We Have A Problem: 3D Printers Are Too Expensive

Hackaday, we have a problem. 3D printing is changing the world but it’s still too expensive to be embraced as a truly transformative technology.

With each passing year, the 3D printing industry grows by leaps and bounds. Food safe PLA is now the norm, with dissolvable and other exotic filaments becoming more mainstream.  New filaments are making it possible to print objects that were not possible before. New CAD software is popping up like dandelions, with each iteration giving novice users a friendly and more intuitive interface to design 3D models. As time marches on, and we look into its future, a vision of the 3D printing world is evident – its only going to get bigger.

3d printerImagine a future where a 3D printer is as common as an ink jet printer in homes all across the world.  A future where you could buy filament from the supermarket down the street, and pick up a new printer from any hardware store. A future where dishwashers, refrigerators and bicycles come with .stl files that allow you to print upgrades or spare parts. A future where companies compete to give the market easy-to-use printers at the cheapest price.

Is this future possible? Not until the technology changes. It’s too expensive, and that’s the problem you’re going to solve. How can you make a 3D printer cheaper? A cheap printer could change the game and make our future a reality.

Where do we need cost savings?

To get you going, here are some parts of common 3D Printers which think need to find cost-saving solutions.

XYZ AND HOT END MOTORS

Stepper motors are going to run you about $15 each. Is it possible to use cheaper DC motors with some type of position tracking while keeping the cost down?

HARDWARE

Threaded rod is probably the cheapest way to move your XYZ axis. What about couplings and guide rods? Check out how this guy made a CNC out of parts from his local hardware store.

ELECTRONICS

No arduino with Easysteppers here – too expensive. We’ve just seen a super cheap controller a few days ago. If we use something other than NEMA steppers, it will radically change the typical electronic controller for our super cheap 3d printer.

EXTRUDER

What is the cheapest way to melt and extrude plastic? What about using thermistors in place of thermocouples? Let’s think out of the box with this, and see if we can get away from the typical stepper motor based extruder. Remember, everything is low cost. If we have to sacrifice some resolution, that is OK.

So there you go. Let’s hear your input on the issue. We need to make 3D printers a lot more affordable and we want to hear any ideas you have on the topic in the comments below. Do you think this is in our future and why?


The 2015 Hackaday Prize is sponsored by:

219 thoughts on “We Have A Problem: 3D Printers Are Too Expensive

  1. Too expensive? I’ve paid more than the cost of a 3d printer for a ‘normal’ printer many times.

    Paper printers are only as cheap as they are because they’re sold on the razor blade model. I don’t see 3d filament being quite as ‘keyable’ as ink cartridges…

        1. Just purchased my first, one that didn’t have the DRM’d crap in the machine or carts. Rejected ALL the above that deal in that blackmail (IMO).

          IMO, NO ONE should buy / support those who sell that crap. Those that do get their own rewards :-).

      1. So true. Been there, cracked that (at least I tried..) I did rewrite the “fresh” EEPROM content, but the printer itself keeps track on which cartridges were spent. So much no no for tagged overpriced filament boxen.

        1. Hm, how hard is it to hack the printer’s own firmware? Presuming they’re using a normal microcontroller somewhere. Even if you had to replace it from scratch, there’s already firmware available for all the other 3D printers, that’s designed to be modified for specific printer details. If the printer’s using a common microcontroller, can’t be too hard. Maybe even as simple as changing a couple of bytes in it’s storage EEPROM.

          Yes that’ll harm a business that’s trying to bring down the initial purchase cost of a printer. But also to make more profit by needlessly locking people in. Same reason Microsoft keep changing the standards of their Office every time. Nobody’s needed a new version of Word for it’s features since the early 90s. Companies, the big money, are forced to buy it for compatibility with documents other companies will send them. If people would just standardise and choose one of the other formats to save in, which Office supports, there’d be no problem. But people are literally too lazy and stupid to choose from a dropdown, thus billions of dollars.

          So what I’m saying is, fuck ’em, for choosing the greedy monopoly business model. Incidentally the thing about drug dealers giving away free samples is a myth. There’s plenty of people who get addicted by themselves, who’ll pay good money for low-quality drugs. Prices are the same for everyone. That’s right, heroin and crack dealers are more ethical, purer capitalists, than Microsoft.

          1. microcontrollers can be fused such that the code from them cannot be read.

            at least not by usual programming methods….

            i you are willing to go to such lengths to hack it, you are willing to build your own, and then you dont care about hacking some commercial application.

          2. Yeah I thought that, hence I mentioned the free firmware that’s available for 3D printers in general. Since they’re fairly standard 3D printers, shouldn’t be impossible to tweak some existing software in.

            You can still rewrite a microcontroller’s firmware though right? It’s people stealing the software that the MCU’s manufacturers were worrying about. Worse comes to worse, just replace the chip with a new one. Comes to even worse than that, just stick in a new driver board. Shouldn’t be too expensive compared to what you save, depending on how much filament you tend to go through. Whether it’s cheaper than just buying a different printer, who knows? But destroying the DRM is a good idea in general, it means you can compare it to other models evenly. Although for the amount of trouble involved you might not want to bother.

            Still, you’d be a hero to some if you solved the problem. You’d also piss off the manufacturer. Which might encourage others not to bother with this stupid pricing model. Or they might just double down and try and make it even harder to hack. But since a microcontroller just isn’t that expensive, it wouldn’t be easy to do. After all the mechanics and motors are going to be like anybody else’s, driving them is a well-solved problem, and still an active interest for some people. It’d also mean you can add in whatever features you like, and give you more control, if you wanted it.

        1. I wonder if it’s a reference to the FTDI debacle. I’m inclined to agree that most users probably weren’t affected. If your machine was built around an Arduino that uses an FTDI clone, it could have been impacted by the driver update, but that doesn’t seem very common. Most DIY 3D printers are based on Arduino Mega 2560 or clones. The R3 uses a 16u2 chip, R2 uses an 8u2, neither are FTDI.

          1. James, as long as the existing driver works, and the hardware’s OK, should it matter? Are you saying there are drivers that actually stop older hardware from working? If there is, I’m amazed but also not really surprised! There’s business for you. If it increases profits, they’ll have to do it. No choice, cos one of their competitors will. Then once they have, everyone else has to follow through. And they can’t just live with making a bit less profit, share holders will start dropping them. Especially if a computer is making the decision, nobody wants to invest in the second-biggest profit.

            And that, competition and share prices, are why capitalism forces companies to do whatever it takes to make the most money. As long as either it’s legal, or their chance of getting away with it makes it a worthwhile risk. Companies will never become ethical of their own accord, if it costs them potential profit, if it’s not the most efficient move. They’re unable to. The only thing that works is using the law, which is (supposed to be!) the will of the people, to force them to act in the public interest. Ethical capitalism is bullshit, it’s impossible, and it doesn’t take a genius to realise that.

            To return to the point, briefly, there’s often websites that keep hold of older versions of drivers. Usually it’s not the butthole ones that try and charge you, that expect you to pay for something they didn’t make themselves. The amount of desperados on the web would be funny if it wasn’t so depressing to think that *somebody* is buying their “service”.

          2. I got my $10 perfectly good scanner from Goodwin because hp stopped supporting that printer beyond WinXP. So it is very real for OS upgrade obsoleting hardware especially if the hardware isn’t open sourced or the vendor doesn’t care about its customers.

          3. Ah wait, you mean Microsoft (for it is they!) refusing to let old drivers work on Windows, right? Scanning the web seems to say that it’s a system setting you can disable. Maybe a virtual machine’s the answer, or a USB installation of Linux running Wine? I realise that’s a bit desperate, although OTOH I wouldn’t expect a company to write new drivers for long-obsolete hardware. They shouldn’t get rid of the old stuff, but software costs money.

            I’d expect better service for some photo-calibrated expensive printer, than something that cost 30 quid, but still.

          4. “I’d expect better service for some photo-calibrated expensive printer, than something that cost 30 quid, but still.”

            HP is notorious for intentionally breaking older stuff, no matter how expensive it is. I had a very expensive color laser that stopped working with Windows 8. It prints, but the scaling/rasterizing engine is working horribly wrong and all the color prints come out with diagonal banding.

            HP’s response “we’re working on it”… and that was two years ago.

