Kicking The Tires Before You Buy: 3d Printers

So you’re looking to buy your first 3D printer, and your index finger is quivering over that 300 US Dollar printer on Amazon.com. Stop! You’re about to have a bad time. 3D printing has come a long way, but most 3D printers are designed through witchcraft, legends, and tall tales rather than any rigorous engineering process. I would say most 3D printer designs are either just plain bad, or designed by a team of Chinese engineers applying all their ingenuity to cost cutting. There are a few that are well designed, and there is a comparatively higher price tag attached.

I’ll start by going through some of the myths and legends that show up in 3D printers. After that I’ll go through some of the common, mostly gimmick, features that typically hinder your printer's ability, rather than adding any useful function. Next I’ll go onto the things that will actually make your printer better. Finally, I’ll add some special consideration if you’re a beginner buying your first printer.

Most printers have very mechanically weak designs. You will never be able to just throw a spindle on 99.9% of 3D printers and get a CNC machine. 3D printers are typically not designed to take loads, they are not designed to handle dust, they aren't designed for alignment, or anything else needed for a CNC machine. Printers get away with an enormous amount of mechanical sin because of two things. One, there is barely any load on a 3D printer. Two, the additive manufacturing doesn't need anything but better-than-nothing position control to work. What I’m getting at is that most printers are terrible, and they will work anyway.

Bushings will work just fine. Really. A self-lubricating bushing on a precision, hardened rod will work super well for a 3D printer. It will be just as precise, just as smooth, have less maintenance, and cause no issues compared to a proper bearing. The LM8EUU bearings used in most 3D printers are usually made to such terrible tolerances that they could be making your printer less accurate.

Most of the time, none of this matters anyway. Because, your bushing or bearing is likely to be pushed into a flimsy piece of 3D printed plastic; which will instantly negate any precision advantage from either solution. If a printer advertises linear bearings or bushings going into a steel or aluminum part you may get a precision and rigidity advantage over another printer choice. Otherwise, it's unlikely.

Another thing that isn't going to do any good is a regular 608 skate bearing riding against an extrusion. These bearings have to be pre-loaded to provide any precision. They are designed to take up some axial misalignment. In order to get rid of this axial misalignment, you have to pre-load the bearing with a spring or bolt, pressing the inner race against the ball and in turn the outer race. Both must be firmly held. If this isn't happening, which in 90% of the 3D printer and CNC designs that use this method it isn't, you will face at least some misalignment. The perceived benefit of these bearings just won't have the desired effect on the precision of your movement.

In conclusion, the type of bearings in most printers won't make much of a difference unless they are properly restrained, loaded, and aligned; which is expensive to do.

A NEMA 17 stepper motor is probably overkill for a 3D printer. They just happen to be the cheapest and most readily available size. Most 3D printers that have trouble printing with a smaller, weaker motor are badly designed and need the extra power. I don't see many printer manufacturers advertising smaller motors, but some have tried to pass off larger ones as an upgrade, which is dubious.

GT2 timing belts, or even low-stretch string, are plenty for your printer if you want to keep the cost low. Even if you have a precision ground acme lead screws with a properly adjusted and pre-loaded lead nut, you still won't get any advantage unless you hold them properly, which is expensive to do.

That isn't to say, if you had a Class 0 ball screw on a properly ground and squared three hundred thousand dollar mechanical movement that you wouldn't see more precision. It's just to say that if you put a ball screw in your machine and hold it in place with a mostly hollow block of PLA, it's meaningless.

Also, I will mention, if the choice is between a bit of hardware store all-thread and a ground or rolled lead screw with lead nut of any description, it absolutely, 100%, will make your printer more accurate. Especially for Z movements where the weight of the extruder or bed is pressing down on the lead nut, pre-loading it.

Makerbot did this puzzling thing with the Replicator 2X. They built an expensive steel frame for their printer, and then held all the moving parts with bad injection molded plastic. They’re so useless that there's a whole ocean of aftermarket businesses replacing the plastic parts. Most 3D printers have their primary movements made with 3D printers. Which means that you have a mostly flexible material trying to remain rigid. See the problem? It's going to flex under load.

