I enjoyed this article, but perhaps not for the reasons the author hoped. This quote, "With some tweaks to my G code and use of a nozzle that is more suited for higher resolution prints I am confident that I could produce a part that would be very close to the quality of the Stratasys produced part." really summed it up. It sums to "I can tweak my setup and get this part to be good enough relative to the expensive machine." and that is great. What is missing is it is every friggin' part. Seriously, all of the $2,000 printers have quirks that hit based on part geometry, especially in ABS. Part too long? need to add hold down tabs to prevent warping. Too many overhangs? extra supports. Too long a tool travel? Up the hotend temperature so that the extruded material doesn't cool down to much.
You can make great parts but every machine is different, and every part you make has a slightly different requirements. "Real" manufacturing tools are all about the repeatability and setup. Pretty much any machining center can take a gcode file, look at it, and adjust itself to make the best possible rendition, or warn you if it can't do so and meet its advertised tolerances. A hobby machine you print it, you tweak the parameters, you print it again, maybe you tweak the gcode a bit, and finally you print a really nice part.
What gets you repeatability? measuring. There are so many things that you can measure and hobby machines don't. Who puts glass slides (accurate to a tenth of a micron) providing position feedback on the x, y, and z stages on a hobby machine? Nobody, the slides and readers cost $1K just by themselves. There has been some motion toward monitoring the feed of the filament to give a "stripped filament" warning, but who measures the actual flow? the temperature of the melt chamber and feed path?
I really enjoy my 3D printer, I'm printing out a damper for a meat smoker as I write this, its going to be glorious, but its also the 3rd time I've printed this particular part to tune it up. We'll see how it works on the smoker.
As the article says, if you want precision and repeatability, you don't use a FDM in the first place, you will mechanize the piece.
As an engineer I use my FDM machines in order to detect "bugs" in my design, easily and cheaply. I don't need to cut a block of aluminum that volumetrically encloses the piece I want, unless I am certain the design is perfect.
The great advantage of FDM is the ability to print with support material anything, not just 2.5 D designs like normal milling. 5D milling is way more expensive and complex.
tenths of a micron in FDM? You are lucky if you get real 5 micron accuracy in a FDM printer.
I print most of my designs 2x or 3x oversized to really understand the prototypes. In the end I will use SLA machines lost wax and machining to finish the process.
i haven't ever used or owned a 3d printer and this part
>This quote, "With some tweaks to my G code and use of a nozzle that is more suited for higher resolution prints I am confident that I could produce a part that would be very close to the quality of the Stratasys produced part." really summed it up. It sums to "I can tweak my setup and get this part to be good enough relative to the expensive machine." and that is great. What is missing is it is every friggin' part.
is sooo true regardless of discipline.
/rant on
sure, you can make vim/emacs/sublime or macos/linux/windows behave just like you want, but you have to do this on every box you ever touch (disclaimer: i use vim). as i get older, the more i want my tools to just work, without tweaking for hours that requires deep knowledge about the system i'm customizing, to the point of willing to pay money to get the customizing part done for me. in this case, the $200k machine most likely isn't used by a single person who knows everything about its quirks, so it's all the more important for it to work correctly without tweaking, on the first try. i wish more software was like that.
There are tools out there that address your specific complaint re: software repeatability. A combination of Ansible/Emacs/Git dot files can yield a consistent environment across machines. Howard Abrams has some great material on how he does this with Emacs (howardism.org).
I applaud the effort, but this is not the best way to save the few hours it takes me to get a new machine up and running that I have to take every year or so.
What happens if your build script breaks? You have to work around it then troubleshoot the problem. Already there goes your time savings.
I keep my dotfiles in a repo, that seems to be at the appropriate level of abstraction. Easy to set up, easy to add to, I can understand everything its doing, if it breaks I can localize the problem and fix it exactly where it's happening.
Trying to expand the scope to include arbitrary package management? I don't think the tooling is there yet, that will make this sort of thing very painful. The dotfile convention has been brewing for decades.
