Design questions

Hi!

I've seen a couple of corexy 3d printers and was wondering what the best design of axes is in terms of the resulting print quality.

1. Z axis design - are the belts used on UMMD better than ball screws? I can see the advantage of the Z axis design with 3 ball screws, because the bed can be auto-leveled, but aren't the belts betters in the resulting print quality? Can ball screws be the same or better quality than the belt solution?

a) 2 belts - like on UMMD - [drmrehorst.blogspot.com]
b) 3 screws - kinematically coupled bed - [www.youtube.com]

2. Floating X axis - is it really needed to have the X axis floating. I mean have it mounted with a bearing block on one end because of thermal expansion. I guess 99% of printers have the x axis firmly attached and it seems it doesn't cause problems.
UMMD x-axis: [4.bp.blogspot.com]

1) it depends on multiple factors. Belts are driven by something- pulleys, gears, etc. If those are poor quality, they will create repetitive errors in the Z axis of prints. Of course, you can have poor quality ball screws, too, but they're less likely to create repetitive patterns in the prints. Belts are a LOT cheaper than ball screws, especially if you're using more than one, and especially if you want an arbitrarily long Z axis. Ball screws may, and belts definitely require some sort of brake to prevent bed drop when motor power is cut. There are all sorts of approaches, depending on whether you want to keep the bed from moving or let it down gently. I used a worm gear drive in UMMD because it prevents the bed from moving at all and the cost was favorable compared to buying a motor with a brake.

1a) I used two belts in UMMD because I used two linear guides for the Z axis. The linear guides I used are rather large and don't have any detectable play so I felt they were sufficient. It's easier to set 2 guides parallel than 3 or 4. I put the guides near the center of the Y axis because I didn't want to cantilever the bed off of one side which tends to cause the bed to bounce like a diving board. I have essentially created 2 cantilevers, but half as long.

1b) Many people think "I want autoleveling, so I need to use multiple motors, drivers, cables, etc.", but what they really want (whether they realize it or not) is to not have to keep releveling the bed after every print. If you're building with a flat bed plate, like cast tooling plate, you can lift the bed with a single motor and don't have to worry about the bed tilting and requiring releveling before every print. Autoleveling is a self-fulfilling prophecy- as soon as you use multiple motors to lift the bed, you have to use autoleveling because the bed will tilt a bit every time you turn the power on because the motors jump. If you lift with a single motor and it jumps, all screws/belts jump equally and the bed remains level.

If you use a flat bed plate, such as cast tooling plate, using an intelligently designed mount will allow easy leveling, and it will be stable over temperature. An intelligent mount has three support points- the bed is a plane and three points define a plane. Lifting is the same- three screws/belts are ideal. A kinematic mount allows easy manual leveling, and allows the bed to expand when heated.

2) If you build the XY mechanism using linear guides on a rigid frame, you may find that the mechanism will bind when ambient temperature changes. That happens because as the frame warms the Y axis guide rails move apart and as it cools they move closer together. If you heat the printer to print ABS, PC, etc., it's a bigger problem. Linear guides are great bearings but don't tolerate much position error of that type. Letting one end of the X axis float prevents binding as the Y axis rails move closer or farther apart. If you use round guide rails, they'll probably flex enough that you won't need to let one end float, but then you have to deal with the sagging, etc., that comes with them, too. Many designs mount an X axis linear guide on an aluminum plate or tube. In theory, that aluminum piece will expand along with the rest of the frame of the machine (also aluminum) so you won't have any problems with binding (though you might have problems with it remaining straight due to the differential expansion of the steel and the aluminum). In UMMD, there's only the steel guide rail for the X axis, and it doesn't expand as much as the aluminum frame, so the floating end of the X axis is needed. UMMD has never had any problem with the XY mechanism binding.

Edited 3 time(s). Last edit at 08/09/2022 11:01AM by the_digital_dentist.

---

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

Thank you very much digital dentist.

