Glow Plug Turned Metal-Capable 3D Printer Hotend

At this point, most readers will be familiar with fused deposition modeling (FDM) 3D printers, and how a plastic filament is pushed through a heater and deposited as liquid through a nozzle. Most of us also know that there are a huge variety of materials that can be FDM printed, but there’s one which perhaps evades us: you can’t load a spool of metal wire into your printer and print in metal, or at least you can’t yet. It’s something [Rotoforge] is working on, with a project to make a hot end that can melt metal. Their starting point is a ceramic diesel engine glow plug, from which they expect 1300 C (2372 F).

The video below the break deals with the process of converting the glow plug, which mostly means stripping off the metal parts which make it a glow plug, and then delicately EDM drilling a hole through its ceramic tip. The video is well worth a watch for the in-depth examination of how they evolved the means to do this.

Sadly they aren’t at the point of printing metal with this thing, but we think the current progress is impressive enough to have a good chance of working. Definitely one to watch.

Previous metal 3D printers we’ve featured have often used a MIG welder.

Thanks [theFinn] for the tip!

37 thoughts on “Glow Plug Turned Metal-Capable 3D Printer Hotend

    1. Melting point of ferrous metals? Oxidation of same? Need for shield gas.

      Why not start with solder?
      I bet you could do a messy job of that with a solid metal hotend, direct drive and heating element from a cheap iron. Aluminum hotend melting point isn’t that low.

      Speaking of which, he’s attempting copper?
      Why not aluminum wire?
      Start with easiest/lowest temp.
      Walk then run.

          1. That still seems quite doable. And solder might be much easier to re-melt and reuse than plastics.

            Re having kittens: never forget that, although life is a bitch, the puppies are cute.

      1. Yeah I’d assume that whatever they are doing needs to happen inside a chamber filled with inert gas. You can get a reducing flame on a torch.. Not with a glow plug or other methods of heating afaik. Gotta keep all those greedy grabby oxygen molecules out.

        1. the glowplugs internal bore is small, and tightly toleranced. the high temperatures inside it exclude air via thermal expansion and thus create an somewhat anoxic environment inside the glowplug itself.
          the thermal cycle at the build plate surface is very fast (single digit to tens of milliseconds) this means that there is little time for oxidation as deposition occurs.

      2. With copper, one can join (braze) metals without flux. I didn’t believe it until I tried it. It’s not as strong as brass, but it works surpsingly well.I didn’t really try to melt copper onto copper but assume that could work.

        I think I’d have a really hard time melting copper onto copper well and cleanly without melting the previous layer while using oxy-acetylene as I brazed with. Induction heating the material for the new layer seems like it would prevent overheating the existing layer.

        1. You can’t induction heat copper. Ferrous metals only.

          Brazing, by definition, means you partially melt the base metal.
          Aluminum brazing rods exist, not the same alloys as the base metals. Worse than welding in many respects, but ‘torch’.

          1. Pretty sure the first induction heating demo I ever saw involved dropping a slug of ALUMINUM into a pair of coils. It hovered between them, melted, and started to glow. They killed the power, and it dropped into a bucket of water, hissing.

          2. Video examples of inductively melting copper can be found online.

            Brazing does not involve melting the base metal. It’s a hotter process than soldering but is essentially very similar.

            If the ‘torch’ comment was directed toward me, then yes, I considered torch to be implied when I mentioned oxy acetylene. I also tend to say “arc welding” rather than “electric arc welding”.

          3. Brazing:,is%20defined%20as%20brazing%20(Ref.

            Diffusion of brazing metal into base metal is characteristic., not with soldering.

            Induction heating works effectively on ferrous metals _only_. Sure with enough power 1% energy transfer can heat up other metals, but not effectively. Try an aluminum pan on an induction stove.

            ‘Torch’ was in reference to brazing being much more convenient in the field.

          4. IIRC Induction heating for iron is (was?) typically at around mains frequency while for al i think it was about double and for a crystal pulling machine for synthetic rubies (laser rods) the al2o3 (doped w chromium oxide) was about 900hz, and as far as i can recall efficiency was related to the tuning of a tank circuit/coupling of the load within limits of course.

