Z mechanism options

[Michael Hackney]

I'm starting to research building a Core X-Y to compliment my multiple delta and cartesian printers. I understand the Core X-Y mechanism and will use a "non-crossing belt" arrangement that Carl Feniak documented. I'm also leaning towards Open Builds Slot too. My question focuses on options for the Z axis. Carl's implementation uses a cantilevered bed. I'm planning a 12"x12" bed with aluminum heat spreader, borosilicate with PEI and heater so that's a lot of mass. I'd like to use high quality ball screws both for the precision and for the speed (rapid "Z lifts foresting control like on my deltas). I'm thinking of a 12"x24"x24" build volume so the Z will have a pretty large range of movement. Here are some of the options but perhaps there are others (and why I'm posting this!):

1. rails and guides at each side with parallel screws
2. rails and guides at each side with a single screw on one side
3. rails and guides at each side with two Z steppers and belts (most likely 9mm) one on one side
4. rails and guides at each side with a single Z stepper and belt (most likely 9mm) on one side
   

One nice thing about the Core X-Y design is that it seems to be very modular since the X-Y motion control is isolated from the Z control. With the basic frame and Core X-Y mechanism built, you can incorporate basically any Z mechanism you'd like.

thanks in advance,

Michael

[John Gelnaw]

People who've built a larger C-bot have gone for the "triple c-bot", using three leadscrews (usually one each side, and one on back or front), with a belt driving all three leadscrews off a single motor.

That's the route I'm going, but I've been stalled in Real Life, and haven't been able to get back to finishing my build.

[Michael Hackney]

Thanks, interesting! I googled it and found some photos and BOM. I like that arrangement. He doesn't say if he's using ball screws or simple threaded rod.

[whosawhatsis]

For other options, see the discussion here: https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!topic/3dp-ideas/qvMgYblsKh4

This discussion is based on the premise that rather than using multiple leadscrews or belts, you can use a single actuator plus a passive system of cables and pulleys to ensure that three points on the plane of your platform move together.

--
You received this message because you are subscribed to the Google Groups "H-bot and CoreXY 3d printers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to h-bot-and-corexy-3d-...@googlegroups.com.
Visit this group at https://groups.google.com/group/h-bot-and-corexy-3d-printers.

[John Gelnaw]

On Saturday, February 20, 2016 at 7:29:05 PM UTC-5, Michael Hackney wrote:

Thanks, interesting! I googled it and found some photos and BOM. I like that arrangement. He doesn't say if he's using ball screws or simple threaded rod.

Lead screws, which are in between.  Usually, the tr8*8.  The ones from zen toolworks are popular, because they have a turned down end.  The openbuilds version is cheaper, but without the turned down part.

I went the openbuilds route (using their threaded block and collar), but as I said, I haven't finished mine, so I can't say if it works well or not.

[adam paul]

The Zen tool works lead screws are used in the triple-c.

[Michael Hackney]

Thanks John and Adam. I prefer turned down ends and have made my own turned end screws for my CNC converted mills and lathes. 

I'm thinking with the 1' x 2' bed size that 3 ball screws with two at the front (2' edge in front) corners and 1 in the mid back would be a good compromise arrangement to support a bed this large.

Another question I have is around "leveling" and auto-leveling. What do folks use on these Core X-Y machines? I've converted all of my delta printers to use FSRs for auto calibration and my TAZ 4 Cartesian printer to auto leveling using a 3 point FSR system.

[Jetguy]

The whole point of building the Z state right is auto leveling is then no longer required or even remotely needed.

The axis moves as a plane, has no reason to warp or distort, and thus no need to auto level.

You can add it if you want, just saying that the compensation should be nearly 0.

[Michael Hackney]

Jetguy, can you provides some more detail/information on what you mean by "building the Z stage(sic) right"? 