A few printer manufacturers have gone out of their way to make sure that load bearing parts, bearings, etc. are placed in metal. These are very not reprap printers. It can't be helped though, unfortunately the laws of physics trump design ethics.

The first layer of your print is the most important. It doesn't matter if you have a raft. The error won't get taken out in a few layers. The truth is, that any error in a lower layer of your print is likely to be transmitted to a higher layer. So you must have a level bed. Most printers with large beds have a huge piece of aluminum flat stock that's not held to any tolerance, which in turn has three springs with screws in the center, holding up a circuit board without any tolerance, holding a glass plate without any tolerance.

You are likely to have a bad time. For most of these movements, the smaller the movement the less error you will see. If a cheap rod has a straightness error of +-0.5mm over 300 mm, you are likely to see that 0.5mm at the end of 300mm, but maybe only .16 over 100mm. When you add in all these cheap mechanisms you start to have an impossible to solve tolerance stack. After all you are trying to keep a nozzle 0.18mm +-.02mm away from a glass plate at all points. That's pretty tough. A smaller printer will give you better results than a large printer that costs the same money. Auto bedleveling can compensate some for this.

Also, large prints eat a ton of plastic. Expect to be risking 30-40 dollars on a large print.

Weird mechanisms won't help you print better. Again. You are trying to position something accurately and repeatably. Every time something is added to the equation it's one more thing that can go wrong. So anyone advertising a new innovative mechanism who isn't simultaneously awarded accolades for groundbreaking work in engineering, is probably selling you a gimmick. If you are into it, that's fine, but don't expect an improvement in your print quality. We’ve been designing things that move in a straight line for a long time. It's known.

Note: I’m not talking about delta printers here, they work for 3D printers specifically because a delta movement is one that can't handle loads but is very good at fast, accurate, positioning.

It sounds amazing in theory, but the truth is that most dual extruder set-ups are useless in practice. Just pause the print, and change the filament. If there is a misalignment between nozzles; it will ruin your print.If the plastic drips out of one of the nozzles; it will ruin your print. If one of the nozzles jam; it will ruin your print. Or if the added weight of the extra extruder messes with your mechanical movement. It will ruin your print. Dissolvable support is nice in theory, but it's a huge mess and the results are questionable. You’re better off buying good software with the extruder money you saved than you are buying that extra extruder.

The extruder is the magic that makes your printer. To buy a cheap, poorly made knock-off, and then expect good printing operation, is baffling to me. Again, there's no clever hack here. An extruder is well made or it isn't. We’re down to physics again. There haven't been many huge innovations in automated lathe capabilities. It will cost about the same to make it in the US as it will China. So, it's very unlikely that the import extruder had anywhere to cut cost other than material quality and precision. Just buy a name-brand e3D v6 or j-head, or whatever is well-made and has a quality inspection step.

Seriously, the knock-offs are so bad they crushed the spirit of the designer of the J-head. He doesn't want to design it anymore. For those of you who bought the knock-offs, go sit in a corner and think about what you’ve done.

Repeat after me, "I cannot hack physics. Physics doesn't think I’m clever. It works or it doesn't. It's made well, or it isn't". Engineers aren't insane, they aren't out to hurt your feelings. They aren't hoarding secrets so they can charge money for no reason. All these things have been scientifically tested to be true or not. It's not a system you can game, only one you can compensate for.

Every cent you don't spend on good parts is time you spend compensating for those cheap parts. For example. I spent two hours trying to figure out why my printer was skipping steps every few layers. It ends up the cheap pulleys I bought were at fault. The manufacturer saved some money by injection molding the pulley teeth, and pressing those on an aluminum core. No surprise, the plastic teeth broke. I ordered more expensive, all-aluminum, pulleys from a reputable source, and haven't had problems since. I saved myself maybe three dollars by buying those plastic pulleys. I ended up spending three hours and an additional ten dollars to fix one pulley. Seems pretty silly to me.