          5. As for disable device drivers signing, that’s a dumb way of doing it. No amount of fiddling with settings would let you run x86 driver in a x64 windows environment. A VM let you run x86 code without compromising security for the rest of your system nor kept you from updating the host OS.

    1. I have to agree. What I see more likely in the near future is the local shop that has professional-grade 3D printers and has a library of stl files of parts. When the washing machine breaks, the average person will go to the local shop and have them print a repair part.

      1. 3D printing a spare part of something proprietary is no different from copying a DVD. If the industry sees the latter as theft, they will see the former as well.
        They’re not jumping at 3Dprinters users neck like they weren’t during the audio taping period because the technology isn’t there yet. Give some time to the technology to mature and after some changes (for worst) to the copyright laws tons of lawsuits will follow.

        1. I think your “copying a DVD” analogy isn’t quite right. I can’t buy 3rd party produced DVDs (at least from anyone reputable). But I can purchase 3rd party replacement parts for my car, washing machine, vacuum cleaner, etc. That is an industry already, and one I can see expanding with the use of 3D printers.

          A shop purchases a 3D printer, and subscribes to a repository of STL designs for replacement parts. Manufacturers could even submit STLs directly and bypass the need to manufacture and distribute parts. It could be a win-win for both OEM and hobby/repair/maker shops.

          1. Then how would the original manufacturers get a cut (just forget they won’t want it) if they let the designs in the wild? Any attempt to monetize each time a spare part is produced will inflate the price and create the market for “illegal” copies.

          2. @qwerty
            Once again: why can I buy aftermarket parts compatible with my car? Until you answer this, any reasoning along the lines of “but how will they monetize” is begging the question.

          3. broken plastic parts, like tail lights, handles, knobs etc come to mind as car parts and other parts that can be made my 3DP, both by people and the major manufacturers. Now thay are made and stored until wanted = lots of wasted space and time. They usually do not need the great strength of metal

            Metal parts tend to cost more when 3DP is used, so metal car parts will be hard to make with 3DP.
            Still, things are new, perhaps one day we will have multi element 3DP and we will be able to print memory and hard drives etc – currently impossible, but who know what 50 years will do?

    2. Or you could order it from a company that manufactures spare parts using traditional plastic manufacturing techniques which is astronomically cheaper, faster, and better made like you can today. The only time I could think that you’d want 3d printed parts would be if you had some obscure appliance which spare parts aren’t available for.

      1. Transport for small parts(0-150g) from china has a fixed cost of 0.5- 2dollars, not to mention the travel time and packaging. If you have a 3d printer at home it costs less, you get it faster and you can even customise it.

      2. You would want to print parts that are too expensive to purchase. Those parts are either for equipment so new third party replacements aren’t yet available, or they are parts where no third party replacements exist, due to the part being specifically designed so that it is unique to one particular manufacturer/series/model. The “proprietary” parts in other words. Those are the ones you won’t be able to download stl files for.

        1. And right there along with 3D printers are 3D scanners. If a part isnt available yet you pop it into the scanner and make your own .stl file. Maybe even tweak the design to reinforce the failure.

          Things are changing- either legal strong-arming to keep things as they are or business models change to adapt to this brave new future.

      3. “Traditional” plastic manufacturing techniques, predominately injection molding, are VERY expensive. Thousands or tens of thousands of $$$ to produce the molds, pennies to produce each molded part. It’s only economical if you are producing many units of the part.

      4. Exactly. I don’t have the money to buy a pro-grade 3D printer along with all the supplies. I don’t have the space for it in my condo either. Most of these tools have a pretty good learning curve if you want great results and I’m only willing to become an expert in so many fields really. I do the part picking/design, schematic and PCB (in Altium), I write all the code (and documentation, bug tracking, source control, etc) and the basic 3D design. I also build the prototypes (both with cheap kapton stencils and solder paste dispenser) which often have small pitch and QFN parts (iron, pre-heat and hot air mainly). I also figure out how to test them (jigs, procedures, code, etc).

        I prefer to let someone else worry about how to manufacture the necessary parts themselves — be it traditional injection molded parts, CNC machined parts, sheet metal bending, laser cutting acrylic, etc. All this specialized equipment is really costly, takes up space and requires a lot of effort to turn out great results time after time. Most of the time, it’s not a 3D printer that will produce the parts I need either (metal cases, acrylic, etc).

      5. And yes, for a small part the cost amortizes out to a few cents. Plus $15-25k for mold design and retooling fees.

        If your product has enough volume, you can afford the tooling fees. If you can’t, then 3d printing it is. 3d printing also allows to have a library of parts damn near close to infinity, where molds are heavy pieces of steel or aluminum and are very heavy and bulky.

      6. Yes, we know that a rear plastic tail light housing that weight 3 pounds, costs less than $10 to make by GM. Yet if you break one, GM will charge you $300 or more for it as a scarce sole sourced part. A 3DP could do this a loy cheaper

    1. how on earth did I not see that before now…I’m guessing these folks are indeed using optical encoders and DC motors, given the hints on their kickstarter. Even if they’re only right in 50% of their claims, I think the tiko is def a landmark in cheap 3d printer production.

          1. yea, saw what the replaced them with, too bad. I was hoping for linear optical encoders and DC motors, which would have been significant progress, and reuse industry expertise (most 2d printers use these)

        1. Read their Reddit AMA, it was quite a spectacle to watch them get jab after jab from the Reddit community. The problem is that they’re using cheap non-NEMA stepper motors with plastic gear drives on them, that have been known to wear out after limited usage, and are about as slow as a clock motor. So their printing speed is going to be low and the MTBF is going to be pretty low as well.

      1. I was thinking optics and a registration transparent ribbon (with 600 dpi vertical lines etched or printed on it) like many old HP printers used for keeping the head properly registered with each pass. I haven’t torn one apart in a while to see if they still do this, but the motors in those things were never very beefy and seemed to do a very good job, even with high print rates like draft mode.

      2. kickstarter 3d printers have an awful history of under/never performing. The print speed on that will be terrible, and delta printers are notorious for being difficult to trouble shoot. Also not having a heated build plate will limit your filament options to basically pla. I allmost backed the tiko then ended up getting a prusa i3. 300 bucks gets you a decent 3d printer today that is upgradable to be an outstanding 3d printer.

        It is a nice price point though.

        1. I have a printer without heat bed, and I printed a lot of stuff besides PLA. Flexible filaments, bronzefill, woodfill, PET, nylon, even small parts of ABS.
          The Tiko is so small, that ABS doesn’t even have enough space to warp :-)

        2. $300 isn’t bad, but that’s more than I spent on a TV. I’m not saying we need a 3D printer in the range of the RasPi, but we need that type of difference in price that the RasPi is compared to an average computer.

          Why? Not everyone has disposable income. The RasPi is the price of a hardback book, meaning libraries can loan them. A printer in the 1/100th to 1/10th price range would encourage the same thing; a printer that is cheap enough that loaning time on it is not something only a hacker space can manage. Because, let’s face it, lots of rural areas don’t have hackerspaces (other than people’s garages) but do have people who could make use of 3D printers.

          On my fixed income, the low price models don’t have the capability and that high price models are out of my range. My dad almost bought one, when we thought that making some knife handles that would be less painful to use with the various nerve damage I have, until he tried to ventilate his garage and found it too difficult. Lots of little “make things better” projects are in mind if we can find someone with a printer; the bigger projects are some custom game pieces to then cast in aluminium.

          1. Failing at ventilating the garage interfered with getting a 3D printer?

            Dryer duct and a PC case fan should pretty much solve any air-quality concerns.

      3. Using threaded rod is great for a slow moving axis such as the Z axis is Cartesian printers. It would be WAYYYYY to slow for the rapidly moving x and y axis. Would you want to wait 48 hours for a 20 mm calibration cube to print? Belts are cheap to buy or source from household appliances.

          1. If you’re concerned about the thread pitch, the simple solution is to turn the problem on its head.

            Instead of turning the rod, use it as a rack & pinion system.

            Take a 6 mm metric rod. It will have 1 mm thread pitch. Then turn a 10x10mm cylinder out of aluminium and cut a 5 mm groove 3 mm deep in the middle. Then drill a hole through to match your motor shaft.

