CoreXZ

I am still in the process of cleaning up all my files so that I can publish my design but I thought I would share. (The files will eventually be here.) It is proving to be a rock solid printer for me. It uses the CoreXY geometry in the XZ plane with a little twist. It gets me 3X the resolution in the Z direction.



I have printed as fast as 200mm/s. I get great surface quality at 90mm/s.

I use 100lb test braided fishing line. 24 625VV bearing/pulleys. 3 spools.

Guitar tuners are used to tension the system and to adjust the tilt of the gantry.

The firmware has been written and tested.

222x200x222mm print volume

After using Mendel style bots and using this, I would highly recommend this one. No more screw artifacts. The gantry can't get accidentally tilted. One less stepper. Two steppers push the effector back and forth but are mounted on the frame. You can move fast in the Z direction allowing for non-planar printing. Moving the printer during a print causes no artifacts. (Even flipping it upside down.)

I will probably design an all printed version at some point but the laser cut frame is very cheap.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Quote
nicholas.seward
Guitar tuners are used to tension the system and to adjust the tilt of the gantry.
The gantry can't get accidentally tilted.

Nice, the cables are in an anti-racking configuration. For those who are older, and were taught drafting with a pencil and paper, the cable routing is reminiscent of that used on the horizontal bar used to draw straight parallel lines.

Edited 1 time(s). Last edit at 07/04/2014 12:10PM by A2.

Very nicely done. Cost and BoM?

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Wow... This looks like it has the potential to be the new popular standard for RepRap.

Any plans to have kits for sale? The video had me throwing my money at my monitor.

I forgot to post some YOUTUBE CANDY.

@MrDoctorDiv: I am organizing the BOM. I suspect the cost could range from $350-$600 depending. I will post it here when I get it done.

@Feign: I sure hope that I help inspire people to ditch lead screws and belts. They really don't make sense for the light weight problem we are faced with. I did build this as my personal response to the Prusa i3. Time will tell if people start to switch over.

I do have plans to kit this printer. I am shooting for the end of the month. I plan to only make a few a month. I am trying to balance my day job with this without taking away my time to design and build crazy things.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Quote
nicholas.seward
I do have plans to kit this printer. I am shooting for the end of the month.

YESSS!

Quote
nicholas.seward
I plan to only make a few a month.

NOOOO!

Quote
nicholas.seward
I am trying to balance my day job with this without taking away my time to design and build crazy things.

It's understandable I suppose. You've got some lucky students, man. You should really look into getting some kind of manufacturing partner to turn out a steady line of kits though. Ideas get taken a lot more seriously when they've got sales numbers attached.

@Feign: I would certainly welcome any manufacturing partners. The design is open source after all. I would be glad to tweak the design to fit anyone's stock and setup.

However, I know this design is tons better than the standard Mendel/Prusas but that doesn't mean everyone will believe me. I will have to make enough units of my own to drive that point home.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Very interesting and apparently very practical Nicholas. Please put me down for one of the first kits when you decide to offer them.... I'd take one tomorrow if it was available.

How would you implement leadscrewlessness for CoreXY? Just wondering your point, it's not that difficult, but you have more design experience than I do by far.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

@MrDoctorDIV: I have about 20 ways this could be done. However the basic concept is you need to use the same tech that keeps the gantry from racking/tilting. However, you need to do this in two dimensions. This and this will give you a basic idea.

I have to be honest. I really dislike moving the Z platform like this. All the really well engineered solutions are too complex and all the simple solutions are compromises. If I had to pick, I would use two rails on one side with a lead screw and a cantilevered bed like the Makerbot. You can always just replace the leadscrew with a block and tackle.

IMHO, avoid a Z moving bed. You will either compromise or add cost and complexity. (Don't let me stop you. There is probably a good idea out there waiting to be discovered that is hiding from me because it knows I don't like it. :-)

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

@ nicholas

pardon my ignorance... Am i getting it right that it is bad to move the platform in the z direction but ok in the y direction ( prusa i3 ). Wasn't it the other way around ? What is the logic behind not moving the platform in the z direction?

I was planning on building a corexy with a cable drive z like the tantillus , thinking it was a good solution but you got me confused..

reprap forum tantilot

thanks

Edited 1 time(s). Last edit at 07/11/2014 03:17AM by ekaggrat.

The Z-axis bed needs to be leveled and orthogonal, and if not perfectly aligned in the Z-axis, it will have pronounced projection error the further it moves away from the nozzle. Moving the bed only in the X-axis (longitudinally) does not have projection error, it only needs to be leveled.

