Possibly building 3D printer - seek a bit of advice ...

Hi All,


I have decided to enter the world of 3D printing but am a bit unsure of which of the many paths into 3D printing to choose. Thus, I am posting here - hoping that one of you may help narrow the scope of options ...


... As it is I built a CNC mill a couple of years ago which proved to be quite successful in that it holds a precision of ~3/1000mm within a square of 100x100mm and ~2/100mm in its full milling volume. Built in cast iron I am very pleased with it. BUT it also took A LOT of time to design & build - and although I am tempted to build a 3D printer I cannot set aside (this) much time to do so (50 hours could be fine, though).

On the other hand I am looking for a quality 3D printer which is very reliable in use and will allow me to make precision 3D prints with many different filament materials, including high temp filaments, and heavy weight filaments (e.g. copper filled filaments). Absolute dimension precision matters as well as surface finish.


Additionally, I have a somewhat limited budget, as I would prefer to not go beyond around USD 500. This is something that I reckon does not make things easier, as it, as far as I can see, unfortunately excludes Voron, SK-Go, & Ratrig builds.

One more limitation is that I am in no way a programmer - nor do I have any experience with setting up firmwares etc. on e.g a Raspberry Pi. I am aware that this is less than optimal - to amend this I have instead considered buying a Creality Sonic Pad and set it up to work with the printer I get/have. I can see that on Github there are firmwares for different printers - and on youtube videos on how to load a new firmware onto the Sonic Pad. Even without experience I think it looks accessible to do.


One possible advantage is that I have access to some 40*25mm rectangular aluminum rods than can be precision milled on the CNC mill. I reckon such rods could make for a rigid 3D printer structure.


With the above in mind I am considering:

- either buying an Ender 3 S1 Pro which I may eventually rebuild to a larger volume and more rigid construction, or

- e.g. something like the TwoTrees Sapphire Plus (https://www.twotrees3dofficial.com/en-de/products/corexy-3d-printer-sapphire-plus-sp-5-dual-z-axis?variant=42712549720284) - again with the aim of modifying it along the way.

Given that I will be printing heavy objects I have excluded the Ender 5 because it only supports the print bed on one side and also has a bit too small print volume.


However, knowing that when it comes to precision everything matters I just wonder if either of these 3D printers constitute a sufficiently good basis to build on? I.e. is the software/ hardware good enough to form the basis for something like an eventual SK-Go quality printer .... Or, if there may be a better suggestion for a good start-out 3D printer that may eventually be modified to become really good?

Also, I would, if possible, appreciate some information on which "brands" are considered to make precision 3D printer parts: i.e. glass-filled "tooth" belts, the "rollers" & bearings that these belts "roll" on, sufficiently good ball-screws, other bearings that require precision, etc. This would allow me to improve the 3D printer in due time.


Well, I hope the above is not too "fluffy" and that I may meet with someone here who may know/sense where I want to go with this - and have some input ...

In any case thanks for reading & maybe replying

Cheers, Jesper M

3D printers are a lot easier than CNC mills, so I expect you'll have a good time whichever way you go. For heavy prints, something with a fixed bed would be ideal, but bed Z works too if it has good support like that Sapphire Plus. Bed Y may struggle with inertia toward the end of the print.

For high temp filaments, you need a hotend that doesn't have PTFE tube in the melt zone. And probably an enclosure, which is good for any printer. Maybe even heated chamber, but the bed heats it to some degree. The Sapphire has an optional acrylic shell, but it seems overpriced, especially since it doesn't even enclose the top (always more troublesome than the sides since the filament still needs to reach the moving printhead). I'd probably cut 3 trapezoidal panels and a small square to make a pyramidal top, with a bowden adapter threaded into the center of it. Use reverse bowden from the spool to there, and then free run of filament down to the printhead after converting it to direct extrusion (bowden is terrible, don't trust anyone who says otherwise).

