Fabbing mechanical parts

Fabbing mechanical parts

Fjr of Mobile, AL, wanted to build a trebuchet. He soon discovered that it was a huge hassle because the parts he could find weren’t compatible with each other.

I was talking to my brother and as I talked I was frustrated at how hard it was to build things. The biggest problem in my opinion is, everytime I want to build something, it pretty much always involves some kind of shafts.

Shafts have several functions:
1. They hold components where they need to be. But they also need to be held in place on the shaft.
2. Shafts give various components a common rotational axis.
3. Shafts transmit motion from one component to another.

If you want to place a component on a shaft and have it perform reliably, it needs to be held in place on the shaft. If you want a component to spin smoothly, then you need a bearing placed on your shaft. The bearing needs to have the same inner diameter as the shaft’s outer diameter, within a few thousandths, if not things will vibrate and wobble and all kinds of problems. Most shafts get stacked on with a different diameter for each component. Thats alot of custom machining.

I don’t have the resources to machine custom shafts each time I have an idea, and I’m not alone in not having them.

What’s needed is a standard, a system.

Fjr created a number of possible parts to his “Simplified Mechanical System,” including the two above. What do you think, readers? Do we need a series of standard components for maker projects? Leave a comment.

28 thoughts on “Fabbing mechanical parts

  1. ... says:

    You might try looking to the optics industry for inspiration, they have established a system where you start with a table that has 1/4-20 holes on 1″ centers, and then add any number of standard components (post mounts, stages, connectors, etc). With the system it is possible to build very complex experiments easily without the need for any machining.

    For examples check out thorlabs.com, they have a huge assortment of components

  2. Forklift says:

    So… a more robust Lego Technic, then? Sounds good to me! I would think this problem might solve itself, though, as homemade, open-source, and commercially available fabbing equipment (mini-CNCs, 3D printers, etc.) become more common.

    Just one request. Make it in metric, please!

    1. fjr122 says:

      It is metric. The skate bearings its designed for are 8mm inner diameter, 22mm outer diameter and 7mm thick. Nice round metric numbers. :)

  3. packrat says:

    Well, not looking at all, tbh.

    There is an expansive library of parts that are easily interoperable out there. Companies like Misumi and McMaster can provide pretty much everything you could ever want to build machinery. Mind you, you’ll have to do a bit of work to sort out exactly what you need, but once you get the hang of it, and learn all the names for the various parts available, you can build all sorts of stuff. There is also the prices involved with buying from companies that stock “everything.”

  4. fjr122 says:

    I have used McMaster. Very handy catalog to find things and know what to call parts. I built something a couple years ago using skate bearings. Both the metric and standard bolts were *close* to 8mm not exactly. So I eventually found and ordered some “shoulder” bolts. They are *exactly* 8mm up to the shoulder then they were threaded 4mm for the remainder of the length. Those are currently $5.71 on mcmaster ( http://www.mcmaster.com/itm/find.ASP?tab=find&context=psrchDtlLink&fasttrack=False&searchstring=90278A427 ). I needed 6 of them!!!

    I also ordered some precision 8mm rod while I was at it (also not available….anywhere!!). But it was too hard to cut at all. It’s currently $16.28 on mcmaster http://www.mcmaster.com/itm/find.ASP?tab=find&context=psrchDtlLink&fasttrack=False&searchstring=1265K64 ).

    Here is a bearing same size as a skate bearing on mcmaster ( http://www.mcmaster.com/itm/find.ASP?tab=find&context=psrchDtlLink&fasttrack=False&searchstring=5972K501 ) its $4.06 an 8mm mounted in aluminum http://www.mcmaster.com/itm/find.ASP?tab=find&context=psrchDtlLink&fasttrack=False&searchstring=8600N13 is almost $15. It would take 4 pieces and screws in my system to emulate the same function. But you could replace the bearing if it wore out. And it wouldn’t even be $1. And you could tear all the pieces down and build something else with it if you wanted.

    When I said I lacked the resources one of the main resources I was talking about was money. I was building something that needed 4 mounted bearings and 4 of those shoulder bolts.

    That’s almost $80 for just for a simple assembly.

    Surely we can do better than that!

    1. packrat says:

      Point is, though, that there is “a standard.”(really standards) They’ve already been set by the ASTM, ANSI, SAE, ISO, and whatever other standard-setting boards there are out there.

      Understandably, the costs are a bit painful for being able to rapidly build/prototype your design.

      Your comparison to the Arduino in the blog post is more apt than you realize.
      The Arduino setup is, comparatively speaking, an expensive way to build things, considering the microcontrollers used in them run about $5 a piece. You are actually paying quite a premium for the ease of rapidly building and programming your project.

    2. Chris G says:

      There are several other places to source parts that I have used:

      Stock Drive Products (http://www.sdp-si.com/)

      SmallParts (http://www.smallparts.com/)

      and others as well.