Trying this will inevitably force you to make compromises, and compromises defeat the whole point. The work it does is fundamentally equivalent to a package manager, and existing package managers haven't managed to get it done yet.
If every Mac developer can be convinced to use a package manager, and every package manager can be scripted, (looking at you, Mac App Store) and every app with a modular architecture also either plugs into an existing system or implements its own package manager, then this becomes viable. But you still need a modular and easy-to-manage way to manage app configuration too. If you can get them all to use dotfiles, awesome. But they don't now.
Right now, invariably you're reduced to a pile of hacky workarounds. Any automation will have been done at the wrong level and will break constantly.
The last time I needed to set one of these up, I started to write down a checklist of things to do. I gave up after I realized that I was just pulling the checklist from memory, it already exists and writing it down wasn't going to do much good.
Sounds like your job involves making small amount of similar text manipulations on a large number of computers. Which is different than what the tools you mentioned were design for - i.e. for doing lots of different jobs on a small number of computers.
As a dev, I use two computers +/- a few shells for most of my work. Little enough to manage configurations by hand. I'm more like a machine shop from GP's comment - machine shops do jobs to the same standards, but internally, each one is likely to be unique in its setup, tooling and process.
What you're ranting about is lack of tools suited for the kind of job you do. My needs are different, so my complaints will be opposite - I'd like tools to be more like Emacs; capable of being molded and integrated to better fit any given job I have to perform.
actually it's not that bad - i've got 3 boxen that i use regularly and that's it. the problem is that those machines get EOLed every once in a while (say, 2-3 years) and each time i get a replacement i dread about some config setting that i set to solve a problem or whatever and then forget about it. it's a problem that happens often enough to be annoying but also seldom enough to not be worth automating, because software evolves, etc. i've got a few most important dotfiles on dropbox.
"It sums to "I can tweak my setup and get this part to be good enough relative to the expensive machine." and that is great. What is missing is it is every friggin' part. Seriously, all of the $2,000 printers have quirks that hit based on part geometry, especially in ABS. Part too long? need to add hold down tabs to prevent warping. Too many overhangs? extra supports. Too long a tool travel? Up the hotend temperature so that the extruded material doesn't cool down to much."
No, this is like the crusty machinist who knows just how to wiggle his hips and shake his elbow to get a 1/2 thou on the worn-out bridgeport in the corner.
FDM should let engineers make prototypes without bothering that guy (whose time is probably 3x more valuable than theirs, even if his pay doesn't reflect it). This sounds like the printer is just a more compact version of the worn-out mill.
Plus you have to account for the printer's quirks right in your 3d model which is annoying and error-prone. (Yes, I know every package has a way to segregate those changes, etc. but still.)
The RepRap project and the derived printers tend to share the opposite philosophy - the original idea was to print parts for more printers, so naturally parts were designed to be 3D printed. The RepRap logo is a compromise design for part holes without support (no longer necessary, though).
Most RepRap owners ('hobbyists') are using the 3D printer as the end manufacturing method, not just for prototypes, so this is a pretty reasonable tradeoff. Note that the original article fell into this category - the 3D printed part was not a prototype, but a jig used in production. (looking at that part, it could have been designed to not require any support material at all)
Welcome to the wonderful world of quality control :)
Making something is easy, making something with a high repeat accuracy and very low fail percentage is super hard.
I learned this the hard way myself with an IO expander for the Atari ST that I designed. I needed it to drive the lathe/mill controller, the ST simply had too little IO capacity so with a bunch of decoders and latches I was able to drive a much larger number of motors and relays.
The first one worked like a charm. So did the second and the third. But the fourth, we couldn't get it to work no matter what. Part variability led to some signal arriving just that much earlier causing the latches to remember data that was still stale. The reason was - in retrospect - quite simple, the design had assumed that the input to the latches had long stabilized before the latch pulse was sent. But because on the ST the latch pulse took a different route as well there could occur a tricky situation where the output of the buffer driving the data bus on the far side would still be rising. To make things worse, this would not happen repeatably, only with certain patterns of 1's and 0's. In the end, a single buffer in line with the latch pulse caused enough delay to stabilize the input to the latches and that solved the problem.
