt in drill by its head and press the file against it while it is rotating until there are no more threads). Then take a thick washer or some sort of drilled disk or even a printed part that is dipped in silicone on its head and insert it on the smooth end of the bolt. And voila, non-slip and no rotation.

Edit: the screws used in table clamps that have the swiveling, free-rotating cones would be ideal: fill the cone with silicone and you're set.

Edited 3 time(s). Last edit at 06/07/2016 07:19PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I kind of made a more realistic (I believe) support, one that involves some aluminum angles and strips of cork for insulation and grip to at least confine the glass/ALU plate to one direction (Y). I am still thinking about how to confine the X axis around the bolts as there is space for it. If you have any suggestions I am all ears



I also need to find an angle that is shorter that 12mm on one side and 20mm on the other side, so an asymmetrical ALU angle.

Edited 1 time(s). Last edit at 06/11/2016 05:51AM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I am not sure if this will help for your printer but the picture below shows what I use for supports and adjusters. At three points are the adjusters - not too different from in your diagram, but at the other three points are 6mm diameter stainless steel posts which are a close sliding fit to 3mm thick PTFE plates to take out any side loads.



This design started as a quick kludge to overcome the print bed taking up random positions in X and Y but is now fitted to all of my printers. Although the geometry is not great, it is able to accept any point of the hexagon being depressed by 15mm without sticking.

Mike

Hmm, yeah, not sure what can I take from your setup but I like the levered adjusters (can't see how the thumbwheels do the work there) and also I see you use piezo disks for nozzle contact sensing... or at least this is what I can take from your image. Nice setup though.

My setup is quite simple, 3 rods and 3 bearing-blocks that also act as square tube clamps and level adjusters. The issue I faced with the previous design is that I didn't have any way to constrain XY movement of the bed and now I have added the ALU angle to constrain on Y and I could use the adjusting bolts on the 2-rods side to constrain the bed movement on X. I can't do much for X on the single-rod side so I must somehow take advantage of the two adjustment bolts' heads on the other side.

I would also like to know if the cork the bed rests on is ok from the leveling pov.

Edited 2 time(s). Last edit at 06/11/2016 04:14PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

I do remember seeing 300mm of cork being used under a laser setup to measure something called "Mossbauer effect" which probably means that it is good for anything. The adjusters are not on levers, just a parallel mechanism to avoid introducing any additional compliance into the system.



Piezo disk not marked but under the pressure pad.

Mike

Quite complex but nevertheless pretty professional. I am trying on the other hand to make something very simple.
Here is a very simple way to constrain laterally (on X) the build plate:



Due to the cork the heat should not creep into the ALU angle so an ABS simple part and an M3 allen bolt should do the constraining. I do believe that some sort of insulating pad should be there at the tip of the bolt so the heat doesn't creep back through the allen bolt into the plastic.

Edited 1 time(s). Last edit at 06/11/2016 05:39PM by realthor.

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Quote
the_digital_dentist
The problem with schemes that use a switch to detect a head crash is the same that problem that causes the z=0 switch to be ignored. You can't rely on malfunctioning electronics to behave the way you want.

The problem with mechanical methods, such as allowing the head to tilt or have a spring(s) to allow the extruder/hot-end or bed to move out of the way is the limited range of motion they can provide.

The only guaranteed way to prevent hot-end damage is to physically stop the bed and extruder nozzle from moving close enough for the two to touch. That requires installing finely adjustable, sturdy mechanical stops for the Z axis. If you physically block contact, you're counting on the motor to slip and not tear any teeth off the drive belt if the axis is belt driven.

That prevents damage due to the nozzle crashing into the bed plate, but doesn't help with the nozzle crashing into a print or clip used to hold a glass plate on the bed.

I don't think there is a mechanism that can protect from all possible crashes without severely compromising print quality.

Or install the nozzle 1mm below the X Rods so that if the bed hits the nozzle the flex of the rods will allow for that 1-ish mm and the bed will remain pushing against the rods. Will 10mm rods flex for 1mm if the bed forces against the nozzle before the heatbreak gives away and bends?

---

RepRap Lander concept on Concept Forge
RepRap Lander concept on RepRap Forums
My Things, mostly experimental stuff

Lol...Just put X arm on a hinge...then you might only need to level bed in one direction.

Edited 1 time(s). Last edit at 06/12/2016 04:47PM by MechaBits.