I'm a pragmatist. I can't believe that there is something magical about Core X-Y that eliminates the need to occasionally level the bed - especially on a 12" x 24" bed. At the very least, auto leveling would be useful for setting the Z=0 before starting a print. Any heated print surface is going to "move" due to thermal expansion. Even on my TAZ 4 Cartesian printer with a 1/4" MIC-6 ground aluminum (very flat), silicone heater and borosilicate there is enough variability from day to day that I set the Z=0 frequently. I have outfitted it with 3 point FSR auto leveling and haven't looked back.

[Jetguy]

The problem is, you just combined several problems into a single solution statement.

Core XY has nothing to do with a Z stage. They are independent concepts and mechanisms.
A multiple screw Z stage connected via a timing belt is one of many ways to ensure a large axis plane such as Z can be moved as a plane throughout it's travel.
Bed warping is totally independent of Z stage. The assumption is, you use glass or some other surface which is assumed and reliably flat within a common sense tolerance as a plane.

[Jetguy]

I'm sorry that's probably not the answer you wanted but you compared problems you have with a totally different printer, totally different construction method, radically different design and it's just not even a proper comparison.

A less than optimal design is called that for a reason- it's going to have minor repeatability issues.

Building a design is MORE than just the design, it's construction methods, alignment procedures, materials and quality.

Again, your problems on a TAZ 4 can be avoided by a proper design that is assembled with precision components and aligned properly. You installed a compensation mechanism to account for these errors- yet you knowingly have to reset the compensation several times.

That experience is not going to carry over outside your TAZ 4

[Jetguy]

Again, ;lets really take a look at this and expand your understanding.

You have a TAZ 4

It's basically an i3 derived design. Or call it an oldschool Mendel.

The problems include round rods for all axis, a cantilever frame between Y and Z axis, potential for twist of non-planer rail alignment in nearly every axis.

On a moving bed design, the Y axis rods form a plane, but if the 2 rods are not in perfect alignment, there is a twist in the roller coaster track that the bed rides on.

So yes, I totally expect you to have all the issues I've described with the Wanhao I 3 and were even worse when I moved to 3 point leveling.

It's no wonder you have problems and really and truthfully are going about it the wrong way because you don't understand how the geometry is going wrong.

While we can have these problems on any printer, the construction and design of the TAZ and many printers like it just make it worse.

Further, the typical 3 bearing Y axis method to prevent over constraint, works to a major disadvantage and actually makes the problem 10 times worse.

When  the bed is fully forward, the side with 2 Y bearings has the bearing very near the bed leveling adjusting points. The side with 2 bearings is effectively the main guide.

The other side with only a single bearing is the follower. It's job is to act as an anti-rotation plane around the axis of the main guide rod.

The problem is the differential placement of the bearings VS the typical 4 corner bed leveling points.

So, on the single bearing side, the single bearing is in a different place compared to rod length, and spans between the 2 leveling points on that side. IF the rods are not dead parallel in a true plane. the bed will dip and follow a very odd pattern of misalignment.

[Jetguy]

Other factors that come into play.

In my designs, every single linear guide rail is at the outer most mounting point on the frame. True for Z and for X. Y forms the gantry on mine.

So a fundamental difference is the build area is inside my liner guide rails, where as on most printers (if we take your TAZ4 example) the linear guides are well inside the area of the bed.

This means a 1 percent twist or error in rod alignment is amplified by the much larger bed compared to bearing mounting points and rod path.

I can have a 1% error and by the time you get to the nozzle and bed relationship, the error is reduced to less than 1%

In TAZ 4, a 1 percent error is amplified- maybe a few points to greater than 1 percent error across the plane- but especially great at the far corners.

Honestly, once most people move to a glass plate on silicone pads, if you see significant bed warpage/error, it's your frame and mechanics, not the bed.

So again, this isn't just Core XY, it's everything.

It's overall frame construction (I build on a granite countertop with the printer upside down to ensure flatness of the X Y frame plane).

My supported linear rails/guides are also at the furthest possible distances in the frame structure.

Every single thought is put into understanding how forces and energy are transmitted through the frame and linear guides.

[Michael Hackney]

I understand that. I am separating out the Z mechanism because I understand and have a good handle on the X-Y mechanism. 