(Of course not with every board...).
Anyway, that was electronics, this is mechanics but the same principles apply.
Something funny is happening in youtube land, classical music.
There are a whole pile of 'youtube wonders' that can play an enormous variety of material but that have never in their lives done a live concert. It took a long time before one of those miracle people owed up to the recipe: record hundreds, or even thousands of times. Pick the best.
That's the difference between an actual pianist and youtube miracle workers: a pianist can do what they do repeatably, youtube miracle workers would end up being booed out of actual performances.
Repeat accuracy is very hard, mechanics, electronics, playing the piano, the subject hardly matters.
I've been reading all kinds of books and articles about quality control last year and I believe that in a mature market it is the only absolutely critical element in any production process. If all your competitors are doing good QC and you don't you're dead. Everything else can be fixed but without repeat accuracy you'll be fighting a lost battle.
Another way to look at this is that for real world stuff where there is a whole company dependent on the output of the 3D printer the cheap one is too expensive, and the expensive one probably is cheap.
That's because if you were allowed to make a million parts with each and you could take the best part from each set of parts they'd be very close. The 'best' part of those two might even be from the crappy machine. But if you needed just 50000 parts that would simply work your cheap printer (or even a small army of them) would wear out long before completing the order, and that's what makes them expensive.
Partly this was an illusion created by the combination of affordable
video equipment and the Internet; young jugglers now kept their
cameras running all the time, so if they hit a trick one time out of
100, they could upload the proof and make themselves look like gods,
even if they’d never be able to execute the trick onstage, like
Gatto could.
when I teach innovation classes we talk about the difference in mindset between making 1, making 10, and making 100 of something (and obviously this would scale to 1000+ but that is kind of beyond a lot of students). Until you have experiences with scaling manufacturing you have no clue.
I used to do consulting work on this (my background is in manufacturing). Typically I would come into a company at the prototype stage and work with them to redesign the prototype to make it manufacturable...I loved doing it but wow does that mindset have to change about what can be accomplished. There was an article on HN a while back about why you can't manufacture like Apple that I really should have saved, it makes this point beautifully.
This is a key component in why hardware start-ups are hard. Proof of concept: dead easy, as long as physics does not make your product impossible. Prototype: a little harder. 0-series: already quite hard (that's roughly what the story above was about). Mass manufacturing: super hard.
It's also why manufacturing samples should never be taken as indicative of what a plant will actually put out and you still need to sample the actual product to determine if quality is acceptable or not.
Lots of companies doing business with third world producers have found this out the hard way. (And then there is malicious substitution as well to content with.)
And requires skill sets that seem really similar but are not actually all that similar. They are linked through engineering design but yeah...they aren't.
"The Stratasys printed part was printed with ABS material, 100% infill with a standard layer height on normal detail settings. Not having direct access to the machine I was not able to extract any real nitty gritty details on the overall print set up. For the Lulzbot print that I created I used HIPS material with a 0.5mm nozzle, 0.1mm layer height, 25% infill, 4 top and bottom solid layers, 45 mm/s print speed with standard acceleration, 240C extrusion temperature, 110C bed temperature, and an extrusion width of 0.6mm."
This is an apple to oranges comparison. I work with a Lulzbot 5, a Lulzbot Mini, a Stratasys dimension elite, and two polyprinters on a daily basis and have been working with 3D printers for over 10 years as part of my job. Why was a material like HIPS ever used for comparison with ABS if both machines can print with ABS? Why was the extrusion width 0.6mm on a 0.5mm nozzle? Why is he comparing a 100% fill print with a 25% fill print? Why didn't he mention that the Lulzbot Taz 5 he was using has a $500 print head upgrade to let it print soluble support material as well? This whole article just reeks of either someone who is either very new and inexperienced to Hobby 3D printers or was intentionally trying to make the TAZ 5 look bad. I have done similar comparison studies myself and have come to drastically different conclusions. We sold the dimension elite a while ago.