            Now you have a roller. Run it along the threaded rod and notches appear. Press it down hard and the beginnings of teeth form. Take a 6 mm thread tap and rock it sideways in the groove to cut the teeth out to depth.

            With these dimensions, it should produce 31.4 teeth, which is not an integer number, so rubbing the roller against the rod over many revolutions wears it down in diameter until there are exactly 31 teeth – then the gear starts to mesh and stops wearing out. At this point, you introduce the thread tap and use it to carve it out properly.

            Then you have a pinion and a matching rack with very little backlash, that goes 31 mm per revolution.

            The smallest gear you can make depends on your motor axle. You need to leave a bit of skin between the center of the groove and the axle, or the pinion will snap in two, unless you leave the axle hole blind. Then the minimum number of teeth for a 6 mm rod is about 25 if you want the pinion to grab at the sides of the rod. Less if you make the groove shallower.

          2. Furthermore, you can use two hall-sensors to sense the thread pitch of the rod in a quadrature configuration to find out your relative motion along the rod, so you can eliminate the gear backlash in software.

    2. Disclaimer. I work for Ultimaker. (Also known as the Cura guy, developer of one of the main Open Source 3D printing solutions)

      So. I know what parts cost at bulk. I know what assembly costs. I know what material costs from a supplier. I know what support&shipping costs. I know our machines are expensive.

      En result. Tiko will go the same way as the MakiBox. Most likely not shipping even half of them and then going out of business.
      http://www.fabbaloo.com/blog/2014/10/30/makibox-is-done

      1. Are you able to expand on the topic for the benefit of 1000 whiny internet commentards? I know facts & informed comment are unpopular but it would be great to hear the low-down (even in general terms) from someone who has been there & done it.

        1. We use more expensive parts (after all, our cheapest option is way more expensive). And A we have more margin, yes. After all, they do want to pay me for open-source software development (something which most of these cheap printers just leech off)
          We’re also making sure that we’re not supporting any child labor.

          But, just stripping our machine down the the BOM, we are on more then these machines. For an Ultimaker, you cannot get the parts for $180. But an Ultimaker is a bit of a more complex beast. So. Let’s do different semi-math.

          For $180. I could most likely get. Quality electronics. Motors. A single roll of material and a half decent hotend. Buying in bulk, and accounting for a small failure rate.

          What you lack is, assembly, shipping, casing, calibration, sales&support staff, finance, software-dev, lost shipments.

          (My contract prevents me from talking exact numbers. Sorry)

          1. there are 2 external interrupts documented at http://www.arduino.cc/en/Reference/AttachInterrupt

            if you want to go into the pin change interrupts supported by the atmega, you are beyond what your hobby arduidiotino tinkerer can do. and the rest of pin change interrupts are not as versatile as the 2 external interrupts.

            i can read a manual, and i can use it, but i specifically referd to the “arduino” which is not the “atmega 328p”, it is the entire of the ecosystem, and what it supports.

        1. The problem is laziness and/or the lack of skill set think outside of “Arduino”. Can’t get there if you confine your solutions.

          If the AVR chip can’t service all the interrupts fast enough, go bigger. Lots of chips out there that have better peripherals/performance for a cheaper price. (e.g. ARM uC)

          You know a small CPLD/FPGA or the PSoC can make peripherals for these things. If the Arduino framework fails you, use something else. No one making consumer 2D printers resort to buying Arduino boards and over priced breakout boards.

      1. So what? Even if Arduino can’t hack it, someone’s building a DC servomotor printer using an ARM Teensy. Once it matures, switching controller boards would be easy and pretty inexpensive.

      2. Take a look at TI’s InstaSpin technology. Essentially, it turns almost any kind of motor into a servo and does a lot of the hard motor positioning work for you without encoders. I think anyone inclined to build a 3D printer from scratch would be able to figure this technology out.

        http://www.ti.com/tool/instaspin-bldc

        I saw a demo of of this motor control technology at the Embedded Systems Conference in San Jose last year. A TI rep was running a conveyor belt with a small DC motor without an encoder or any other sensors. He jammed the conveyor which caused the motors spindle to stop spinning. When he released the belt, the motor spun faster until the belt reached the position it would have been in had he not jammed the belt. I think the development board was measuring back-EMF from the motor to precisely track the position of the shaft.

        You may be able to use much less expensive motor DC motors without the need for encoders, while still providing precision motion because the system tracks rotation, and speed, while also allowing faster speeds and higher torques than a given motor’s datasheet’s ratings call out .

        The development board’s cost may be a deal-breaker depending on the definition of expensive is for a project.

        Any thoughts?

    1. Cost performance reasons is my feeling.

      A DC motor with an encoder that is powerful as a ‘average nema 17’ stepper would be a complex specialized device once you include a motor driver for it, tuning, etc.

      Cheap $100 inkjet printers use razor blade model as mentioned elsewhere – they sell around cost and make it back on ink. This also has the advantage of encouraging users to upgrade to new printers constantly, as the cost is down the road at all times.

      Meanwhile a DC motor with an encoder requires mechanical and electrical concerns that most 3D printers simply cannot take on.

      Stepper drivers do take current into account.

      Also, reading encoders straight from the main MCU in a motor controller via interrupts is not really how it is always done. There is also special purpose chips for that, or the option of multiple MCUs on the motor driver board.

          1. You can slow it down in code or speed up as needed… Just need to code a distance to PWM fqctoring…

            After looking at it again i forgot to say motor for extruder is missing… So might cost a bit more than That…

            I like the hot glue gun extruder idea but dunno how you would add a motor to that setup easily…
            Maybe design a servo to pump the glue gun and have a hole restriction for filament size…

            Dollarstore glue gun… With 4 servos and a contoller maybe $61 USD before shipping?

  2. The problems isn’t the cost, it’s that 3D printing is a huge pain in the bum.

    2D printers became ubiquitous because you could make a document, press print, and have a physical copy. Super simple. With 3D, we’re not even close.

    Also, with a 2D printer, it’s very very easy to make a document. For making a 3D object, there’s quite a learning curve. If the only way to make a 2D document were to manually code it in LaTeX, then we probably wouldn’t have seen as many 2D printers either.

    Solve these problems and even Grandma will be printing trinkets.

    1. Specifically, These are the issues that I can think of off-hand. Feel free to add your own:

      1. Getting stuff to stick reliably. I know people have their recipes for success, but they really only work for a specific filament. Can you imagine the Epson Stylus Pro manual saying “If you want to print black and white, then spray hairspray into the printer. If you want to print graphics, then stick a piece of blue painters tape over the front of the printer and cover it with a towel for 20 minutes before printing.”

      2. Smarter slicing software. There are SO many settings that it takes quite some time to become proficient. Also, you often need to think about how to slice an object during the design phase. Do you need your walls to be a particular width? Then you need your design and your slicer settings to match, or the walls will be printed wrong. Can you imagine telling your mum that the reason her birthday invitation got printed wrong is because she didn’t have the printer DPI set in exact multiples of her font size?

      3. Smarter machines. Is your filament moving around before it cools to below Tg? Stick a fan on the carriage! But it’s not that simple. Some parts of the print (e.g. small infill sections) will get loads of fan time, whilst perimeter corners might have the carriage whizzing past without enough cooling time to ensure the corners remain sharp. So what do you do? Edit your slicer settings to slow the perimeters down, re-slice, and have another crack at it. Imagine your having to re-print your resume because you had the carriage speed too high and the serifs came out wonky!

      4. Easy (consumer easy) 3D modelling software. This one is hard because the nature of 3D is trickier than 2D, but we can still draw parallels. Word doesn’t require you draw the entire page structure, it has a button for margins and one for style. If we were able to create a content-based 3D modelling package then this would be much easier for people to grasp. E.g. “Bowl ->Serving->Art Deco->30cm->Blue”

      1. LOL #1.

        Good points – I think mostly these fall under product maturation. Think back to before there were PnP devices. Getting a printer to actually print way back in the old days was something akin to this (minus the hairspray, but I do recall needing tape on occasion). One way to handle a lot of these issues is to simplify the software and consumable side by finding just the right standard hardware, or by publishing and refining a really solid standard hardware specification (like Amiga One did) to which economies of scale can be applied to bring the cost down or the really good parts and then the collective focus on software can ensue to make a 3D printer OS that handles all of the nitty gritty you outline as a matter of course. Not sure about Aquanet as a consumable but we can work on it.