A couple of advantages of a completely stationary bed are:
Requires less or no calibration between prints.
Imparts no vibration, allowing more viscous/fluid deposition.

@ekaggrat: A2 is exactly right. Moving the bed as Mendels do makes it easy for leveling purposes. However, some people take issue with moving that much mass around quickly. Theoretically, you could have a tall 2kg print that you would have to slow down so the steppers don't skip or the part doesn't detach. However, in practice, nothing that we print normally weighs anything compared to the bed itself and the steppers are good enough to at least move the bed. It is a compromise that only really works well for 3D printing. It doesn't work as well for CNC milling.

I also agree with A2 about going for a stationary bed if you can. That will stick you with a column delta, grounded delta, tripteron, or some dumb printer like this.

For the record, it is not bad to move the bed in the Z. Most printers with a big build area to footprint area ratio do it this way. You may look at a printer like my CoreXZ and say that it would have a better ratio but you have to consider how far the bed can stick out in both directions so the Y footprint dimension is always more then twice the Y work area dimension. I have seen people squeeze 200mm range out of 300mm with a Z bed. Additionally, it is not that a Y-bed is better, it is that the Z bed will either cost more or cause more headaches. It is really a simple matter of making the Z bed rock solid. There is nothing inherently wrong with the Z that I would say don't use it. My whole point is that I don't like for the cost/complexity reason. Also for the record, it is only marginally more expensive and more complex. I am just a member of the Simplification Church. If it can be done better with less parts and or less money, I have to do it. I spent a month on the CoreXZ trying to find a good way to remove 2 bolts and 2 bearings that were theoretically not needed but in the end all other solutions become more complex when they were removed. It is hard being this way.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

@nicholas

thanks for the clarification .. I know a delta is a perfect setup and already have a 3dr delta .I was just curious to see a corexy machine's performance and wanted to build one just out of curiosity . i am now leaning more towards the core xz now.. Designwise is it possible that can be slimed down more and the frame be supported with some foldable extensions?



thanks.

@ekaggrat: It can most definitely be made into a folding bot. The XZ part of it is quite slim. I tried to make it happen but at the end of the day it conflicted too much with my other goals. Of course, you will have to completely redesign the frame. Sounds like fun.

Just to give you an idea of how crazy I went with the simplification process... There are only 10 laser cut pieces and only 6 printed parts that matter.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

@nicholas

will wait for the final design files ...

@nickolas Great concept! Since building my i3, I've had nothing by aggravation from the flawed z axis.

I've started putting together a CoreXZ design using V-slot. I'll post it up once the preliminary design is finished.

Edited 1 time(s). Last edit at 07/18/2014 06:14AM by LMColl.

@LMColl: Sounds like fun. Let me know if you need a design review or anything.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

@nicholas

Brilliant design, just started planning a reprap variant. What is the behavior of the z axis when the printer is powered off? Does the gantry stay in place, or fall towards the print bed? Is there any noticeable error from the fishing line "walking" on the spools as they wind/unwind?

Edited 1 time(s). Last edit at 07/20/2014 06:02PM by NathanaelXYZ.

@NathanaelXYZ: I have had instances with low string tension that allowed the gantry to fall about 10mm. With high/regular tension, the gantry usually never budges. In any case, the farthest it has fallen is 10mm so at the end of a print, I park the gantry at the top and off to the side and have yet to have a problem.

Before I go into the calculations, string walking on the spool for the most part just changes the tension in the system. Of course, a change in tension of a system with more string on one side of the spool than the other will result in a slight displacement if we model the string as a spring. To give us an idea, I am going to calculate based on the false assumptions that the strings don't stretch and only one side of the string leaving the spool matters. These values will be over-inflated but we know for a fact that the error will be less.

Y-stage:
Distance from spool to static redirecting pulleys: 75mm
String walking: 3mm
Mechanical advantage: 2

Worst case error: [75-(75*75+3*3)^.5]/2=30um (With symmetric loading, I would guess we never even hit 1/4 of this value.)

XZ-stage:
Distance from spool to static redirecting pulleys: 400mm
String walking: 10mm
X mechanical advantage: 1
Z mechanical advantage: 3

X worst case error: [400-(400*400+10*10)^.5]=125um (This error would be from a move from one corner of the XZ range to the opposite one. If you look at the error when you just move across in the X direction the biggest walk is only about 2.5mm which leaves you with an error of 8um and that 8um is an overestimate as stated above.)
Z worst case error: 125/3=42um (This error would be from a move from one corner of the XZ range to the opposite one. If you look at the error in the Z in just one layer then you get 3um. Additionally, if the X is held constant then the biggest Z error from a top to bottom move would be 94um.)