I'm not familiar with the tech but having read the article I feel he was very positive about the TAZ 5 and so doubt he was going out of his way to make it look bad.
Hi, I wrote this article. To answer your suspicions.
1) Realistically they are within an hour of each other regardless of infill so the total print time was negligible in this experiment. A few changes on both machines could make either or print much faster.
2) Yes, it was my intention to have the same layer heights and overall print settings as close to each other as I could.
From the article: "One defect I observed on the Lulzbot printed part was the blade portion of the tool had a split in it from not bonding correctly. This was most likely a cause of pushing the 0.5mm nozzle to its limits in terms of layer height but a little glue and some pressure would fix this split no problem."
Er, that's called a "reject" in manufacturing.
This is the basic problem with low-end filament-type 3D printers - weak layer bonding. You're welding a hot thing to a cold thing, which never works well. And there's considerable thermal expansion, and thus shrinkage, in the material, which produces internal stresses between layers. That's where things crack.
"This is the basic problem with low-end filament-type 3D printers - weak layer bonding. You're welding a hot thing to a cold thing, which never works well."
Bonding always works well for me and many others. The author of this story choose to use a material (HIPS) which really should still be considered experimental for most FDM printers. I've personally never had a lot of luck with it and instead choose to use either PLA or ABS with a heated chamber & with those two materials can build very strong parts with no boding problems whatsoever. Here's an example of a violin I've recently printed:
Note that I've now printed five violins and in the 100's of hours it has taken to do so not had a single failed print even when printing thin (typically .40 mm) shells.
And there's considerable thermal expansion, and thus shrinkage, in the material
That crack is clearly due to cooling and shrinkage pulling the layers apart at the edges. (The effect is more pronounced at the ends. Large parts generally curl upwards at the edges, like cooked pepperoni, just less so.) The fact the author immediately doesn't know this marks him as something of a noob. Also, he should infill the crack with extruded melt from the print head, not just clamp and glue it. Otherwise, he will be loading the part with bending stresses. (This is also a noob level lesson.)
This is an unfair comparison.. Having used a Stratasys Objet of around a similar price point, the amount of detail you can get is incredible -- down to fractions of a millimeter.
Sure, for the prints that were pictured in the comparison, either can do relatively well, but the point of higher end printers is for higher resolution prints, different materials, shapes of the print, etc. To ignore all of those use cases and focus on a generic print that a hobbyist would make, of course a hobbyist printer would suffice!
Exactly - this is a pretty simple part, without fine details that would actually test an expensive printer.
Its like comparing how Facebook or Word works on a netbook vs a top-end desktop.
And I'd say furthermore, although not knowing how the part is used - with such flat features and 90-degree corners, probably the best way to make this would be sheet metal, which with a waterjet cutter, press break, and spot welder could be produced in 1/100th the time and cost.
No mention at all of the strength/rigidity/hardness of the jigs. The professionally printed one was made from ABS, which is a harder plastic than HIPS. It is also possible that the professional machine bonds the plastic filament to the workpiece better. The best way to test this would be destructively.
Of course, if the jig is just going to be used for a short period of time, structural integrity doesn't matter. And the hobbyist machines are just going to get better as technology improves.
Also missing from the discussion are tolerances. If the Stratasys is able to print meeting high significantly higher design tolerances, that alone could justify the price tag. Certainly in other areas of manufacturing, adding zeros to the accuracy/repeatability of a system often adds zeros directly to the price tag.
If you can get a print hot off the printer and immediately say "LGTM" without further ado, that may just mean your application isn't very demanding.
Hi, I wrote the article. I am really enjoying the comments everyone has contributed. One major detail I left out was the amount of failed parts that came out of the professional machine prior to getting a good one. I was unaware of the reject rate at the time I wrote this. In addition, the production tool ended up having resin added and then sanded to cover up some of the imperfections on the "professional" print.