Lastly, don't buy a printer with a cheap z-axis. Does it look rigid? Does it look stable? Does it look like the most expensive movement on the machine. If the answer is no, skip it. It's only second to extruder for importance. My printer has two home-depot threaded rods for the Z, likewise, my prints always look terrible in the Z. There's no way to fix them without upgrading.

Do we have to repeat the mantra again? Just like with the extruder. You are, at the beginning and end of the day, out to extrude some plastic. So why race to the bottom to buy the cheapest filament you can possibly buy? Find a reputable vendor that makes filament in your country with real engineering specs and buy that and only that. Try

out a few brands, but don't go buying the cheapest stuff you can. You will fail parts, or jam your extruder, or have a bad time.

Things that typically indicate good quality filament is: An error of out-of-round that's less than 4% and an size tolerance that's less than 4%. Or approximately +-.04 mm on 1.75mm filament. Filament that mentions quality inspection, laser micrometers, and other expensive things thrown in the manufacturing process. Also, if the company can name the source of their stock pellets, point to a datasheet, or give engineering specs, that is a really good sign. You don't want recycled plastic for your filament.

Also, the cheapest filament is usually black, this is because you can grind up any color plastic and dye it black. Buying cheap black filament is a great way to get a literal rock or bit of grass in your extruder. Quality black filament will be virgin plastic.

A 3D printer must position a nozzle accurately. It also needs to do this at a reasonable speed. So it must take the mass of a printer nozzle, ramp it up to a speed, and then ramp it down to a completely opposite speed. The force doesn't vanish, it is transferred to the belt, to the pulleys, and down to the table or frame. The more rigid the machine is the less it will flex when doing this. It will also vibrate less, which will show up on your prints as less ringing. This is done by having heavy materials that don't bend in the parts that see force. If your printer has an aluminum extrusion frame, but the manufacturer cheaped out and 3D printed the brackets that hold it together rather than opting for the cast ones, your printer will still have some flex to it.

Also, a ridiculously tall reinforced axis will wobble when the printer is moving. Cantilevers are also bad. It will show up in your print. I like the prusa i3 and the printr bots, but they have a max speed. This is a mechanical nightmare of rigidity and alignment.

This is something that is very hard to do in most printers. You must be able to square each axis to the other. Or, in the terms of what this means practically for your printer, if you print a large cube, every side should be perfectly square. There should be no parallelograms.

Most printers aren't designed to be squared. How to square a machine is something for another day, but for now I’d recommend watching some videos on YouTube of machinists squaring their machines to get a feel for it. It's an art, and most real machines are designed to compensate for it. That's why a knee mill is more-or-less in the configuration of the ordinate axis drawn for a 3D plot. It's really intuitively easy to figure out how to tram it. However, once you do something like making a gantry mill, it becomes more difficult. For example, what happens if the whole frame twists a little. You’ll be perfectly square at one end, and out at another.

I will mention that this is a little more of a problem for Cartesian machines, a delta machine can compensate for this sort of misalignment a little more easily, as long as you have the rails parallel to each other and perpendicular to the base of the robot. There are some alignment subtleties for these as well.

As mentioned previously, it is really important that your bed be level. Even after squaring your machine, you may still get some misalignment over time in your bed. Software bed leveling will adjust for these small misalignments really well and make printing much more pleasant. You still want your bed to be within 0.5mm of the nozzle, but you no longer have to spend hours getting it within .01mm.

The machine needs a brain. It would be nice if this brain was reliable, documented, well designed, and supported. For example, what if a wire breaks in your stepper motor cable while the machine is operating. (This oddly specific example happened to me.) If the board has a protection diode, nothing much. However, if the driver chip is sitting there unprotected because a company decided to save 25cents a board by neglecting that part, you are going to lose a driver chip as well. You want to remove as many unknowns from your printer adventure as you can. The electronics are the most complicated part. It may be good to spend money on them.