What I was trying to say - the bed itself can warp even with ground AL and/or borosilicate. Sorry, I didn't mean to imply the Z stage.mechanism itself. But, that said, even if the bed is perfectly flat, there are multiple ways it can be off-level from the X-Y plane and my experience has been that this will likely happen. I hope this makes more sense. 

[Michael Hackney]

Thanks Jetguy: 

Let's assume that I do know what I'm doing. I get and appreciate your points and one of the primary reasons I want to build this Core X-Y is precisely because the bed only moves up/down (Z) eliminating a lot of the issues you point out. I have all of my delta printers completely dialed in and can reliably print any part, any time, in volume on them. I haven't spent that much time with the TAZ to reach the same level of proficiency. (however, I have overhauled my TAZ 4 with higher quality precision rods and ball bearing linear bearings and beefed up the frame. I also overhauled the bed. In practice, the TAZ is actually pretty reliable but I don't use it daily like I use my deltas.)

I also build on a granite block (I have a pretty well equipped machine shop with 2 CNC milling machines, 2 CNC lathes, a CNC router that I built or converted from manual machines) so I understand all of your points on assembling a rigid and precise frame, twist and torque, parallel guides, etc. So let me back up to the original intent and question of this thread...

I know I can build a good rigid frame to support the Core X-Y mechanism. What I don't know is what are some of the best (or the best) options to implement the Z axis motion control to move a 1' x 2' bed up and down precisely and maintain it's level to the X-Y plane reliably and repeatably? The cantilever table design that I've seen seems problematic, especially as the print area/bed increases in size.  Perhaps in practice that's not the case - that's what I'm trying to figure out. The 3 screw design seems like it would support the bed better but perhaps it is overkill in practice.

I also know that on all of my printers the bed remains pretty darn flat but there are still day to day minor deviations that require setting Z=0 so I can print the parts I manufacture reliably. But again, this is on my delta printers which are different beasts. The white parts on this fly fishing reel require very precise control of the first layer so the "open mesh" is not smooshed too much (this is purely asthetic). I'm able to print this part with near 100% success on my delta printers but I don't use the TAZ to make these.

Cheers,

Michael

On Thursday, February 25, 2016 at 9:49:38 AM UTC-5, Jetguy wrote:

[Ryan Carlyle]

Triple lead screw, all the way. Two screws in corners with adjacent linear guides (or V-wheels on extrusions) and one screw on the opposite middle. Don't add any extra guides on top of the two mentioned, they will only add over-constraint and binding possibilities without any meaningful stability benefits.

Two screw with screws in the middle of opposite sides and rods or rails adjacent to the screws (NOT in the corners) is acceptable but not as good.

I personally don't think thermal expansion is an issue unless you have a heated chamber -- a Cartesian bot with a Zmin endstop is simply much more geometrically stable versus thermal/environmental effects than any delta. The max endstops in a delta put all the frame expansion between the datum points and the bed, which is the worst possible way to keep a good first layer. Delta probes make up for the fact that you're relying on the simultaneous accuracy of three separate switches that are at the opposite end of the printer from the position you actually care about.

If you're worried about it though, put a sensor on your CoreXY carriage and use it as a regular Zmin switch for homing. That will compensate bed height variation such as removable plates. You can go ahead and use bed tilt compensation at that point if you want, but that CAN add artifacts to prints. So it's better to get a reliable mechanical tram. With a mostly-metal triple lead screw printer, you'll tweak the leveling screws maybe once every few months.

[Michael Hackney]

 Thanks Ryan.

I have decided to go with linear guides on this machine. Not sure which ones yet but I'm using these very cool Speed Demon guides from LM76 on an all-metal Rostock Max-like delta. I'm using the SGR-10 with 3 roller blocks and they are smooth as silk. I'd probably use these for the Y and Z rails and maybe the wider SGR-20 for the X (which will be the short ~12" dimension) attached face-on for the carriage. But open to feedback on this idea.