What was the reject rate? Also, some of us have asked some other questions about tolerances and material properties in other comments on here; could you answer them? I'm sorry about the somewhat intemperate comments of some of the other commenters.
I can't give you an exact reject rate but it went through at least 3 iterations before being acceptable. As for tolerances and material properties (overall strength of the part) the Stratasys machine is superior in both aspects. As I mentioned in the conclusion of the article, I would like to do more experiments on the parts including dimensional analysis and possibly a tensile or compression test. Having 1 data point (1 tool from each machine) wouldn't be the best results for an accurate comparison but I would still be interested in the results.
A few years ago in grad school I spent about a month designing a 3D printed case for a Raspberry Pi + Camera [0]. We used a Makergear M2 (a $2000 printer at the time).
The process had some frustrating moments, but overall it was an amazing experience. 3D printers allow for a wonderful iterative design process. Come in in the morning, pull parts off of the printer, measure, test fit, etc. Tweak in CAD, or start working on the next part, and start printing your changes/additions before lunch. By mid afternoon the print is done, and you usually have time to do another round of changes that you can start printing before you leave for the day.
Of course, this is all assuming that your printer is working continuously and that you're ok tossing a lot of filament from those test prints. My experience with the M2 was that typically I could get it dialed in and it would print great for 2-3 weeks before it started misbehaving. Sometimes it was a simple adjustment (re-level the bed, etc.), but most of the time there was no obvious indication of the problem, and I'd spend several frustrating days leveling, calibrating, cleaning, and generally tinkering until it started working consistently once again.
There are also plenty of issues with the print quality and precision, but I've already written more than I intended, so I'll skip them. Needless to say, it was sufficient for our needs. IIRC, our final cost per case was $4 in material (PLA) with a print time of about 6 hours.
It's a fascinating technology, and it's mind blowing how far it's come in the last decade. I'm not sure that the FFF/FDM style printers will ever get to the point where they're ready for use by the general public, but the overall technology is here to stay. I wouldn't be surprised to see 3D printers become as common as microwaves in my lifetime.
I have never seen a 3D printer, although I have seen a model (of a house plan) made with a 3D printer. You folks who live near civilization and 'make' communities are so lucky! :)
Are there any 3D printers that integrate a milling machine?
I imagine a machine that prints the model slightly oversize, milling and then perhaps finally warming the surface to polish it, all as it goes...
I would love to have access to a hobby 3D printer as I have this idea that won't die about making a cylindrical version of the Enigma http://wiki.franklinheath.co.uk/index.php/Enigma/Paper_Enigm... with glowire or alternatively tracks for ball bearings, as a cheap present people can buy at, say, Bletchley Park gift shop. Silly thing to get fixated on.
possible, but it probably is more a "mill with a 3D-print attachment" in terms of construction and price. A mill has to have higher stability, stronger motors, ... than a pure printer, driving the price up.
There are combo machines but they are almost always worse than having two machines, and then moving the part from one device tothe other. Then, machining is an issue- your CAM/slicer need to work together.
> Are there any 3D printers that integrate a milling machine?
No, and there likely never will be. Why would you wait for hours to then still have to mill down your final product when all the advantages of 3D printing are then lost and all the advantages of milling are lost?
The advantages of 3D printing are that as an additive process it will work in many environments where a subtractive process would not, the advantages of milling are that - as a substractive process - it can make a final result that is substantially stronger than most of the things you could do with a 3D printer and that it most likely is much faster.
Re-working or finishing like that is typically reserved for casting and other cheap mass production processes.
"By combining both, additive manufacturing via powder nozzle and the traditional cutting method in one machine, totally new applications and geometries are possible."
Don't people often just use acetone or another solvent to smooth the surface of chunky prints (outdoors, I hope)? How effective that would be would vary depending on the plastic, but it seems easier than milling (if not especially precise).