        Maybe the right question to ask, HackaDay, is “which hardware is best?” Find the best in class and promote that so that supply and demand can start driving the price down. Product manufacturers tend to feel pressure to reduce quality to meet demand – if you start with published standards for quality, and challeng suppliers to compete for the role of “official supplier”, you might have the beginning of a relationship that brings costs on best-in-class parts down.

        1. I concur with [AD Homined] – create a “best in class,” especially for those (such as me) that haven’t bought one yet. By pulling all the Hackaday experience into this “one review” it will help those who want to make the right choice in their first 3D printer purchase. And, as mentioned above, hopefully help to drive down the price for future 3D printer enthusiasts.

      2. While I agree 100% with what you’re saying, I still think that my success rate with 3D printing is in the same ballpark as with inkjet printers. I have terrible luck with paper jams or paper just not being picked up or genuine ink cartridges just not working or being detected or lasting anywhere near as many prints they should..

      3. ” Imagine your having to re-print your resume because you had the carriage speed too high and the serifs came out wonky!”

        There speaks a man who’s never used multiple-pass on a dot matrix…

        1. Try bold, double-strike, underline on a Diablo 630 (connected to a Xerox 820-II running CP/M and WordStar). That took 8 strikes per character for headings. Looked nice though.

      4. Don’t even get me started on bi-directional vs monodirectional printing, rich black vs true black, Raster processors, RGB to CMYK conversion, dot gain, using the wrong side of single sided paper (i’m not even talking about photo paper there), screen fonts vs postscript, expired paper and a boatload of other things I don’t even remember from my Graphic Communication Technology course work.
        To put it simply even inkjet printing is harder than you think it is it just doesn’t fail as catastrophically as delaminating in your hands or glooping all over the print bed. People still go to professional printers when quality matters, businesses cards and wedding invitations are the easy examples.
        So to close up this analogy eventually Grandma may be able to print off a passable new selector knob for her washing machine but she’s not going to be printing a new nylon bearing for it.

      5. Word doesn’t require you draw the entire page structure, it has a button for margins and one for style. If we were able to create a content-based 3D modelling package then this would be much easier for people to grasp.

        Content-based? W3 has been trying for that for what, a decade? And yet people still insist on using bold instead of strong (5 letters shorter), or italic versus emphasis (1 letter shorter).

        Remembering “way back in the day”, there was a GREAT digital format that made printing painless, but almost no one used it, so it never caught on. Does anyone even remember postScript (.ps)? The problem with PostScript was that it was a proprietary format owned by Adobe, and used by Apple for the LaserWriter. It was too expensive to implement in lower cost printers, so it never became a standard. (Unfortunately, Adobe learned it’s lesson, and then gave the world the abortion known as the Portable Document Format (.pdf)).

        1. It wasn’t the language that was proprietary, it was the compute effort needed to interpret it. PS was basically open, but for the typeface hints section that allowed Adobe to create nicer looking fonts, but the manuals were available and the language was well documented.

          PS is still the basis for PDF, just with the general purpose language elements removed – no loops or variables. The graphics commands are still in place.

          If they had found a way to make a good interpreter that didn’t need as powerful a CPU as the computer it was attached to already had, it would have been OK. Of course, if MS hadn’t wanted to ensure there were no competitors, it would have used Display Postscript instead of the Windows graphics API.

    2. “The problems isn’t the cost, it’s that 3D printing is a huge pain in the bum.”

      I agree with this but not with your reason. Most of what I have printed has come straight off of Thingiverse. No designing necessary!

      But.. getting all the settings just right.. getting it to stick… the printing process itself is still rather difficult. When that gets easy like it is with paper more people will want to own printers. Then maybe mass production can help with the price issue.

    3. Its also a quality issue.. Ive spent around 20K (me and my clients) on 3D prints in the last 3 years. All on high quality Objet and SLA prints.

      The quality is so much superior to FDM, it makes it look like something made of play-dough tools.

    4. OH! Also dont forget how difficult it is to make 3D designs.

      Ive worked as a engineering designer with bigish 3D design teams – even design engineers struggle to express themselves in 3D design tools they are experienced in – expensive tools.. Heartbreakingly expensive tools.

      Someone might make a opensource 3D package that can make shapes of complexity, and document them, let alone allow real 3D engineering workflows – but when? How?

      SketchUP is the example of what happens when you remove complexity from 3D modeling. You are limited to simple shapes.

        1. I’ve had the task of fixing some output from sketchup, removing doubled polygons and edges, “welding” edges and vertexes that are just sort of “leaned against” each other, deleting extraneous internal edges in what are supposed to be solid or “water tight” meshes and more various problems.

          Blender is free. Caligari trueSpace can still be downloaded, here’s an active support forum http://www.united3dartists.com/ IIRC there’s some free version of 3D Studio? There are plenty of free 3D modeling programs, many of them can export to STL or other formats suitable for slicing, with varying levels of success.

          One thing several of them have are rather non-standard interfaces, usually due to a past history that stretches back to Amiga, old Macintosh or Linux. Rather than changing everything to conform to Windows or OS X GUI standards they just folded, spindled and mutilated until they had a reasonable facsimile of the software’s old GUI. Lightwave is another one that started on Amiga, though isn’t free.

          One thing to note about trueSpace, it has the Lightworks render engine and before Microsoft bought it and made it freeware, it sold for less than the Lightworks plugin for Lightwave.

          1. Haha, well you got me there. Sketchup does do a lot of that.. But I have plugins that can sense it and autofix or at least tell me where the issue is. Also over the years I’ve developed habits that prevent those from forming most of the time. It’s not an idea solution but..

            I’ll check those out though, they look interesting, thank you!

      1. Try AutoDesk’s Fusion 360. I think it’s more “powerful” than SketchUP and it’s free for hobbyists and for small companies that make less than a certain amount (The amount is more than $100,000).

  3. dc motors with encoders, while cheaper, are a whole lot more inconvenient to use.

    the development cost in terms off time is huge, but manegeable for developing a product (almost torturing, for a single-shot project). It took me days to calibrate the PID controller for my project which does exactly that – uses a dc motor with encoder instead of a stepper:

    http://andreicociuba.ro/2015/04/30/self-typing-typewriter-1/
    (ok, i admit, i could have measured and tuned the controller numerically, but that would take even MORE time, and only worth doing if you are doing a production run, not even remotely worth for a single item)

    the advantage here is increased average velocity at the expense of path accuracy.

    It is easy to make a motor get to the precise position. it is HARD to make it go along a specified PATH. in control engineering terms, exact positioning is reffered to as Stationary Error, and the path would be dependent on what engineers call Rampp error, or Velocity error, which is nearly impossible to be brought to 0, instead designed to be fixed, predictable and compensated for elsewhere.

    it is even harder to SYNCHRONISE two of them (one per axis).

    I haven’t tried it out, but i immagine that splitting a curve in sufficiently small linear segments might take care of that, but simultaneously, it might cause jagged movement, and vibration. (its all thought experiment here, and certainly worth trying out, and i WILL do that when i have suficient time to spare)

    And then the various problems will be dependent on the speciffic path, which is, by definition unknown at design time…

    and the more i think of it, the more complicated it gets, and the more worthwhile trying.

    1. Yeah, plus using brushless DC motors will not save you a penny. And using cheap brushed DC motors instead of cheap steppers will inevideably shorten the lifespan of the printer crowned by that last failed print.

      1. I guess you meant brushed DC motors. Brushless is a very different kettle of fish, I don’t think I have ever seen a printer use those (the controller would be quite crazy). And they are also hard to classify as “DC” – they are basically 3 phase motors.

          1. The problem’s usually precision of movement. That said, the brushless gimbals on drones are pretty sharp, thanks to the accelerometer feedback, and they’re ungeared. using BLDC motors for a printer is something I’ve been considering doing; While control’d be a nuisance, they’re smooth and fast, and fairly cheap. I’m sure there’s got to be a way.

          2. you essentially have nested control loops. the controller uses an analog/pwm input for setting the speed and does its best to keep it steady, and you give it the required input to achieve positioning precision, in your own PID loop from your microcontroller

          3. The controllers you are thinking of work using so-called “sensorless feedback”. This works great for rapidly spinning motors, and can be done pretty cheaply, but it fails at low speeds. Even worse for 3D printers, the algorithm to start spinning the motor may do all kinds of undesirable things, like briefly spinning backwards. You pretty much need both a feedback device and a controller capable of using that to commutate a 3-phase motor.