1a) so the z axis belt is used because it's cheap especially for longer z axis? I thought the decision for the belts was to avoid the repetitive errors (wobble) that the screws create. Is it then ok to use the screws if you focus on quality prints and not so much on money? I mean if I buy a standard ball screw, will it be better than the belt with Chinese pulleys? Btw. I have Rino at home with Nema 23 motor on it, so the bed falling wouldn't be a problem.

1c) convinced me that I actually didn't really want a self-leveling bed, but flat, precise, only once-leveled bed that would work forever :-)

2) Then why don't people use it on their printers? I've probably only seen it on UMMD. I also noticed that you used wider rails or at least it looks that way in the photos. I have some Hiwin MGN12C's that are not that wide, maybe they could be used and wouldn't 'wobble'.

One question: Does the preload class matter for the linear rails?
I found this table: [motioncontrolsystems.hiwin.us]


The rails I have at home have the light preload (0.02 Cdyn Light Preload). They do not move as "freely" and are not very quiet. Perhaps they are more intended for CNC than for a 3D printer.

Edited 3 time(s). Last edit at 08/09/2022 11:12AM by printman.

If you have bent screws in Z, they can create repetitive errors in the Z axis of prints by causing lateral shifting of the bed/print layers. That problem can be prevented by anchoring the bed to the Z axis bearing blocks and letting it float on the Z axis nuts. If you drive multiple screws with a single motor, you have to use belts and pulleys to connect the motors and screws. If the pulleys are out of round, they will cause repetitive Z axis errors in the prints in the form of varying print layer thickness.

In a belt driven Z axis, out of round pulleys will do the same thing they will do in a multiple screw driven Z axis- cause variations in layer thickness. The Rino has excellent quality gears and mechanical support so it isn't an issue, and in most machines, it won't let the bed move when motor power is cut. Unlike screws, belts will never cause the bed to translate horizontally.

I believe that using belts for the Z axis, especially a long Z axis, is cheaper than using screws to do the lifting, and considerably easier to build.

When I designed and built UMMD I was mostly concerned about reliability and print quality and much less about speed. UMMD's reliability has been excellent. I typically print at 80-100 mm/sec with a 0.4 mm nozzle and about 30-40 mm/sec with a 1mm nozzle, and start long prints before I go to bed at night, so it doesn't matter if they take 4 hours or 7 hours (or 23 hours or 37 hours). I don't manufacture prints for a living- this is hobby stuff. I can wait until the print finishes, however long it takes. I used 24 mm wide rails for X and Y in UMMD, primarily because they were available cheaply at the time I was shopping for rails (that's the nice thing about designing/building from scratch- you can use whatever is available without worrying about obtaining specific/expensive/too-cheap parts on someone else's BOM). They're pretty big and heavy, so they increase moving mass, as do the F608 bearings I used for pulleys. The width of the X axis rail helps keep it from twisting under the weight of the extruder carriage, but a smaller rail screwed to an aluminum tube would probably be OK, too.

You'd have to ask others why they don't use a floating X axis in their printers. I suspect most have never considered it, and most printers you see are open, and not heated for printing ABS, etc., so it may not be necessary. I have seen posts here by people who built XY stages similar to mine without the floating X axis complaining that they had to adjust the linear guide positions when their shop heated up in summer (they built the machine in winter when the shop was cold).

Preload in linear guides is for high precision positioning mechanisms. It's fine in a 3D printer as long as the motors have enough torque to push the bearing blocks, but a corexy stage requires a bit of force to move without adding preload, especially if you throw in extra pulleys like some designs do to allow more convenient positioning of the motors. Preloaded rails may limit the top speed you can achieve if you want to print super fast. The preloaded type tend to wear faster than the regular guides, though a 3D printer is a very light load for most linear guides, so it may not matter.

Edited 1 time(s). Last edit at 08/09/2022 05:45PM by the_digital_dentist.

---

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