Wow... I have never thought of the process in the linguistic terms "additive" and "subtractive". That actually changes the mental model for me. I have always thought, in less abstract terms, about removing shapes (carving out) and optionally sticking external parts (bolting on) to get the desired product. The specific add and subtract "language" presents it in familiar, abstract, and, more importantly, composable model.
If his conclusion were true companies like Shapeways would only use cheap hobby printers. But ofcourse it is not.
It's far from a perfect analogy, but with servers Google demonstrated really well that the cost of high-end 'professional' equipment can be a waste of money where cheap, commodity kit can be used at scale instead. It's not completely unreasonable to think that a farm of cheap 3D printers could be used in place of one high-end model eventually.
People are freaking expensive. I think that is something that you miss easily.
200k sounds expensive for a printer, yet that is the cost of 1 guy for three years (assuming western country).
The 200k machine just has to be a small bit better than the 2500 machine for it to pay off so fast it is ridiculous. A commodity approach to printing sounds like it would involve a lot of quality checks and other really pricy manual things that would eat up that 200k in under a year.
If you can remove all human costs, then yes, commodity gear is likely better, because you are paying for gear that knows that humans are expensive, so even small savings in humans translate to a ton of cost, and so the professional gear is priced accordingly.
The 200k machine just has to be a small bit better than the 2500 machine for it to pay off so fast it is ridiculous. A commodity approach to printing sounds like it would involve a lot of quality checks and other really pricy manual things that would eat up that 200k in under a year.
Owning the 200k machine has to be better than owning 80 of the cheaper machines for it to be better. In a lot of scenarios that will be the case - if you need the improved resolution, reliability, etc that the 200k machine offers then the cheaper machines just aren't an option. In time though, as the cheaper machines improve, the expensive machine may stop being a better option.
> Owning the 200k machine has to be better than owning 80 of the cheaper machines
No, his point was that with the cheaper printer you might need to hire one guy to look after the quirks of the 2500 machine and customize each part so that it works fine with the cheaper machine. If the 200k lasts more than 3 years, it's already paid for itself, because after 3 years the cheaper machine would cost you 2500 for the printer and 200k for the guy.
the difference in 3d printers and servers is rather hard to compare: different servers can compute the same result, faster, slower, but it'll be the same. not exactly true with 3d printers, where a result isn't binary.
Servers aren't actually binary, either. Your cheap server might serve the same result (binary), but performance might degrade under load more rapidly (nonbinary), the failure rate might be be higher (nonbinary), and the support for cheap hardware might be nonexistent (nonbinary). There's also the issue of SKU drift/mix as your cheap hardware fails and gets replaced with a different/newer SKU (adds maintenance cost, is also nonbinary).
At the end of the day, the cost difference plus the ability to mitigate/hide the problems via software makes the commodity hardware the winning option, but it's not just a matter of "same bits come out; good enough".
Commodity hardware is also not really cheap. These are 1U/2U servers that cost thousands of dollars. These are cheap relative to mainframes, but they're not consumer devices.
In fact, the $2000 Lulzbot printer used in the article is made of lots of 3D printed parts, manufactured on a farm of Lulzbot printers: https://www.youtube.com/watch?v=v_jUObUGLTA
The difference here is reliability. Commodity kit is generally reliable, just slower; whereas cheap 3d printers are more likely to fail a print/produce junk.
Meanwhile the assembly line and workers expensively wait for the assembly jig to be printed. And the contract due date doesn't automagically move out a day because you had a printing problem, so now you're paying (more?) overtime once you finally get a working jig.
There is a solution to that, when your labor and capital costs are huge but the cheap printer is 1/10th the cost of the fancy printer... simply install and use 3, 4, maybe 5 cheap printers, keep the best print, and pocket the savings.
Yeah, but 3D printers are also really slow. The part in the article took 18 hours (on both printers) and in my lab we've had similar experiences.