            Plus side: there are cheap chinese incremental encoders out there for ~$15 – search for “incremental encoder” on ebay. Someone with the right know-how could implement a complete feedback loop in an FPGA based on this – if you’re especially clever, you could run all 4 motors with a single FPGA and an Arduino to support it. I estimate the total cost for doing this is in the range of $30 per motor and $80 for the controller and power stage.

            Unfortunately, I get paid to design motor drives, so I can’t exactly write an open source controller…

    2. The problems you mentioned have already been solved. First one is motion planning. Simply put when you from point A to point B you accelerate until you reach your desired speed and then decelerate (at the correct time) to finally stop at point B. If you control not only the position but also velocity (and even acceleration) you will have highly accurate travel. You can run multiple control loops, i.e. an inner velocity loop and outer position loop. More advanced stuff like feedforward mechanisms are also incorporated to get tighter responses. Synchronized multiple axis moves are again handled by motion planning. You are on the right track with the dividing the path into multiple parts. You just make sure that numbers add up to the correct acc., velocity and positions at each interval.

      Most of higher end CNC machines use servo systems (which basically are motors with encoders and a control system).

      I would suggest you take a peek at picservo documents. It’s quite a good read.

      1. Most high end CNCs use servo’s because there is a certain point in power requirements where the costs cross over from steppers being cheaper to servo’s being cheaper.

        Small steppers + drivers are dirty cheap. What’s used in the current gen printers is only $5 a piece in medium bulk.

    3. “the advantage here is increased average velocity at the expense of path accuracy.”

      So simply put. You’ll be producing crappy prints. Path accuracy is SUPER important. I had a small bug in one of the Cura version (software generating toolpaths for 3D printing) and this caused inaccuracies in the paths, only tiny inaccuracies, but enough to make horrible prints for some people.

        1. There are a plurality of 3D printers that don’t have DRM though. DaVinci and 3D System’s Cube series are the only hobbyist machines that I know of that has DRM.

          On the coffee front, Keurig posted a regret for doing that. There might be others. But at least there’s hope there too.

          1. I bet Keurig regrets that it didnt completely work… Not that they did it!

            But yes, the DRM on filament in the low cost 3D printers is an anomaly, not the norm.

            However, most if not all consumer\maker 3D printers are guilty of patent infringement, so everything is a cluster-f to begin with.

          2. Can you say what patents those might be? Because I’ve never heard of such alleged rampant patent violations. I know there’s the Afinia lawsuit, but the patents I’m aware in that are rather specific and I’ve never seen anyone violate those patents, at least since then. I’ve heard that slicers removed that technique.

        2. Oh, I know. Luckily they have not seen a whole lot of traction, yet. If 3D systems gets their way, we will all be paying $180/kg for crappy filament because we have not choice. Right now the typical user is savvy enough to see through this false economy. I hope it stays this way long enough to get the gougers out of the market.

    1. Comparison ignores two things:

      1) Paper printers have only two things to move: a light-weight head (on one axis) and a PIECE OF PAPER! FDM-style 3D printers necessarily have more to move around; SLA is the path forward.

      2) It took a long time for paper printers to achieve high quality output for a low price, and there were millions of low quality output, expensive printers sold to pave the way. The demand for 3D printers will not be as high as that for paper printers for a long time (if ever), so that road paving is not a sure thing.

  4. The core way to make them cheaper is mass production. Motors are bought with bulk discount, controllers are standardised, frames and housings are injection-moulded by the thousand.

    This does not fit with the rapid design evolution and hacker mindset though. I’m inclined to say you can’t have both. It’s already extremely cheap within it’s existing market. To make it cheaper means redesigning for the consumer market. That means higher reliability, more standardisation, less production overhead.

    We might get a few long-tail economies, but super-cheap doesn’t fit the hacker market here.

    1. We already have proprietary printers and yet there are still Repraps. There’s no reason there couldn’t be a mass-produced printer market for the ‘norms’ and still open source printers for the makers. I think it’s still difficulty of use combined with not having sold the masses yet on just why they NEED a 3d printer that keeps there from being mass-produced printers.

    2. It is always a trade off – you can either have a team of extremely smart people money can buy to spend years of refinement on cost reducing a product or unorganized movement of unknown skills end users trying to copy/paste from each other and doing lots of trial and error. There is only so much the lazy path can lead to.

    3. Exactly my thoughts. it all comes down to one word: BULK.
      Selling a million identical machines gives you the power to buy, produce and ship bulk. No kickstarter will ever accomplish this. I think even stratasys/makerbot will not make this transition. both are companies making only tens to hundreds of machines per day. Making thousands or tenthousands is a really hard transition essentially reinventing a company.

  5. Maybe someone can build a supercheap printer using a cheap mass produced trashcan as housing and frame. That would save us a lot of money plus the manual labour of throwing away the junk it prints as well as manually throwing away the printer once the rest of the cheap materials have worn down. The initial idea of 3D-printed wealth for the masses is noble, but this will probably happen at the cost of sustainability, just like it happened in the world of 2D printers.

    1. Wealth is not for the masses.

      Plastic trinkets are.

      Only financial and legal reform can help the masses at this point – but they are now thoroughly propagandized by the rhetoric of the capitalists.

      1. So edgy.

        It’s terrible that an entire industry has sprung up around consumer-grade 3d printing. I pity the folks who left their day jobs to start 3d printer manufacturing companies, what with the employing people and the sharing their R&D with the open source community. Why would they do that when they could patent their innovations and troll upstarts? And the chumps who have contributed their time and money to making prosthetics for children, how horrible!

        The saddest of all are the “hackers” and “artists” sharing their designs so that anyone can reproduce them free of charge, or even worse, reproduce the very machine needed to steal the shirts from their backs.

        These atrocities could be avoided if people would stop pissing away their time and intellectual capital, become wealthy by any means necessary, and fund the People’s Angry Mob Pitchfork Factory. Then we can take down the man. Whoever that is today.

  6. DC motors optical encoders are standard fare in modern inkjet printers which have abandoned steppers long ago (except for the scanner head), but that’s a 2d system, made by companies with plenty of R&D budget, it’s not a trivial task. Steppers work just fine.

    Thermistors are already used commonly, thermocouples I imagine would be less common usage (has the author mixed the two up?), thermocouples would have an advantage for high extrusion temperatures and resilience, but thermistors are far easier to use, and commonly available, hence why they are used (typically 100k NTC).

    As for a controller, a Chinese “Arduino” is a couple of bucks, and stepstick type drivers are similarly a couple of bucks a piece. That’s more or less all there is to it when you get down to it.

    The price has some room to fall yet, but ultimately the motors and hardware have the largest cost contribution, especially the motors. You can buy from Aliexpress a kitset wooden i3 for significantly less than $200 US, plus shipping, including a kilo or two of filament and more or less all the tools you need to assemble it – I know because I have one, it works well. It’s not consumer ready, but it’s cheap for those who want a 3d printer, if they have the technical aptitude to use one, they should be able to manage assembling one – and if anybody does, they might want to read my notes on the process first: http://sparks.gogo.co.nz/prusa-i3-from-rp3d-notes.html

      1. The main requirement for a scanner drivetrain is that it move a constant mass at a constant velocity except when homing. The feedback loop should be simple. Besides, mass manufacturing drives down the cost of making a custom encoder wheel / strip and opto modules, much of which is subsidized by ink sales.

    1. The problem is that the those cheap DC motors with encoders are not that much cheaper than a $15 stepper (did you check how much do those encoder strips/wheels cost?). Also you will need a fairly beefy motor, the small motors in cheap inkjets move only a comparably lightweight head, not a heavy extruder assembly.

      What is worse, a stepper can be driven by a cheap Arduino and a stepper controller, a DC motor + encoder need a lot more CPU oomph – it needs PID for positioning and hardware support for reading the encoders. Forget about trying to read the encoder strip/wheel using pin change interrupts on an Arduino, that’s way too slow and it takes missing only one pulse to screw up your print. So a more advanced controller is required, with hardware encoder support. So what you save on motors you will pay on the controller hardware.