Bottom line: even if the error was this bad, the printer would be acceptable but I fully expect the actual error to be an order of magnitude less. Just the act of calibration would effectively 1/2 most of these error values and the symmetrical loading will further diminish the errors.

PS I read your blog and wonder why you have decided to go with a friction drive. The spool solution as shown above causes little error and will never slip. I look forward to seeing your progress.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

@Nicholas Seward:
Thanks for the numbers, and for checking out my blog! The friction drive is just an experiment. It only cost ~$5 to throw in some m5 aluminum spacers with my mcmaster order, vs. $21 plus shipping for three spools from tridprinting. I will switch to spools (especially in light of what you have said) if it doesn't pan out.

I have also thought of a way to reduce the size of the x-gantry, at the cost of some added complexity:


-Only one cable is shown. The other would be mirrored.
-The pulleys are in numbered order. (The 7 next to the drive motor was cropped out.)
-Pulleys are spread across two "planes:" front ("F") and back ("B").


There are a couple more details on my blog.

First, I would not worry about the added complexity. It is very easy to drill holes and route the string so that isn't really a problem. However, I think you can keep the string in 1 plane and get smaller.

Here is the arrangement that I now use. It keeps the string in 1 plane with only the addition of 2 bearings. The gantry is about as compact as it can be in the X direction.



Sidenote: My first prototype uses double stacked 608s. That is by far the cheapest solution. 25 cents per bearing for a total of 50 center per pulley. What I didn't realize until the prototype was built was just how much weight that would be. For just a few more cents I am now using much smaller and lighter 625vv bearings. It ends up costing about $3.60 more in the end. (The biggest reason I use them is that they just look like they belong. The 608s just didn't look like they belonged.)

Edited 1 time(s). Last edit at 07/23/2014 03:56PM by nicholas.seward.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

This is an intriguing design and I'm right in the middle of designing a rebuild of my printer using aluminum extrusions and openbuilds vslot for the linear motions. I'm thinking I could implement a variation of this somewhat easily, but I'm guessing that the Marlin firmware would know nothing about how to operate it. Would that be correct? I know it has corexy support, but probably not corexz.

Bruce

@brucehvn:

I have hacked together a Marlin branch for my machine. I don't claim that it is a general solution but I would be glad to share it. I would spend more time on making the firmware handle generic transformations but ARM chips seem to be the future and Smootieware is good at that already.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Quote
nicholas.seward
@brucehvn:

I have hacked together a Marlin branch for my machine. I don't claim that it is a general solution but I would be glad to share it. I would spend more time on making the firmware handle generic transformations but ARM chips seem to be the future and Smootieware is good at that already.

Thanks Nicholas. I'd like to take a look at your branch. So, are you using a smoothieboard and smoothieware now?

Bruce

Quote
brucehvn
Thanks Nicholas. I'd like to take a look at your branch. So, are you using a smoothieboard and smoothieware now?

I am not using it yet but I am moving in that direction.

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Would it be so bad to move the Z rails and bearings to the other side of the gantry?
Then the pulleys could move out further toward +-X.
The string forces wouldn't be in the same plane as the guide rails, but not far away.

Quote
DaveGadgeteer
Would it be so bad to move the Z rails and bearings to the other side of the gantry?
Then the pulleys could move out further toward +-X.
The string forces wouldn't be in the same plane as the guide rails, but not far away.

The short answer is you can do this.

The long answer is that I struggled with this design for months. I went through many many iterations. It was a game of Whack-A-Mole. I would fix one thing and break another. The main thing would be if you flip the rail and linear bearings to the other side of the gantry it will do one or more of the following...

1) expose the string
2) require tall spacers and screws for the 625vv's
3) increase the width of design (if the gantry and back plate are on the same side of the Z rails)
4) require something other than a laser cut gantry so you can make a place to screw in the pulleys (right now for simplicity I designed it such that you could use 1 laser cut piece for the gantry)

I hope I missed some simplification. I surely tried. (Although, I think I got it pretty simple.)

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

Does Smoothieware support CoreXZ yet?????

@vreihen: No, but I assume it can be done easily. Wish I had time to do it. I am probably going to save my new Azteeg X5 for my CorEXY. (It is a CoreXY that drives the extruder with a belt from a static stepper.)

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