For the Z - two linear guides mounted on the frame uprights at opposite ends of the long dimension (24") i.e. the corners. Need to think about which surface of the upright extrusion these should be mounted on. I'm leaning towards having the guide block face the opposite corner along the short rail rather than face the opposite corner where the other guide is mounted. The screws positioned adjacent to these as shown below. Now guide on the rear - just the screw to not over constrain. But I'm looking for experienced feedback on better arrangements. 

 

I completely understand about over constraining the screws. Made that mistake on the long Z axis of one of my first milling machines (Z=30") and it was a nightmare. Question though - should the Z stepper be mounted on the bottom with the upper end of the screws floating (that seems to be what you are describing) or the opposite way. Other than getting the stepper out of the way of the top, it seems having the fixed/driven end of the screws up top where most of the printing action will be going on (I wold rarely print anything over 4" tall) might be a better arrangement. But I can see arguments for the bottom arrangement too. 

No heated chamber. I'm looking forward to rounding out my 3D printer experience with they CoreXY machine. I'll start without and do a good job on the design and build/tram and see how it performs. 

thanks,

Michael

[William Cook]

Very timely comments for me, I am following this thread closely

On Thursday, February 25, 2016 at 11:51:46 PM UTC-5, Ryan Carlyle wrote:

[Steve Johnstone]

Great post and perfect timing!

I’ve just started a core xy design based on everything I’ve learnt over the past couple of years. I’m shooting for a 300 x 300 x 600 (z) build volume in a heated chamber.

Based on my experience and everything I’ve read, the z-stage is going to be critical to high quality printing. So far this is where most of my time has been focused.

I want the chamber to be as small as possible so that volume that needs to be heated and controlled is minimal.

The structure is going to be pressed sheet metal rather than extrusions and at the moment I’m favoring this for a possible z-stage – 

I'm thinking of using IGUS Double DryLin rails and two screw, belt driven from a single stepper motor - LINK

[Steve Johnstone]

Sorry, pictures got hidden -

Great post and perfect timing!

I’ve just started a core xy design based on everything I’ve learnt over the past couple of years. I’m shooting for a 300 x 300 x 600 (z) build volume in a heated chamber.

Based on my experience and everything I’ve read, the z-stage is going to be critical to high quality printing. So far this is where most of my time has been focused.

I want the chamber to be as small as possible so that volume that needs to be heated and controlled is minimal.

The structure is going to be pressed sheet metal rather than extrusions and at the moment I’m favoring this for a possible z-stage – 

I'm thinking of using IGUS Double DryLin rails and two screw, belt driven from a single stepper motor - LINK

- hide quoted text -

--

[Michael Hackney]

Steve, you may want to reconsider the IGUS rails - in my experience they really are not precise enough for our printers. Check out the SpeedDemons I linked to above, the wide 35 version would be great for your design and these are very cost effective and well made precision linear guides.

[whosawhatsis]

+1 for not using IGUS, especially for Z.

I've actually found V-slot extrusions to work best for Z, because you can get them really tight and make a stiff linear guide (one that won't fall under gravity, even with the screws detached). This can maximize any backlash in the screws, but doing so makes that backlash consistent and predictable, and unless you're using a platform compensation scheme so that your Z axis is constantly moving back and forth, backlash doesn't matter except insofar as it is inconsistent. This method otherwise gives you a much more steady Z axis (both in the Z direction and in X/Y than anything that slides more easily and relies on gravity to provide tension on the screw.

--

[Ryan Carlyle]

I think motor on top and screw hanging down is maybe slightly better for performance -- it puts more mass up near the XY gantry to absorb reaction forces and it puts the screws in tension so they're marginally less likely to whip at high speeds. But I think these are both very minor effects, and arranging sync belts and wiring around other components will probably make the holistic design favor putting motors at the bottom.

[Steve Johnstone]

Thanks for all the advice.

OK - Igus are a no go.

Michael and Whosa - I really like the idea of been able to adjust the tightness of the carriage. One of my big concerns is getting everything aligned and parallel in the z-y and z-x planes. This type of configuration would be great as I've only two rails to worry about.

What size lead screws would you recommend for a 600mm build height?