      It is certainly possible to do this and there have been printer mods that use these DC servomotor (because that’s what motor + encoder really is – a servo) arrangements. However, it is not going to be any cheaper, unless produced in enormous quantities, with “pre-packaged” controllers.

      I think that this race to the bottom is pointless – you can go only so far on the quality of the hardware before the machine becomes an unusable piece of crap. There is a reason why most sensible printers start at around $500 – it is hard to make a printer that is viable mechanically and economically (price of a printer is more than just sum of the part costs – someone has to build it, ship it, provide support, pay bills, etc.) for much less than that.

      Moreover the idea that everyone is going to own (and need) a 3D printer is just stupid. It is a really unhelpful fallacy that is pushing the 3D printer market into unsustainable extremes, trying to drive down the prices of the machines regardless of the basic economics. How many of the those Kickstarter and Indiegogo startups promising to deliver a cheap printer actually did – and are still around, not bankcrupting themselves in the process?

      The comparison with the $50 mass produced inkjet is a false analogy – everyone needs to deal with paperwork to some degree – bills, taxes, administrative paperwork, letters, etc. and most people know how to read and write.

      However, 99% of people don’t need nor posses the skills required to make use of a 3D printer. And downloading pre-made gizmos bought online to print them at home? Sorry folks, ain’t gonna happen. This is a wet dream of a few companies like Makerbot, but it ignores the fact that mass produced, injection molded objects are going to be always cheaper. The odd object can be always printed using a 3D printing service – often from the same website where the model was available from.

      1. it is totally possible to read at least one encoder with any arduino (they all have at least 2 external interrupts).

        that’d only be good for one axis, so either go for multiple arduinos in one setup, or go for the due which has interrupt capabilities on all pins.

        a basic PID control loop takes next to none of the CPU power, especially on an arduino due, and it’d totally be feasible.

        the problem is that your run of the mill PID controller is not optimal for pathway control. only position control. which means it’ll get to where you tell it to, but you have to find some cludge to be able to make it to go there through a specific route at a speciffic speed. (see my previous comment)

        it is possible, but then you’d need nested control loops, and the type of setup that is more than the usual “i can arduino, such electronics, much programming” person.

        it’s deffinitely possible, though i wouldnt be entirely convinced you get it to be cheaper. especially when you start factoring in your development time.

    2. 2D printers also don’t have the same kinds of loads 3D printers have. There’s very little comparable between the two kinds of machines once you factor in everything that’s going on. 2D printers have found a need such that at one time, everyone had one. And everyone probably still has one unless they’ve made the conscious effort to get rid of them rather than letting them sit in a closet. 3D printers haven’t found the killer app that makes everyone think they need one. It’s possible it might never justify its place. But 2D printers used to start at far more than what 3D printers start at now, esp. considering inflation, so there’s hope yet.

      Thermistors are dirt cheap too. It’s hard to argue they’re holding the price up when they cost so little in volume. They’re only expensive if you buy them singly. Heaters are cheap too. There’s nothing that’s really expensive these days unless you splurge for the premium parts. Thermocouples require a special amplifier that costs more than a thermistor. Thermocouples are really a premium option that’s not necessary for basic operation. It’s better for more high end use.

      The unrealistic demand for cheap is annoying. The machines have been getting cheaper, you wouldn’t notice if you’re not paying attention to the history. DIY 3D printers used to cost a lot more than this to build. Expect it to get cheaper faster than a certain rate and that risks getting a lower quality machine than if you’d waited.

  7. I am sorry but saying that something is too expensive without marking down the numbers is not sensible. So how much is too expensive? 1000? 500? 300? 200? 100?

    I like to pay for things how much they are worth instead of some form of subsidized model where the real and excessive profit is sought from selling the media, so the article is spot on on arguing how to make the actual cost of the printer come down.

  8. “A future where you could buy filament from the supermarket down the street, and pick up a new printer from any hardware store.”

    One of our supermarkets already sells one model of 3D printer. The cost was €999 for the printer and around €30+ for the filament roll. The quality of the sample prints did look quite adequate (on par with the Ultimaker based on my experience).

      1. You mean for the cost of materials? That might be the cost of the oil that goes into making the material, not counting any of the processing costs needed to make anything out of it.

  9. Well, if we think outside the box, maybe FDM is not the best way to cut cost. How about cheaper SLA? Or an entirely new technology?

    As others have pointed out, 2D stuff might not apply to 3D: i never think that my document is unprintable, the DC motor moving the cartridge around does not have to do it at constant speed or be in any precise place at a precise time, printing a document takes virtually no time…and more.

        1. First, I realize you’re being sarcastic. But I think the mechanics are accessible to more of us than the chemistry of photosensitive resins. The resin people will probably just do what they do no matter what we argue about. If they can leapfrog filament materials in cost, strength and safety, that’s great. It might be a while though.

  10. Not too expensive. Now good ones that are not finicky and a pain to get set up? Yes those are expensive. but if you want a cheapie that is not very accurate, you can get cheap ones all over the place.

  11. Well, they can give them away for free, if they force you to buy their inflated filament. (Filament has a base cost of under $2 per pound, yet I see it sold for $30 per 100 grams.
    This brings the Gouge-me-day-and-night Hewlett Packard model to mind.
    Threaded rod is fine, but slow. The faster slew speed of steppers and linear motors offsets the cost of worker wages who wait while threaded rod turns.

    Home/hobby use = OK to be slow. Any sort of production places a premium on speed. Yet, even at home, we want speed. Few people want to wait while fine threaded rod does its job, cheap as it may be.

    The expiry of patents and faster production methods is bringing the price down.

    New patented methods always save you time, and ask extra $$.

    1. I thought it was assumed that mass production was far from one of the problems 3D printers are trying to solve. Borrowing from your analogy, you wouldn’t use a $50 super-subsidized-by-ink-more-expensive-than-gold HP DeskJet to mass produce books either.

      Sadly, the state of the art in FDM relies on filament with tight diameter and shape tolerances across its length to control extrusion volume. This adds $$$ to the cost of raw bulk plastic and additives.

      That said, you’re either not looking hard enough for filament or are cherry-picking prices to make some sort of case. $30/100g is ridiculous. I can drive 20 minutes to Fry’s and pick up 1kg of passable PLA or ABS filament for under $30, or mail order better quality stuff for about the same price shipped. That’s still a big markup over raw materials costs, but few of us have the facilities or time to extrude our own filament.

  12. “A future where companies compete to give the market easy-to-use printers at the cheapest price.”

    That may be the way it was in your Econ101 textbook, but here in the real world, the primary reason a manufacturer lowers prices is to put financial stress upon their competition. If the competitor can’t match the lower cost, they lose customers. It’s like a “reverse auction”, the producer with the lowest price at market gets the customers. Now, on the surface, you might say that’s the same thing, until the competition is driven from the market. Once alone in the market, the manufacturer is free to raise prices to the highest point the market will bear.

    A manufacturer may also acquire the competition, and then as a “cost of doing business”, raise prices, see above.

    Or they can expand into other markets, and then because of the cost of doing business, again raise prices.

    1. >A manufacturer may also acquire the competition, and then as a “cost of doing business”, raise prices, see above.

      Or they can acquire the competition, lose the community, and have their stock tank because of a stupid acquiring, see stratasys and makerbot

  13. Kits for 200mm * 200mm * 180mm FDM printers are available for ~$300, including 2kg of filament, a 5-axis capable controler and drivers, an LCD screen, and SD card. That’s absurdly cheap, and even more absurdly cheap compared to just a few years ago. I think this is a problem that’s solving itself.

    1. I got one, was able to make one correct print and then the bad quality filament full of shit just clogged my print head. I was not able to unclogged it and now I need to buy a new one. Which can cost a lot where I am (Norway), including delivery price. A few parts got broken while assembling it, the cable doesn’t stay plugged because of the wrong connector so you can get a lot of problem when one of the cable from a stepper just fall when your printing. The structure is also too weak so it moves a lot and it’s a real pain in the ass to calibrate it due to these shitty spring loaded screws (and I’m missing one that was not included in the package). The LCD screen have a potentiometer that should allow you to go through the menu but doesn’t go infinite so you get stuck at one point. And I think that’s all I can complain for now. Right now my printer have been sitting, taking the dust since a few months because… Well because I spent too much time raging against it just to have failed prints. My only hope is to repurpose it in a better one.