50 hours and $500 to design and build a high temperature 3D printer? and you want quality? and you want to print "heavy" objects? and you don't know how to set up firmware? and you don't have access to materials?

It's going to take more than 50 hours just to select the parts/materials.

---

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

Hi both,

& thanks for your relevant feedbacks. Well ... the thing is that I don't really seem to find the "right" solution within the constraints that I have time-wise & money-wise. The Sapphire Plus could be ideal but are e.g. the linear rails good enough to constitute the basis for upgrading? Also I have just learned that the firmware appears to be not too good (even in the current version) which makes it less attractive for someone like me.

As to TDD's comments they are also all relevant, although I, as I wrote in #1, have access e.g. to 25x40mm aluminum square rods. However, when being honest I agree completely that "It's going to take more than 50 hours just to select the parts/materials." ... which is the reason why I wrote this: "Also, I would, if possible, appreciate some information on which "brands" are considered to make precision 3D printer parts: i.e. glass-filled "tooth" belts, the "rollers" & bearings that these belts "roll" on, sufficiently good ball-screws, other bearings that require precision, etc."

In my experience with the CNC mill build, what takes the most time is finding quality components that are also accessible price-wise (e.g. the glass-filled tooth belt I understand you may be using).

Right now I am not quite sure where to go with this ... I think I will just wish you a fine evening and then consider things.

Cheers

Jesper

I'd recommend against using ball screws on a 3D printer. The $20 Chinese ones are amazingly good for the price (about .001" backlash on my CNC), but they do have a bit of eccentricity, so you need a rigid frame and rails to prevent that from introducing wobble (not an issue on a mill since you need rigidity anyway), or something akin to an Oldham coupling to allow lateral movement (MirageC on youtube posted a video about that recently). And they're too slow for X/Y on a 3D printer, and don't provide any advantage for Z which never has backlash anyway due to gravity.

Belt driven X/Y is great because it's fast and zero backlash, and T8 leadscrews work well for Z since they're pretty flexible and have some thread clearance so they generally don't cause wobble unless they're really bent and the frame is really flimsy. Some people do belt Z too, but that has the issue that it will drop to the bottom if power is lost, plus has lower full step resolution.

One of the best tools to build a 3D printer is another 3D printer, so one good option would be to buy the cheapest Ender 3, use it to get a feel for 3D printing and to print whatever parts you need for your custom machine design, and then dismantle it and use its power supply, motherboard, steppers, etc. in your new printer. You might even save money compared to buying all the components separately due to shipping cost, and certainly will save time shopping.

Don't expect to get perfectly precise prints just because your motion system is super precise. Thermal shrinkage tends to make things come out slightly smaller than they are laid down. Kind of like injection molding, though less extreme than that since it's single extrusion lines doing the shrinking rather than the whole thing melted at once. Often you can correct for it using the "horizontal expansion" setting in Cura, but sometimes you need different expansion on different features so then you have to fiddle the model itself. But since you have a mill, you can print close and then machine to perfection And print models for lost PLA casting and then machine the important features. That can save a lot of time and metal compared to machining from a solid block.

Hi dekutree64,

Thanks again for your relevant & helpful reply - I appreciate your feedback

I will comment below but while I remember it: I noticed in the "screwstruder" thread that you had commented there: With this in mind - did you ever try out this extruder type yourself? And, if so, did you find it to be "excellent" - or did there appear to be maybe some "systematic" shortcomings? I may of course ask rq3 himself but since we are in contact now I thought I would ask you here ...

Regarding 3D printer design you have also answered one of the questions that have been lurking in my thoughts for some time: why didn't anyone use ball screws for the X & Y axes? It makes sense that speed and eccentricity could be explanations although I would guess that vibration & imprecision of the belts would make belt driven X-Y axes less precise in the 1/10 to 1/100 mm scale. But obviously just guessing here ... and maybe the inherent imprecision of the 3D printing method itself makes this a less important consideration.