      The issue seems to be cost/value. Yes, it does seem to be very expensive for ‘simple’ items like pillow blocks, and shoulder screws, etc… but th adage is that there are three conflicting things in play.

      You can have the parts: Fast, Accurate, cheap. (Pick Two)

      The McMaster-Carrs of the world will give you fast and accurate. Making them yourself may give you Accurate and cheap. A 3D fabber ‘may’ give you fast and cheap.
      The Arduino analogy is most apt.

      This is where it pays to scrounge. Perfectly acceptable bearing blocks, lever arms, etc.. can be found in old washing machines, printers, and the like.

      Also look for using simple bushings instead of bearings where possible. Bushings can take quite a high load, can be self-lubricating, and if alignment is a problem, as the home CNC people have shown, they can be epoxy’ed in place after alignment.

      Good luck on your project, I don’t mean to be critical, I feel your pain….

      1. fjr122 says:

        It does pay to scrounge, its amazing what gets thrown away.

        My wife isn’t really in favor of my “recycling” in most cases.

        It’s difficult to know what to keep and what to toss as soon as you find it.

        I like to believe I would use something and build some gadget with it but the truth is I’ve built very few things, and that stuff *mostly* from new materials.

    3. mpantaz says:

      McMaster isn’t the best choice for metric parts — they tend to have a limited selection. That said, those prices seem high even for McMaster, so I looked up the part numbers… Did your project truly require stainless steel and such high precision? In just a couple of mouse clicks I found much better prices:

      “… found and ordered some “shoulder” bolts.”
      Same size in “Alloy Steel”, part # 92981A203, $1.44 each.

      “I also ordered some precision 8mm rod…”
      Tool Steel 8mm drill rod (Part # 88625K67) with a ±0.013mm tolerance is only $4.88. If you really feel the need for the closer tolerance, simply going with their linear shafting (part # 6112K44, $5.95) gives you +0.0/-0.009mm, which closely matches the tolerances for the bearing you selected. It’s case-hardened 1055 steel, but an abrasive cutting wheel will cut through it, or any of the above quite nicely.

      As for the mounted bearings, if you can live with a nylon plain-bearing, part # 6687K33 is only $4.42.

  5. Rahere says:

    Which raises some interesting questions about the use of standard metric threaded rod in Mendel: half of the problems that has are to do with backlash in the mounts, which is why the second rail is floating in all three dimensions. If you can’t be certain about that, then you’ll never get any precision shaping.
    A better approach would have been to use lead screws – after all, lathe designers have had the option all these years and have stuck with that, and all Mendel is is a mill with an extruder rather than a cutter. McMaster, for example, do 18″ and 36″ fast screws in their ACME range designed exactly for this: use cut nuts and you can rough-advance to where ever you need. Add in their flanges and the complicated 120° mounts disappear, particularly if you replace the triangular end frames with steel strut channel and 60° angle brackets.
    Mendel’s approach is to ultimately be self-replicating, but they’re struggling with getting past the question of head chatter as a result. When half your bill of materials is from a supplier catalogue, you’ve got a way to go: they should go for output and then home in on their target.

  6. Rahere says:

    Let’s home in on Mendel’s floating mounts for a second. They translate bind into perpendicular displacement, and as we all know, Murphy dictates that these tolerances will accumulate unidirectionally into maximum difficulty in assembly. Bind is itself caused by non-axial movement in the platform: by contrast, on a lathe the platform bed itself is used as the reference point, and the lead screw follows. What’s needed is floating bearings for the lead screws – it doesn’t particularly matter where the thread thrusts against, simply that it thrusts fairly consistently from the same plane.

  7. metis says:

    the reason high tolerance hardware costs more is it requires a lot more accurate work to create. do you *need* all of your shafts to be on press fit bearings, or can you shim them in or use a compressible pin as a shaft? many one off or prototype projects can deal with a bit of slop until you sort out which bits are critical, and then replace those with a higher tolerance.

  8. fjr122 says:

    I looked on smallparts.com

    This is fantastic!


    That’s half inch, much more common in Alabama…would still need a precision rod though, hardware store stuff still wobbles..

    There’s an 8mm thats similar price too…
    Not a ball bearing, but it would work for many applications..

    Guess I need to hire out a machine shop to help me make one of my axles…

  9. Big Blue Saw says:

    You need to look for precision ground shaft; there are a variety of sources. If you can’t get that and it’s still wobbly, fill in the gaps by brazing, electroplating, or perhaps with epoxy.

Comments are closed.


My interests include writing, electronics, RPGs, scifi, hackers & hackerspaces, 3D printing, building sets & toys. @johnbaichtal nerdage.net

View more articles by John Baichtal
Send this to a friend