      1. It took me about 8 hours to build my kit and dial in the calibration. After three calibration cubes and a few minutes of bed leveling, I was doing 24 hour long prints with no trouble. I was honestly pretty shocked, as I expected an experience much closer to yours.

  14. I’ve been playing with the AS5048A to see if it is accurate enough to use on the joints of a delta style printer so that you can work out the position of the end effector and then you don’t need z calibration or homing. Probably won’t be fast or accurate enough to use in a PID loop while moving, so you would still want to combine it with encoders if you want to use DC motors. If you’re going to use stepper motors you won’t be gaining very much.

    Precision-wise my initial tests using a 3d printed rotating knob shows that leaving it in one position and averaging out readings gives you about 1/100th of a degree before the digits just jump around. No idea what that says about accuracy. But each leg of a delta style printer only rotates through less than 45 degrees so that’s <4500 unique positions per joint.

    The problem I see with DC motors is that there isn't a ubiquitous standard like you get with NEMA-17 steppers. So it could be tricky to specify standard motor specs that people can reliably source in small amounts if you're hoping to make a reprap style printer. In theory if you have encoders and a proper PID controller you can use just about any geared DC motor, but as someone mentioned elsewhere in this thread it's actually really not that simple.

    My hope is that I'll find a way to combine the absolute rotary position sensors plus encoders on the motors to try and come up with a semi-automatic calibration routine..

  15. For an encoder why not have some type of printed pattern and use inards from an optical mouse to scan it. Most of these devices let you get at the actual pixels.

    1. Thats actualy what is used in many printers. You have a steppingpattern on a translucent (?) piece of tf plastic that is mated to the axis and 2 IR diods/Sensors. Its not encoding perse but atleast you can count your direction and steps.

      Get some printers of the same type for upcycling and maybe you could mate the 3 axis to be your xyz framework

    2. If you have tried it, you would know the answer why. I did and here are my findings. All mouse sensors slip, regardless of surface quality, LED or laser. And it’s not a negligible slip, like 1um per metre, it’s huge, unpredictable and fairly large, think +/-10 degrees per turn if it’s tracking a surface of 1-inch rotor, the slip in my experience does not really depend on rotor speed, or at least I could not move it slow enough for it to not slip. The best theoretical framerate one can get in camera mode would be ~60fps for ADNS-9800, in reality probably more like 40fps, which is useless for all kinds of incremental tracking. And if you hope to use it as absolute encoder by printing some readable pattern on an encoder strip, your code strip would have to be narrower than 1mm: that’s the focusing area of the default lens. Good luck printing at least 8 bits, and even if you succeed, aligning it. Yes, you can use a custom lens in theory to focus on a larger area, but we have passed the border of cheap, easy and reliable already.

      Forget optical mouse sensors. They are very good for mice, but it feels as if they were intentionally designed not to be used anywhere else.

    3. Beware that the resolution of the encoder strips of inkjet printers is actually really ugly, they use it as a speed regulation reference and software divide it into the actual pixels. The disk that does the feed is another matter.

  16. I recently bought the stuff to build a Kossel Mini and tried to trim costs buying on alibaba&co.
    Hardware (beams, screws, bearings, rails and carriages): €239.45
    Linear Motion (3 NEMA17, celt, pullyes, etc): €65.21
    HotEnd (1 NEMA17, extruder, drive gears, etc): €79.55
    Elettronics: €32.43

    It’s easy to see what keeps the price up.

  17. Buy filament from the supermarket?

    Well, we’re not far from that, a home store that’s just around the corner from my work just added 3D printers (Dremel) to their portfolio, so I guess I can also buy filament there soon.

  18. “A future where dishwashers, refrigerators and bicycles come with .stl files that allow you to print upgrades or spare parts.” Yeah, like big companies would do that. No, if any part of that is broken, throw the whole thing away and buy a new one. That way you’re spending more money, and that’s the point. When was the last time you bought a monitor or an LCD tv and got a full schematic for the boards inside?

    1. He made a lot of compromises to get there. He was at least originally promoting not using a microcontroller because they were “expensive”. He might have been thinking how Arduinos are $30, but capable microcontrollers can be had for like $0.50 or less. I don’t know if he changed that, because it’s apparent he found out how unreliable and inconsistent audio jacks are, requiring a lot of complicated math to calibrate against the problems.

      1. They did — https://www.kickstarter.com/projects/117421627/the-peachy-printer-the-first-100-3d-printer-and-sc/posts/1242784

        “Big Changes

        As mentioned in the video Peachy Printer no longer uses an audio signal to print; we’ve moved to a completely digital circuit. What does this mean for you? Easier and quicker setup & calibration, less troubleshooting, fewer cables, and a lot more features & flexibility! We are thrilled to be able to bring such an upgrade to the $100 kit, as the digital circuit approach was originally intended only for The Peachy Pro.”

  19. Well for starters you can slice the price of a 3D printer by about 50% if you don’t go for branding. Which will land you a comparable if not superior product. It seems to me that the bulk of 3D printer pricing seems to boil down to paying for a brand.

    You could probably save even more if you sourced the parts yourself but that is one heck of a daunting task for anyone who isn’t used to that sort of thing :P

    1. Brand = support, parts, etc.

      People pay for actually being able to use a tool – not have a new hobby of low-end 3d printer technician.

      The $2500 high end consumer 3d printers barely print better than the $250 consumer 3d printers.

  20. i would agree that 3d printers needs to be more precise, or print stronger prints, but cheaper?

    99% of people don’t have to have one and can use it just like those xero places in late 90’s and early 00’s

  21. I built my mendel i2 for $300 with 2kg of plastic included….
    tho,
    – The motors I used were from stripped out printers
    – I started with printed bushings, but LM8uu are 0.80ea or less
    – 8mm rod is 7.98mm 5/16″ rod is 7.95mm … need I say more?
    – I used a $40 arduino 2560 knockoff, tho they are $20 now
    – stepsticks are $1.50 ea.. I was going to make my own, at that price its pointless
    – a ramps 1.4 board was $80, so I made my own driver carrier (its just a carrier!) board. The ramps baords are less than $20 now
    – I made a hot end for less than $5, but later upgraded to a chinese one cause it was only $15

    So far my i2 has lasted longer than any inkjet I’v ever had, less jam-ups too. I paid $100 for the 2kg of plastic, and IT was still good after a year on the shelf. That all said, I suppose I didn’t have to assemble the inkjet…

  22. As regard steppers…

    When i was young during my university days, i had the assignment to transform an heavy duty DC motor into “stepper like motor”. The size of the motor provided by the teacher has a diameter around 25cm… not exactly sized for modelism. The stepping accuracy was eventually really very good at the end of process development and some cheap gearing helped a lot to additionnaly reduce step size. At that time, the motor was harnessed to a Z80 I/O rack (the one we used in school lab) with a transistorized power interface. Accurate stepping was performed using forward&brake sequence. Any motor coping with the power interface was usable : the Z80 I/O rack program was also able to record forward and brake reaction according to controlled stimulations of the motor. Data were sent to a venerable PC XT using serial line (Yes… i’m a dinosaur…) where mathematical model identification/matching of the harnessed motor was performed and the result was fed back to the I/O rack for use with the required accuracy. I believe the same could be applied about printer steppers with some microcontroller in the back-office… (note : latter we removed the Z80 I/O rack and replaced it by a logic system made of a bunch of Lattice’s PALs)

    Another idea could be to use a sonic wheel on any rotating axis. sonic wheel as using hall effect sensing of rotation using a teethed wheel where one teeth is missing to indicate zero reference. high precision milling for a geared wheel is quite cheap and may be economically good.

    theses ideas may not be so good but may be a step forward for the geniuses laying around here… (and please excuse my english language)

  23. I don’t believe 3d printers are too expensives for the masses. I believe they are too complicated. Most people here can use it but not the masses. Not yet. So far it is not possible to get a model and click on “print” and get the model straight. You need to know your machine, experiment and waste some filament before being able to print something decent. I believe this is why they are not the big revolution the not-tech-savvy was expecting.

