New Printer Design - Could do with some feedback / thoughts on spec!

Hi All,

I'm new to posting here but have been reading a fair few posts from the forums over the last few years!

I'm planning a new printer build soon that is a completely custom design and I was hoping you could look over the spec I've come up with and the ideas I have and tell me if I'm on the right track, if there is anything critical I'm missing or if there is something I could be doing better!

The printers main focus is print quality above all else but other key targets are a reasonable footprint and low noise! Currently I'm planning on it being a coreXY machine but I am open to going the delta route (or something more exotic if there are quality gains to be made!). I'm not particularly interested in doing fancy things like tool changing or multi material currently but if I can leave the option to work on that in the future open then that's a big plus! To get that print quality the main thing I'm focusing on is machine rigidity and a lightweight print head.

So plan so far is:

Motion:
Core XY, (Chosen for the stationary motors!)
220 x 220 mm build plate,
270mm Z height ( the 300 mm MGN rail will travel exactly 270 end to end because of the carriage size!)
All axis running MGN12 linear rail (MGN 9 on X), (Chosen due to the rigidity in basically all directions and rotations!)
Two Z lead screws (Dont want a cantilevered bed due to the deflection!)

Machine Size:
X: 400
Y: 400 (may be a little smaller depending on how the design goes)
Z: Pretty much whatever

Frame:
3030 Extrusion (Mainly chosen for the rigidity)
5 hole corner brackets instead of internal brackets for better squareness Link!

Electronics:
Duet Wifi
24V, 18A PSU
200 Step NEMA 17 (Considering 400 step - anyone got any info on if its worth it?)
Bed levelling with Precision Piezo

Filament Path:
Zesty Nimble Extruder, (hella expensive for what it is but essentially best of bowden and direct drive combined)
E3D V6 (pretty much the goto for everyone)

From this point I guess I have a couple of main questions as well:
Do I go for a static bed and raise / lower the whole CoreXY system or do I mount the CoreXY system to the top and raise / lower the bed?
I cant see a good argument for either option. I guess the lead screw might present a problem should the coreXY system move as it would have to pass through somehow! Any thoughts?

Is there a concern about racking the rails on the Z axis or am I being paranoid?
I don't want the Z axis to jam as it moves up and down. It would wear the bearings faster and also potentially deform the machine / bed / something in a way that affects print quality.

My design in its current form can be seen here! I'm not happy with a lot of things with it and a lot of stuff isn't connected properly or quite the right size! So treat it with a grain of salt.

CoreXY is fine, but moving the whole thing? No thanks. Moving the bed in the Z axis would be so much better. Lifting the CoreXY 8n the Z axis defeats the whole purpose of not having the motors moving.

For the Z rails, (Well moving rails on any axis, really) careful alignment is needed, to reduce wear and noise, but a careful fixed alignment is usually good enough. Some compliance in the bearing or rollers can make up for small imperfections and avoid transmitting any artifacts to the print.

I wouldn't say moving the XY gantry in the Z axis defeats the purpose of coreXY. CoreXY having stationary motors means the print head and moving parts of the gantry are kept light. I don't see how lifting the XY gantry in the Z axis defeats this purpose.

That being said, I'd say go for the moving bed. The XY gantry is much more difficult to get square and level than a single plate (your print bed), so I'd go for making the easier to work with part be mobile to avoid as many twisting and racking issues as possible. That and the print head moving around will produce horizontal forces, so having that mounted rigidly and securely (rather than as a moving platform) seems like a good idea. A print bed won't produce those forces. Again, this is all "theory" and the difference it makes may be negligible in practice.

If you want quality and reliability I'd stick with Cartesian over delta. I know that might be controversial but I'm not saying that without reason. Deltas have a non constant resolution/precision along the XY plane, and any errors in the geometry that occur (diagonal rods not all the same or correct length, towers not all evenly spaced, parallel to each other and perpendicular to the bed, etc. etc.) will lead to hard to diagnose artifacts. I can see dc42 jumping in saying the duet's auto calibration can correct for some of those but that isn't the same as all of those. I'm not saying a delta can't produce good prints, a "sloppy" delta can probably produce prints that are good enough for me, but I don't know what your standards are like.

Also, that "deltas print faster" thing is a load of bull. A Cartesian machine with a bowden extruder has comparable moving mass. Deltas aren't more rigid to cope with the higher accelerations, either, those diagonal arms are anything but rigid. They do look cool though.

You don't need/want 4 linear guides for the Z axis. One near each screw is all you need. Linear guides provide guidance, they don't keep the bed from tilting. Using 4 of them is going to be an alignment nightmare. Even if you manage to get them aligned, when room/machine temperature changes, the guides will move apart when warmed and closer together when cool and the whole thing may come to a grinding halt. Quality linear guides are great bearings but very intolerant of misalignment.

If you're going to heat the machine to print ABS you need to consider how the XY stage will work when it heats up and the Y axis rails move apart because they're mounted on an aluminum frame.

Move the bed, not the xy stage. The only wires that go to the bed are the heater and thermistor. The XY stage has all the extruder wiring the motors and the endstops. I suppose you could mount all the electronics on the xy stage and move them up and down, too, but it usually best to mount the electronics outside the print chamber for easy access and in case you decide to heat the machine for printing ABS.