Ryan - I did think about mounting the motor up top but would make packaging a lot more difficult. I haven't ruled it out and may still do so as the design evolves.

[Steve Johnstone]

I wounder if you could use the V Slots Rack & pinion set up for the Z and do away with the lead screws altogether?

The stepper motors would raise and fall with the build plate so would need some sort of cable chain.

I expect the down side would be synchronizing the motors and tensioning the belts equally??

On Saturday, 20 February 2016 20:53:39 UTC, Michael Hackney wrote:

[whosawhatsis]

While carrying extra mass on you Z axis probably isn't a bad thing (as it is with X and Y), there are a bunch of problems with this particular system. For one, you are using the wheels both as linear guides and as idler pulleys, and they have different effective diameters for these purposes, meaning something is going to be siding and wearing.

Synchronizing the motors should be fairly easy. Driving from the same logic signal (either sharing a stepper driver or using two that receive the same signals) will keep them in sync while moving. When the motors are not energized, they can get out of sync (though if they are physically wired, they will prefer to move together), but the torque would be low enough in this configuration that you could safely run them up against a hard stop and let them skip until they are back in sync. This is ideally done on a machine with digipots, so you can reduce the current to the motors while this happens (to reduce both the stress on physical components and the ugly noises the steppers will make when they start skipping).

--

[Jason B]

On Sunday, February 28, 2016 at 11:53:07 AM UTC-8, Whosa whatsis wrote:

...This is ideally done on a machine with digipots, so you can reduce the current to the motors while this happens (to reduce both the stress on physical components and the ugly noises the steppers will make when they start skipping).

That would also be a pretty neat use of a software adjustable driver.  Put two end stops in series, turn down the torque and home until you get a high signal.

[whosawhatsis]

You'd have be careful about what type of switch you use. If you use a microswitch that will trigger, then allow another millimeter or so of movement before it hard-stops, that's no good. On the other hand, if you have one conductive metal piece (anodized aluminum is no good) hard-stop against another and use them as contact probes in series, that would do the trick.

--

[Jason B]

Oh yes, I'd assume a rather robust hard stop in addition to the micro switch if you are going to be brute forcing it like that.

[whosawhatsis]

But if you have a microswitch and a robust hard stop, there is going to be some non-zero offset between where the switch triggers and where you hit the hard stop. If you hit the hard stop first, the switch never triggers. If you hit the switch first, your last switch to close will stop the movement before that part of the carriage hits the hard stop, and that end will not have moved as far when they all move away together (so if your hard stop is at the bottom, for instance, that corner will end up higher than the others).

[Jason B]

Ahh, but since we designed to ram the hard stops anyway we don't really care where the switches are, they just tell us we're done homing.  So send enough steps after the signal to ensure you've exceeded the switch range of motion and call it done.  In real equipment we have to set limit switches to a jig and/or verify that they are properly trammed when commissioning but something like this sure would simplify things. 

We also have beams that use two laser rangefinders to get position and auto level the axis, but that's only accurate to two or three millimeters on a 15 meter tall z axis.  A bit overkill but nice until someone drops a wrench on to the laser and you have to explain why a new one needs to be ordered.

[whosawhatsis]

Yes, if you modify the software so that the motors keep stepping after the switches are triggered, it should work, though you have to be certain that (and of how far) things will continue to move after hitting the switch. It would probably be more reliable to throw the switches out and just step run the motor long enough that it will hit the hard stop no matter where it started.

[Ryan Carlyle]

Don't really need a double max switch, just take the length of the axis and multiply times the homing speed and run it that long. Or home to Zmin, run 95% down fast, then do another 10% to the hard stop. Then rehome Zmin.

[Jason B]

I'm just figuring it would be nice to save all that z homing time especially since 95% of the time your bed won't be anywhere near full extension.

[whosawhatsis]

It's not that much time, especially if you do it while your heaters are heating up.

--

[Ryan Carlyle]

I mean, if you do a single-start screw a meter long, it'll take a while, but yeah, shouldn't matter if you do it during preheat.

[Steve Johnstone]

Many thanks for all the responses - I've much to think about.