  24. I don’t see why there is this expectation that they should be cheap and widely available… Are laser cutters? Are CNC routers? No… these are professional tools not meant for use by the average person. I think it would be far more cost effective for services to exist that will produce parts for people who need them. How often do you need to print something as a replacement part or upgrade to a device? It’s sure nice that I have the ability to, but what I bought my printer for was to function as a tool to create custom enclosures and parts for my projects. I’m not looking to replace things, I’m looking to create.

  25. Certainly no fans of DC motors here. Engineering and PID tuning are a pain in the ass but it only really has to be done once. A small DC motor, H-bridge and encoder could be had for $2, $1 and $3, plus a plastic gear train. Call it $8, and a wall wart and micro for $5… $40 for all the electronics?

    Assuming you can get a sheet metal frame done (big assumption, but w/e), and you can get the entire frame and rod assembly done for $60, you could probably make a printer for $130 including the bed and hot end, making it sellable for ~$300. There are a lot of ifs in that though.

  26. another cost driver is conspiracy whether it is using price to keep good quality printers out of reach to say prevent gun making or piracy or licensing to handle exotic chemicals that could be precursors to meth or other drug making.

    stepper motors can be gotten from junk dot matrix printers, older floppy drives and hard drives.

    even the daisy wheel printer uses a stepper motor to spin the typing wheel.

    it may be possible to drive weaker motors faster by over volting them and use water cooling and/or inverted logic to run them so that they do not use power at stop mode or even a transistor of some sort on the common wires to cut the power to the motor or driver while at rest .

    then use reduction gearing to slow down and apply more torque.

  27. paying development/prototype prices for mass produced parts sucks.

    Cut out the hundreds of middle men, and bring one printer to mass market, please! Economy of scale, where you at?

  28. I dont’t think the CNC filament style 3D printer is a great architecture. Resolution depends on a lot of different structures and interfaces, so you need a well built machine if you want quality prints.

    I like SLA printing, because all the complex stuff is contained in a little galvanometer package, which could feasibly be cast as one piece, so simple.

    Problem is, UV resin for SLA printing is so damn expensive. Perhaps research into reducing the cost of UV curing resin would do some good?

  29. Agree fully with chaosbc, 3d printers are not to expensive at least not all of them but some are still price gauging just like what is done in any market.

    3d modeling software for the masses is all that’s missing and by masses i mean people who are lazy and wont put the time in required to learn a new skill.

    My Solidoodle 2 was cheap, about 15LB’s of various plastic have gone through with only 2 jams both from laywood, two years of use and abuse its still going strong.

    Cheap and reliable has already happened so what are we doing here trying to reinvent the 3d printed wheel?

  30. I think that a mass-produced cheap version of the Printrbot Simple metal will be the lowest it will go in terms of price without artificial subsidies. So it’s not going much below 200 bucks, I’d predict.

  31. How is this so? Everyone who views your site makes at least 40k a year and decent printers go for around $500.00(not counting all the lazy hype ones from free-money-sucker sites)

  32. Remember when laserprinter used to cost a few thousands or so? Now you can get one for under $100. Remember when a “cheap” color inkjet printer used to cost $500 from local Walmart or such? Can get it for under $50 these day and sometimes free. Stand alone scanner used to cost several hundred dollars too and these day they are part of cheap $50 printer. 3D printer will come down over the next some years and I bet in about 20 years you could get a decent 3D printer with built in 3D scanner and 3D fax support for under $100.

  33. All of your thinking about the interchangebility of stepper driving and DC servo driving is missing one important consideration. What controls what? In inkjet printers the DC driven parts move just with loosely controlled velocity. The inkjet printing head nozzles (having practically zero inertia) can adapt to it with high accuracy in time, using the (slightly uneven) reference impulses from the encoder. It is totally opposite, compared to that, what we would need in the case of a 3D printer. So forget about the DC motors in this task, and use the good old steppers!

    I think (but I may be wrong), the only function – except positioning – which can be driven by DC motor in the 3D printer is, the filament advance. The uncertainties in the motion of the pushed filament, caused by the poor control, (IMO) will not result such crucial imperfections on the workpiece as it would in the case of positioning problems.

    Let me explain another idea on a totally different portion of the problem of decreasing the price of a 3D printer. What about save one stepper and its electronics (and neccessary power supply capacity as well) by driving the Z leadsrew mechanically, by just moving the printhead against a ratchet mechanism at the ending position? It needs to be done once after every layer is ready. Not a big deal, and can be accomplised by a relatively simple change in the driver software or even just in the STL file.

  34. Give every consumer a reason to have one, and then the price will come down. How often does your ‘fridge handle break? How often do you doodle in a 3D CAD program for fun and print out your creation? How often do you need to print an object for an interview, for school work, or for … ANYTHING? The only people who do this are crazy people who spend $1000 on a 3D printer, or build one themselves. Make a compelling argument for every parent to design and print their children’s pencil cases, trapper keepers, and calculator cases, and then every parent will buy a 3D printer and the cost will come down.

    Take drones for example. Two years ago drones were still garage-built frankenstein creations. Last Christmas, every middle aged father of three got one from Best Buy. Why? Because, gee-whiz, that’s cool! And, those drones aren’t “cheap”. What they are is user-friendly and well marketed. Make a 3D printer that plugs in, prints anything you throw at it THE FIRST TIME, and then spits it out for you fully finished, ready to start another print. The hardware is proven, the next step is improving the software and user experience and generating demand.

  35. I think if we looked at dumbed down printers or lower resolution (gasp) prints it would help bring down the cost in a lot of ways. How often to people print simple objects like guitar picks, lens adapter rings, etc that really don’t need a 3d printer that’s terribly smart.

    How cool would it be to be print a chair even if it was low resolution (think printing simple blocks).

    If you could master low-cost/ high resolution printing then run the low resolution part through a higher resolution / more expensive printer to kind of put the trim work on your objects you could speed up your print times and reduce the initial investment in the low resolution objects.

    Or maybe some new techniques for generating your layers. The following example is extremely simplified but you’d get the idea for simple low resolution objects. Take a giant roll of transparency sheet (like for old school overhead projectors) and cut out the layers then stack them.

    Just spit-balling but I think there’s got to be some work done on not over-engineering a product to produce supremely accurate simple objects. It’s like buying a $20,000 laser printer to print out your emails from grandma at home and put them on the fridge. I get the need for SLA type resolutions and complex 3d objects that might not otherwise be able to be manufactured. But average joe prints skulls and phallic objects.

    1. I don’t get the pushing for SLA as “the future”. Sure, it’s nice, but I think the polyjet system is better. SLA is a single material, polyjet is much more flexible method of building a part, including blending material properties like an inkjet blends inks to generate colors. Unfortunately, it’s locked away with patents for a few more years yet. A handful mention metal fabrication methods, which I guess would be pretty far off for anything looking like home fabrication.

  36. 3d printers will always be a niche market and I think we need better printers, not cheaper ones.
    I do see them becoming more popular than now but remain a niche market such as
    hobbyist lathes. Your grandma won’t be running a lathe anymore than a 3d printer unless she is strange.

  37. I believe there are three major issues about 3D printer cost.
    1) Volume manufacturing – 3D printers are not manufactured in bulk like 2D printers, so the cost will be high because parts will be expensive
    2) Print Quality – Active feedback controls are needed, like optical encoders on steppers or servo motors, what have you. Ever end-up printing something wonderful yet to have it completely ruined with a missed step?
    3) Learning curve – 3D printer users basically need to be technicians in order to operate the machine and design tools. 3D printers are CnC machines, so that knowledgebase is required to set-up the machine and generate designs. Also, setting-up the machine and troubleshooting print issues takes a specialized skill-set. To my knowledge, there is no one central location where people can ask about and get well-written directions for 3D printing solutions.

  38. Maybe if someone made a design with some adafruit feedback servos and a cheap controller with some threaded rod ( only limitation is rotation of 180 degrees, but then u could have a set min and max for each dimension, hense no limits needed other than the feedback vaue [ pretty easy to code and have… ] ) Any who a controller and such would be 45 USD for 3 feedback servo motors and an arduino controller :
    https://www.adafruit.com/products/1449
    https://www.adafruit.com/product/2590

    Prob get rod cheap and build yourself a small little printer for under 100.
    Ill sit and design such a small device tomorrow.

Leave a Reply to Genki Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.