Not sure where the break-even point is, but the cost of a failed print is pretty high in terms of sunk time for most prints. On an industrial scale, a machine that prints correctly 99% of the time might be a better investment than two that print say 75% of the time.
The professional printer printed out almost 4x the amount of material in the same time (100% infill vs 25%). If it had the same settings for hollow infill, it likely would have been much, much faster.
The low-cost part is as good as the high-cost part ... except in all of these ways that it's not, that I will outline here. Oh, and I basically just eyeballed them, instead of measuring them. Oh, when pieces were missing due to failure to print, I just kind of assume that they would have printed okay with some tweaks.
I read this whole article hoping to find the part where he tests the tensile strength, impact resistance, stiffness, and dimensional accuracy of the two parts. It never came. Except for the part where he had a delamination ("split"), which I guess means zero strength for that part of the part.
I've been fairly disappointed with the strength of Prusa-Mendel-printed parts I've made. Even if you don't have complete delaminations, it's easy to get poor layer adhesion, which makes the parts fall apart under the least strain. You can fight that by turning up the temperature and filament feed rate, at the cost of dimensional accuracy (which you can compensate for in theory but I haven't done so successfully, being just an amateur) but the higher heat also weakens the PLA. Maybe this weakening is hydrolysis due to water absorption in the filament; I don't have a way to measure that, except printing the same thing a second time after dehydrating the filament, which I haven't done.
I was surprised to find that 100% infill is not always the strongest option, even holding part geometry constant. Lower infill settings produce a more compliant part, and it can consequently withstand higher impact energies.
I don't think you should hold part geometry constant when changing machines or especially entire fabrication technologies. You should play to the strengths of the machine you have. I know that's contrary to the outlook of 3D printing, but it's true. Things like topology optimization can easily generate forms that are impossible to injection-mold, relatively easy to FDM, and much stronger. If you have the option of 25% infill, you can probably get a stronger part by deploying the plastic you saved as ribs outside the main body. And so on.
It would have been nicer to have seen a comparison using the same material as well as a series of parts designed to test for specific issues in 3D printing.
Add to this printing 100 copies of each part and we might have ended up with a useful set of data points.
3D printing, especially hobbyist machines, is still in its infancy and produces disappointing results. A large flat piece warps. The resolution is coarse. It takes too long to produce even a simple piece. The machine needs constant monitoring. It's at the equivalent of the DOS stage of personal computing.
I wrote code to produce STL files. These files are accepted by hobbyist printers. I sent it to a professional shop, they complained one of my triangles wasn't closed, couldn't fix it or tell me which it was. Wonderful.
No contest. Having multiple hundreds of thousands of colors, properties (like heat-resistance, flexibility, different opacity), various different materials, automated easily removable supports, decent precision, and serious speed beats the shit out of your desktop model any day of the week. Sorry.
I'd like to see a similar post ten years from now, see how things have evolved. I didn't realise how precise some of these industrial 3D printers were! This gives me a lot of hope as to what I might have in my house in a decade.
What's also important when comparing 3D printers is maintenance cost and effort. For example: The Ultimaker 2 we have in our Hackerspace needs regular replacement (about every 600 printing hours) of the PTFE Coupler (hot end part). Also sometimes you need to perform a cold pull (https://ultimaker.com/en/resources/19510-how-to-apply-atomic...) if the nozzle is clogged.
Surely the high gloss finish on the black material is drawing attention to a lot of the flaws as well? I'd really like to see a same colour comparison between the two.
I see cheap 3D printers like I saw linux in 1995: not quite the functional equivalent of a $100K SGI workstation, but still a great way to get experience in the area. You just have to understand that much of your time will be spent tuning the printer, and fixing it, and upgrading it. I don't know that the Statasys fails nearly as often as a cheap printer.
3D Printing is really great for industrial designers who want to prototype products. My uncle is an industrial designer and I remember how costly it used to be for him to come up with a simple prototype. Today a $2500 printer can get one pretty far in a few hours, trully revolutionary. It gives a greater margin for experimentation, trial and error.