      3. I am working exclusively with nonferrous metals. the glowplugs will react with iron above 900 C. So steels are off the table for now. we are working on other hotend designs that may rectify this.

        In the semisolid state metals have 3-6 orders of magnitude lower oxidation rates.
        this fact combined with the very fast thermal cycles greatly reduces oxidation in the deposited material.
        for example:

        Solder alloy printing has already been achieved with aluminum hotends of great size and bulk.
        solder alloys are also, next to useless as anything other than demonstration parts. we have done some testing with them already as easy demo cases for getting temperature control.

        I am not starting with copper. we are starting with 510 phosphor bronze. the semisolid range of bronze is very wide, and the tin acts as a getting for oxygen and a flux for copper oxide. this means its ability to bond to itself (and to glass) is actually very good at semisolid temperature between 930 and 1060 C or so.

        I am also working with aluminum 6061 and aluminum 1100-O.

        The temperature is not the biggest challenge, its the working range. aluminum tends to be thixotropic in a relatively narrow range, which means that its viscosity evolves quickly between the solidus and liquidus which requires tighter thermal control. Bronze has a window almost 5 times wider than most aluminums.
        aluminum’s widest window is maybe 30C wide between 580 and 610 C, bronze’s is almsot 130 C wide.

        1. Thanks for the answer.

          What kind of temperature control do you have on the working part?
          First thought (for me) would be you’ve got to build the printer inside an annealing oven. Run as hot as possible. Being in a 400C environment would change the print head design, maybe simplify. Makes the rest of the printer more difficult.

          Also: Fire is cool! Getting paid to play with fire is even cooler.
          _Find_ a way to incorporate a pulse jet in your device. Many views. Glorious noise. Get to know your local police.

          1. With a type K thermocouple bonded intimately to the ceramic glowplug tip, we can maintain a temperature using model based control to within 2-3 degrees C of target at a heater temperature of ~975 degrees C.
            I intend to improve on this in future by eliminating insulation between heater and thermocouple and having a secondary measurement modality using the ceramic heater current consumption for faster response and online calorimetry.

            its a common misconception that you need a heated build plate or chamber to print metals. because they print thixotropically, they are more akin to chocolate or cheese printing than plastics.
            this also means that they do not require chamber or bed heating. in fact, such things tend to induce warping. not prevent it.
            The rapid quenching of the metal as it exits the nozzle is necessary to keeping grain sizes small and residual stresses low. Subsequent passes over the previous material anneal the previous layers if the temperature control is correct.

            Welders do not generally preheat their whole parts when they weld, nor is this the case when brazing or soldering. The only important place, is that the immediately underlying(and or adjacent) material is at least at the diffusion forge welding temperature. This is typical 0.3-0.6 of T melt.
            because the ceramic heaters have energy densities comparable to a torch flame (about 10 watts/square mm) they have no problem obtaining this temperature in the underlying layer as new material is deposited in the semisolid temperature range.

            I have already tried flame/detonation spray as a method for printing metals in the past.

            it has its own set of challenges despite how cool it might seem. :P

    2. I am the guy in the video.
      for fine wires(and thus low drive power requirements) comparable to what a typical 3D printer uses, the cost and electrical engineering requirements are expensive.
      Heating fine wires requires megahertz drivers or exotic coil geometries like radial halbach arrays. both can be expensive to produce, and the readiated power from metal at near melting is hard to deal with, without increasing the coil size.

      this is before we consider the fact that the vast majority of the energy of a standard induction coil is likely to be lost to heating the previous layers or the build plate.

      the lowest power demand I have seen for induction coils using 0.9 mm OD welding wire, is about 3.2 kilowatts.

      Additionally, I want to print ceramics. Induction heaters will not be effective on ceramics until you reach at least the 10s of gigahertz drive frequencies in most cases.

      here is an excellent paper on the topic of induction heater printing of semisolid metals…

      If you believe in induction heating, I would encourage you to pursue it!
      the method can work, and the engineering work to get it to a price point that makes sense for the home desktop would likely lead to many useful developments!