3030 t-slot should be rigid enough for that size machine. If you get the stuff with square ends and matched lengths you can bolt it directly together and skip the corner plates.

---

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

Replying to a few people here in one go (not sure what the forums policy on multiple replies is)

Quote

CoreXY is fine, but moving the whole thing? No thanks. Moving the bed in the Z axis would be so much better. Lifting the CoreXY 8n the Z axis defeats the whole purpose of not having the motors moving.

It doesn't remove the benefit at all - when the head is moving X and Y the motors are stationary. Its only when the Z moves and id wager that the CoreXY system is lighter than the heavy bed! Check out the Voron V2 for an example of this.

Quote

For the Z rails, (Well moving rails on any axis, really) careful alignment is needed, to reduce wear and noise, but a careful fixed alignment is usually good enough. Some compliance in the bearing or rollers can make up for small imperfections and avoid transmitting any artifacts to the print.

I planned to use the external frame as the guides for the Z rails instead of a separate internal frame like most printers - it means the frame must be assembled exactly which is concerning but maybe I'm overthinking it!

Quote

I'm not saying a delta can't produce good prints, a "sloppy" delta can probably produce prints that are good enough for me, but I don't know what your standards are like.

I want to use the machine for engineering prototypes so an extremely tight tolerance is what I'm after! Gotcha with the deltas - the only appeal was the smooth acceleration and braking of the carriages. Certainly look cool though. I work on an interesting machine at work that is a ‘kind of’ delta which would be interesting to convert to a printer.

Quote

You don't need/want 4 linear guides for the Z axis. One near each screw is all you need. Linear guides provide guidance, they don't keep the bed from tilting. Using 4 of them is going to be an alignment nightmare. Even if you manage to get them aligned, when room/machine temperature changes, the guides will move apart when warmed and closer together when cool and the whole thing may come to a grinding halt. Quality linear guides are great bearings but very intolerant of misalignment.

My fear exactly - would two central rails (like in the images) work ok? Im concerned about the bed tilting either way! Is thermal expansion a major concern though? I will be printing mostly PLA and not in an enclosure!

Quote

If you get the stuff with square ends and matched lengths you can bolt it directly together and skip the corner plates.

My choice of the corner plates was for squareness - its much easier to get a big plate like that square from a laser than it is to cut extrusion square. (also much easier to measure squareness on larger sides)

I seen posts from people reporting that when they set their linear guide based corexy machine up their shop was warm and when winter comes and the shop temperature drops a bit, the mechanism suddenly binds. Or they set it up when it was cold and when the shop warmed up it started binding.

I also used linear guides in the XY stage and made it immune to temperature changes by adding an extra bearing block on the X axis linear guide. I heat it to 50C when printing ABS and it never binds.

I used two linear guides in the Z axis of UMMD and it has been working very well. The bed support that rides on the Z axis linear guides is aluminum like the frame, so it expands even as the Z axis linear guides move apart as the machine heats up. No issues with heat there, either.

---

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

Quote
Ed3D
My fear exactly - would two central rails (like in the images) work ok? Im concerned about the bed tilting either way! Is thermal expansion a major concern though? I will be printing mostly PLA and not in an enclosure!

Like The Digital Dentist mentioned, one linear rail per side can be sufficient. If you're using one lead-screw per side as well and are concerned about end table flex, you could use a linear rail with two carriages on it to greatly increase the stiffness. Though that would give up some Z-height travel, always a compromise.

Ive been looking at the Voron quite a lot recently and its an interesting machine - seems to go against what some of you are saying though. What do you think of the design?

If you use 3 Z leadscrews/motors instead of 2 (1 in the center back and 1 in each front corner) and you use a probe on the hotend, you can take advantage of auto bed levelling.

Quote
gtj
If you use 3 Z leadscrews/motors instead of 2 (1 in the center back and 1 in each front corner) and you use a probe on the hotend, you can take advantage of auto bed levelling.

DC42 has implemented this in his firmware but yet I never heard of someone using it. Note this function is to set the bed (a flat one !) // to the XY plane, not to compensate on the fly for bed imperfections. Also, you are likely to bind the leadscrews if correction is excessive and/or not enough play, no self aligning nuts.
Fact is, people who understand how it works and able to make it work are capable of designing and building a cartesian machine (Delta is a different story) that doesn't require this extra complication, will go with one Z motor, use good components with a sound design, align them properly and won't even need auto bed leveling etc... ..

Edited 1 time(s). Last edit at 09/20/2018 06:08AM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
If you use 3 Z leadscrews/motors instead of 2 (1 in the center back and 1 in each front corner) and you use a probe on the hotend, you can take advantage of auto bed levelling.
gtj
If you use 3 Z leadscrews/motors instead of 2 (1 in the center back and 1 in each front corner) and you use a probe on the hotend, you can take advantage of auto bed levelling.

I agree with MKSA - the 3 point levelling thing is neat but it's treating symptoms not problems - a properly design Z should stay level at all times and the on the fly corrections will do the bed imperfections.