I think the Ender 3 is an interesting suggestion for a start-out printer. As it is I have - besides the TwoTrees already mentioned - considered both the Ender 3 (S1 Pro) and the Ender 5 (S1) but have been ambiguous about them because I already know that their print volume is too small, and also, should I decide to (re)build a 3D printer myself, it would be preferable to be at least able to mount a dual extruder. To this end, as far as I understand this, I would need a controller board with five stepper motor outputs which is not present on either of the Enders. But I may just go with one of them (and maybe also a cheaper version) to get started.

As to my quest for a place to buy "good quality" components I happened to watch CNC Kitchen's Voron 2.4 build video on youtube and here noticed that Voron generates component lists for their various printers including suggested retailers. Here I happened to visit Fermio.com's website - located within the EU, and apparently selling quality components, this may be a huge step forward in time savings! Also looks as if they sell quality belts: [fermio.xyz]

Regarding milling the 3D printed objects to size, yups, it could be an option - but usually the key objects I work with are very thin (2-5 mms) and "wide" (~100*100mm, or bigger) - and I need them to be absolutely flat (in some cases better than 1/100 mm). IME not easy to accomplish even on a good CNC mill since essentially all the feasible materials I have tried tend to bend due to internal material stresses. Fastening the materials in a way so that they do not become stressed is another challenge. I reckon something similar will happen with 3D printed materials and expect to try out mild annealing in due time and likely lapping machine sanding. In this context I also reckon your "horisontal expansion" setting may come in handy

My feel is that I will seriously consider some version of the Enders ... however, should you, or another rep-rap'er have a better suggestion I would appreciate hearing about it ...

Cheers

Jesper

Edited 2 time(s). Last edit at 05/11/2023 12:11PM by evalon.

If you're interested in building a tall printer, a belt lifted z axis can be made to work easily and probably more cheaply than if you use screws to do the lifting. A worm gear drive with a 30:1 (or more) drive ratio will stop the bed or XY stage from dropping when power is cut. See my UMMD design linked in my sig, below. I used a Rino that costs about $118 via ebay, but someone at the makerspace built a printer using a cheaper Chinese stepper with worm gearbox attached and it worked fine. My 30:1 Rino drive gives me 20 um/full step resolution in the Z axis. Unless the bed or XY stage is very large, use no more than 3 screws/belts to do the lifting.

I also recommend using a flat bed plate such as cast aluminum tooling plate on a kinematic mount for stability. If you lift the bed plate or XY mechanism with a single motor you won't need to include a bed sensor or multiple motors, cables, and drivers to power the Z axis and everything will be simpler and probably more reliable (there's much less to go wrong). For an example of reliability, my UMMD printer has not had the bed leveling tweaked in almost 2 years, since the last time I did some work on the Z axis.

If you use a heater that's smaller than the bed plate, the cooler edges don't expand as much as the heated center and that causes the plate to bow. Use a heater that's the same size as the bed plate and bowing is much less of an issue. Mount the heater on the plate using silicone, not the 3M adhesive sheets that most people use.

If you want to print high temperature materials, you'll need to enclose the whole thing and probably heat it, so it is best to design it with that in mind from the start. You can make large ABS prints reliably at about 50C. For other higher temperature materials you may need some heroics to insulate the motors, belts, pulleys, and linear guides from the heat.

---

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

Quote
evalon
I will comment below but while I remember it: I noticed in the "screwstruder" thread that you had commented there: With this in mind - did you ever try out this extruder type yourself? And, if so, did you find it to be "excellent" - or did there appear to be maybe some "systematic" shortcomings? I may of course ask rq3 himself but since we are in contact now I thought I would ask you here ...

IMO it's absolutely revolutionary and I can't believe it hasn't taken the community by storm yet. I have one on my SCARA printer, which is a rather flimsy machine so not the best for precision testing, but it works a lot better than the bowden extruder I originally had on it. You must use ABEC-5 or better bearings or it won't follow the helix on retraction. The carrier can be printed with integral posts and use 3mm coarse thread screws to hold the bearings on, rather than the expensive shoulder screws. It takes some trial and error to get the dimensions to come out just right, but 3D printing is repeatable enough that you can make .001" changes to the blade penetration depth and it makes a clear difference.