  1. Relativity Space has developed aditive metal wire deposition printing. It took whole decade and is driven by extremely complicated thermal modeling where a thing is being printed basically very specifically prewarped and it straightens out as it cools.
    Oh and it is a very stateful process informed in realtine by cameras from IR to Xrays.

    Good luck recreating it. Even simple metal dust SLS is extremely complicated when taking into account heat and dimensional accuracy.

    1. Relativity are using a fundamentally different deposition process. They use a GMAW head to deposit molten metal via arc discharge, with primary arc heating focussed on the workpiece. The OP setup uses conduction heating off a completely different class of alloys, with layers effectively brazed together.
      The difference in toolhead power is also significant: Relativity are pumping kilowatts into their GMAW head, OP is only using a few tens of watts.

      Whilst part warping compensation and real-time part inspection are desirable, they would be just as desirable for polymer FDM.

      1. Ok, I acknowledge the difference.
        And polymer FDM really would benefit from warping compensation. Even with 65C heated chamber and perfect bed adhesion I have warping problems with certain big (>20cm..) ASA prints. There is a tension which slightly buckles thin container bottoms after they pop off bed.

  2. Seen this a week or so ago. In order to drill the sparkplug you need an EDM machine, not easy accesible.
    Another issue (touched by other comentators) is the oxidation of metals heated to melt point in oxigen environement. I always think about some vacuum while forging or casting in order to reduce oxidation.

    Metal 3d printing is at an impass: powder metal gets glued/melted with laser, then backed, not easy and dangerous (metal dust); welding uses more material, equipment needs to resist to and evacuate hot gases, it needs further machining to achieve desired dimensions; electtodeposition (like plating) takes time and the result is kinda porous.

    I have an ideea: use precise machined and polished tiny cubes (or other 2D and 3D shapes), like the gauge blocks, then you stack them in the desired position, compress them to make them to adhere one to the other, and also pass a current through them to increase the fusion, then you’ll get a minecraft representation of the object that you may further machine selectively to the desired dimensions.
    The raw cubes can be manufactured and sold by people with heavy equipment and skills.

    1. I would be very surprised if squeezing some metal blocks together and passing a current through them is enough to create a strong bond. If it worked then they would probably already use that for manufacturing, rather than welding. If it were to work there is a good chance the metal ends up deformed anyway.

  3. Home or hobby 3D printing metal into usefull structural components will never be a reality
    current laser powdered metal flat bed printers
    are producing parts for all kinds of things,but
    you need a use case where complexity and strength needs outweigh cost.
    the laser grade powdered metals are
    wildly pricy,and horribly finicky to use
    so if anybody is realy realy realy gona do this
    then building a ball mill and getting a phd in metalurgy is good place to start before spending
    10000 hrs perfecting the process

    1. MIG welder on large gantry. Only tricky part is getting the welder settings right. Clipping wire.

      IIRC the swamp Germans made a pedestrian bridge that way. Two 3d printer/welders met in middle. I assume they had an attendant to keep them running.
      Might have burnt down, fallen over and sunk in the swamp later.
      For the price of welding wire and gas, you could have gotten that bridge prefabbed and installed.
      Had enough money left over to run out of things to do, in Amsterdam.
      Not the point.
      Sing ZZTop ‘Just got paid’ in Dutch, while motorboating huge breasted eastern European woman.

    2. I am finishing my PhD in materials science and engineering.
      Metallurgy is one of my primary research topics.

      Semisolid metal forming and forging have been a thing for a long time.
      in fact, forge welding and diffusion welding are far older than metal powder SLS.

      The machine we are building and the hotend besides, are intended to perform semisolid diffusion forge welding directly from solid wire.

      a few good papers and a nice wikipedia page if you would like to read about it.

  4. Not gonna lie, I thought swamp Germans was like a youtube channel or something. Further research determined it was a nickname for the Dutch. Googling “swamp germans pedestrian bridge 3d print” did get the results though. ;)

  5. You need high tech alloys and high tech heating to get even close to how FDM works with current materials (the filament nerds to be extruded as a sticky paste – not a liquid). Stuff not currently in the realm of home hackers.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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