Using different material and different infill is a bit of an unfair comparison. Using 100% infill causes more problems with warping due to shrinkage, thus increasing the difficulty. Using 25% avoids this, but decreases stability a bit.
Regarding the print quality in general, I think an Ultimaker 2+ would give you better results (also at a price of 2500$).
Be aware that the high end printers include features that are patented, for example soluble support, heated build chambers, etc, all have IP around them that may prevent companies selling inexpensive printers from including those features directly. This could translate to better results from the high end printers.
This seems like an interesting hobby. How much time is needed to achieve a reasonably high understanding of this work? Not industrial grade expert, but a knowledgeable layman. And where/how can I educate myself?
The best thing you can do is to go to a makerspace near were you live. You will learn so much just by osmosis with the people that know just being around and looking.
Then you need a project that you want to do and ask those guys how you can get there. You will need to learn some CAD, some mathematical concept and you will get something.
Then you print your pieces using their printers. If at this point you enjoy it, you will want to have your own 3d printer. The guys will help you with this too.
For example I am part of Clone Wars RepRap, an Spanish language group. We are more than 5000 people. We help each other a lot. I am also part of German,and Austrian groups.
Without the social environment, buying a printer yourself you will probably find it too hard, specially at first.
I'm not even an engineer, and I have made one. But, making a 3D printer by yourself from ground up (or assembling one with the parts bought from china etc.) takes a lot of time. I only paid attention at my spare times, so it took my months. But I must say that, it is a very satisfying hobby. I strongly recommend it :)
But if you want to buy one, you can find very nice 3D printers at $250 - $300 range from aliexpress etc. It may take hours, a day or a week for calibration (depending on your printer model and printed part quality expectations), and you are good to go.
But I must say that, 3D printers need maintenance and that can take your time as well.
If you want to educate yourself, http://reprap.org/ is the site you would want to visit. Not only for making your own printer, but calibrating and maintaining a bought 3D printer as well...
As a start you can just go to Thingiverse to search for a model that fills your need and send it to the printer. That would take you all of a couple of minutes. Per try, until it comes out ok.
If you want to try your hand at 3d modelling then you'll probably download Sketchup as a start and noodle around while watching youtube tutorials on the side. In about 8h you'll feel confident enough to actually model something meaningful. After ~40h you'll be able to knock together a model quite proficiently.
After you'll have gotten tired of Skethub producing broken STLs and not allowing you to change your mind after modelling a thing, you'd move on to one of the real CAD software package, like Solidworks or Fusion 360 (which is great for beginners, btw). That rabbithole goes exactly as deep as you've got time for :)
Fiddling with the printing itself will take a bit of setup time before each print and then a lengthy wait (hours to days) to get the result, or see it fail in some new and unfunny way. It's not really all that difficult, to be honest, just takes some time to develop an intuition about what's likely to fail.
The downside is they can't work with as wide a range of materials compared to FDM printers, and the materials they can work with tend to be more expensive.
You can make great parts but every machine is different, and every part you make has a slightly different requirements. "Real" manufacturing tools are all about the repeatability and setup. Pretty much any machining center can take a gcode file, look at it, and adjust itself to make the best possible rendition, or warn you if it can't do so and meet its advertised tolerances. A hobby machine you print it, you tweak the parameters, you print it again, maybe you tweak the gcode a bit, and finally you print a really nice part.
What gets you repeatability? measuring. There are so many things that you can measure and hobby machines don't. Who puts glass slides (accurate to a tenth of a micron) providing position feedback on the x, y, and z stages on a hobby machine? Nobody, the slides and readers cost $1K just by themselves. There has been some motion toward monitoring the feed of the filament to give a "stripped filament" warning, but who measures the actual flow? the temperature of the melt chamber and feed path?
I really enjoy my 3D printer, I'm printing out a damper for a meat smoker as I write this, its going to be glorious, but its also the 3rd time I've printed this particular part to tune it up. We'll see how it works on the smoker.