Also, for a delta printer, a more powerful board like the smoothie-board will give you better acceleration than the weaker Arduino based ones. The math is much harder computationally for the delta machines.

Some people will disagree with me on this one because of their software ethos. But buying better software for my printer has improved my printer more than most of my hardware upgrades. We’re extruding a nonlinear fluid really fast through a tiny hole, and it has all sorts of weird physical properties. The better designed our software is, the better our print will be. My preference is simplify3D. I’ve heard that some of the other specialty solutions exceed it in some areas. Do research and do what's best for you. That being said. Slic3r is a wonderful piece of software, and I really admire the work that goes into it.

If you’re a beginner, you’ve likely been told that you can buy a 3D printer for 300 dollars and that it will be easy by some guy on his eighth printer. You likely have visions of printing that PipBoy Model. Well, you can spend 300 dollars, and it will be a while before you can eek out a good quality print. Like anything, spending a reasonable amount of money well, will net you the best results. Don't just buy the most expensive printer you can find either; you’ll end up with a Makerbot and have an even worse time. Do your research, check the reviews, and check the parts people are printing.

You will get some things with a higher quality kit, that will make your time much better. Namely, documentation and support.

Here is the documentation for the official Prusa i3 kit from Prusa Research. It's really good. It will really help when you’re putting together your first printer to have no obvious questions. Most of the support question on the #reprap IRC channel are from people who bought cheap kits that are behaving oddly, or can't tell which way is up with a strange part. Good documentation is always an indicator of good engineer and good management in a company. It's the least fun part, but one of the most valuable in the engineering process.

Buying that Shenzen Duplicator off AliExpress is signing a release form from any risk on the company's part regarding problems you may have with the product. What do you do when the electronics have a chip soldered on backwards? What do you do when you get a bent precision rod? A company that ties their name to their product needs to please their customer.

For example, I know Ultimachine will replace a Rambo if it breaks, almost no questions asked. They will email you back with any questions you may have about the board, and provide tailored support. Can you say that for any of the suppliers on AliExpress? Same for e3D. If you have an issue with their nozzle, they will support you. This costs them money. It's in their best interest then, to make products that don't fail. To spend the money where it needs to go. This has been my experience, and that of others. That's why Lulzbot, Prusa, and SeeMeCNC all use these solutions rather than rolling their own.

Lastly, when buying and building a complicated machine like a 3D printer, you will almost certainly have troubles. So, you want to reduce the number of unknowns as to where that trouble is coming from. For example. If you have a stepper motor that isn't turning. If you bought a Rambo board from Ultimachine, you can more or less assume that it's not the board at fault. That makes it easier to determine that it's probably a connector, a wire, a mis-wiring, or the motor itself. Now if you bought the five dollar motor off ebay, then it's maybe the motor. However, if you bought a Kysan motor with a brand attatched, you can probably get rid of that possibility too.

It's the same as checking whether a computer is actually plugged in before you begin looking for the harder problems. With a cheap kit, that has no support, bad mechanical design, no documentaiton, and no traceable components; there's just no way to debug a problem with it, other than testing for every single possibility.

There's nothing mysterious happening in a 3D printer. However, there are limits to what you can get out of one. These limits are set by real physical properties. To harden a rod takes time, energy, experienced people, and maintained machinery. This all costs money. However, if this step is skipped, there will be real, measurable consequences. Consequences that are very well understood by those working in the field of mechanical engineering. So you might have a few good prints out of your Shenzen special. Then the hardened balls in the linear bearing may eat grooves in the un-hardened rods and jam from the resulting metal dust. That's the reality. That's why a decent printer costs 600-2000 US Dollars.

So do proper research, dispel the illusions, read the reviews, and buy a good printer. Hopefully you’ll have a good experience, and start improving your machine. Maybe, if we’re lucky, you’ll feed the discoveries you make back into the community, and we can all build even better machines.