Quote
evalon
Regarding milling the 3D printed objects to size, yups, it could be an option - but usually the key objects I work with are very thin (2-5 mms) and "wide" (~100*100mm, or bigger) - and I need them to be absolutely flat (in some cases better than 1/100 mm). IME not easy to accomplish even on a good CNC mill since essentially all the feasible materials I have tried tend to bend due to internal material stresses. Fastening the materials in a way so that they do not become stressed is another challenge. I reckon something similar will happen with 3D printed materials and expect to try out mild annealing in due time and likely lapping machine sanding. In this context I also reckon your "horisontal expansion" setting may come in handy

Yeah, that will be just as troublesome on 3D prints. Even on thicker prints, the bottom tends to be very slightly convex after cooling and releasing from the glass bed. And horizontal dimensions on the first few layers are tricky because there's such a fine line between failing to adhere to the bed and smushing out so much that you get "elephant's foot". Although in my case that's mostly due to the fact that neither of my printers has a perfectly flat bed or a bed probe for mesh leveling, so I have to level them to where the lowest point will still adhere, and everything else gets smushed a bit.

Quote
the_digital_dentist
If you're interested in building a tall printer, a belt lifted z axis can be made to work easily and probably more cheaply than if you use screws to do the lifting. A worm gear drive with a 30:1 (or more) drive ratio will stop the bed or XY stage from dropping when power is cut.

Great idea! I wonder if a guitar tuner could be adapted as a dirt cheap worm gear... I may try that on the next revision of my SCARA.

Hi both ... Just a brief message to say that I have seen your posts and will get back either on Sunday or Monday. Have a fine weekend

Jesper

Guitar tuners don't have bearings on either gear because they are made to be adjusted a little at a time and then sit. I would also not expect the quality of the gears to be very high which will result in repetitive artifacts in prints. There's no substitute for quality gears. I tried to make a Z axis worm gear drive using some cheap gears I bought via ali-express. It was awful.. The Rino drive uses very high quality gears and bearings (it's about an $800 part if you buy it new).

I used a ball screw to drive the Y axis in a printer for a while. It could run fast and it was smooth, but it had a resonance at around 50 mm/sec that would cause noise, vibration, and skipped steps. I had to limit print speed to about 40 mm/sec to avoid the resonance because I could not find a way to get rid of the resonance. I eventually converted it back to belt drive.

---

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

Hi TDD & dekutree64,

Another quick follow-up as I am currently quite busy with many different things ... I am still considering what to do but have very little time. I do however appreciate your feedbacks and will get back again when time allows ...

Cheers, Jesper

Couldn't a wiper motor or automotove window motor give its worm gear ? These motors are dirt cheap; or even free.
Only problem is removing the rotor ; an angle grinder could do it. The CNC can machine the couplers.
These worm gears are intended to last very long under heavy loads, so we can assume they are precision parts (steel worm + nylon pinion + self lubricating bronze bushings).
(long ago, built a low flow pump around a salvaged one, and the guts were really looking good)

It would be worth a try, but I wouldn't count on it working for the Z axis. Those things are made to be cheap, not to provide accurate or precise positioning. That means the gears are probably not very well made and you'll see artifacts that repeat in the Z axis of prints every time the worm completes a rotation.
The Rino I linked has very high quality gears and very fine adjustment for backlash (that's what the screws that surround the output gear are for). There shouldn't be any backlash because a printer's z axis is going to be loaded by gravity, but it's nice to know that gears are good enough for that sort of adjustment to be useful in other applications.

I tried making a worm gear drive using some gears I bought via ali-express. The result illustrates the pitfalls of using crappy gears for this application.

---

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