I've been working on this problem for years (literally) and haven't yet found a workable, scalable, pragmatic, and cost-effective solution that would allow me to selectively rotate one or more of 80 discs closely aligned on a single, rotating shaft.

Basically, I have 80 identical "discs", each of which is approximately 2 mm wide, spaced out with approximately 2 mm between each disc, concentrically mounted along a shaft (which can be anywhere from 5-20 mm in diameter).

Unfortunately my background is in electronics and not in mechanics or mechanical engineering, and I'm unaware as to whether or not this can be considered a "solved problem" at this scale.

I have tried a number of different approaches (both conceptually and in prototype) but have constantly been stymied by the tight specifications.

Conceptually, something as simple as a gear translating rotation in one axis to another (perpendicular) axis, attached to each "disc" and actuated by a linearly-moving motor would do the trick:

enter image description here

One attempt I've used is to have a rotating "indexer" fixed to a spindle in the middle of the shaft (with the rest of the shaft being free to spin around the spindle), then moving that linearly while subjecting it to rotational movement when aligned with the disc in question (with a matching cutout in each disc mated to the shape of the shaft/indexer), but the need for extreme precision in order for it to move freely along the axis made it unworkable (along with the fact that the shaft/spindle now needed to be twice the width of the working surface. In the example below, the indexer would be 9b, used to rotate 11H without rotating 11G or 11I:

enter image description here

I know one solution would be a clutch of sorts - if one exists that would fit on/in each of these discs. Then the clutches could be mounted directly to the shaft, and the clutch for each disc (somehow, as of yet that's a big question mark) engaged or disengaged as needed when the shaft is turning to translate that motion to each disc.

I had an idea to use a "magnetic clutch" (if that is a thing) where rectangular strips of magnets would be mounted into channels in the shaft, then tiny magnets would be embedded around the periphery of the cutout in each disc. The idea was that when not "locked" in place by some external mechanism, each disc would (by magnetic force) rotate with a spinning shaft. When externally locked, the disc would then hold in place with the "clutch" disengaged as the shaft spun freely under it. Unfortunately, it seemed that even with neodymium magnets, the friction from the entire system was too much for the magnets and the discs were permanently in an "unlocked" position and would not spin with the shaft.

There is also the obvious solution of using pure friction. I have leeway in what material the discs and shaft are composed of, but I'm not sure if that's something that can be considered a reliable solution as it should last for hundreds of thousands of "disengaged rotations" without wearing out or losing friction.

I feel like I might be overcomplicating the problem. Bottom line: tiny discs, one rotating shaft. Selectively (preferably in parallel, but one-at-a-time is fine) rotating each of the discs, without needing tolerances only achievable in a factory and at a affordable price (let's say under $100).

  • $\begingroup$ It's not clear to me how exactly you're expecting this to operate or what your end goal is. There are 80 discs on a shaft - are all of the discs fixed to the shaft and they rotate together, or are they mounted on bearings and each is free to rotate independently? You want to selectively rotate one "or more" of these discs, which makes it seem like they're all free to rotate independently (else rotating one rotates them all) - what is your input? What is the output? Is the end goal just to rotate the disc, as in a centrifuge, or is the disc an intermediate device? $\endgroup$
    – Chuck
    Nov 15, 2016 at 14:50
  • $\begingroup$ What are your restrictions and/or limitations? What is the input source? What pieces can you move and what pieces need to be stationary? It would seem like the easiest answer would be to just move the input shaft around such that it contacts each disc (but again I'm not clear on what the discs are or what they do). What's your time limitation? Can you bring the input shaft to a stop between stages? If not, how are you envisioning synchronizing the speed between the (stopped, presumably) disc and the rotating input shaft? $\endgroup$
    – Chuck
    Nov 15, 2016 at 14:57
  • $\begingroup$ What speeds are these all supposed to be operating at? What are the other dimensions of the discs? They're 2mm wide... and how big in diameter? How big is the shaft to which they're attached? What is the surface finish on the edge of the disc? Are you allowed to contact the planar surfaces of the disc (top/bottom) or only make contact on the circumference? You have 80 discs that are 2mm wide, and you're looking for a (probably) custom machined piece to interface all 80 discs while they're rotating - you are almost certainly not going to get this under 100 dollars (at least not for quantity 1). $\endgroup$
    – Chuck
    Nov 15, 2016 at 15:00

4 Answers 4


A mechanical solution using movable rods:

  1. Have 7 rods with pieces of rubber around them. The rods should be positioned so that you can move them leftwards or rightwards. When the rubber piece touches the disc, it should stop the rotation.

  2. Place the rubber pieces on either the left side or the right side of each disc. This way when you move a rod to the left, it will lock some discs and release others.

  3. Arrange so that rod 1 when actuated will release the discs 0, 2, 4, 6, ... and lock discs 1, 3, 5, 7, .... Rod 2 should release 0, 1, 4, 5, ... and lock 2, 3, 6, 7, ....

  4. To rotate a single disc, actuate the rods according to the binary number of the disc. For example, to leave disc 7 free to rotate, actuate rods 1, 2 and 3.

  • $\begingroup$ I like it, quite clever. But that still leaves the question of how these discs are mounted to the shaft in such a way that when locked they spin freely about their axis. $\endgroup$ Apr 15, 2016 at 14:54
  • $\begingroup$ If you have binary-selected braking, you can also have binary-selected clutching using essentially the same principle. Lots of small moving parts, however, which might make it impractical. $\endgroup$
    – Dave Tweed
    Apr 15, 2016 at 15:45

Depending on the rotational speed and the torque required, you may be able to use something that resembles a pin tumbler lock (I can't put more than 2 links in a post so no link to Wikipedia..)

  1. The inner, rotating shaft will have holes in it, into which small cylindrical pins will be inserted: rotating shaft

(pins are red, shaft is black)

  1. An inner rod with some sort of a bump on it moves inside the shaft (either the whole rod moves or only the bump.

  2. When the bump pushes the pins outside they fit into holes inside a disc and force it to rotate with the shaft:

engaging and disengaging

(inner rod is gray, disc is blue, disc pins are orange and springs are purple)

  1. When the bump moves, the rotating shaft's pins are brought back to their disengaged position by springs inside the disc's holes (the pins inside the disc are not really necessary, they are there to show the resemblance to a lock).

That's just general idea and maybe it's a little gimmicky, but it is another approach.

  • $\begingroup$ Thanks for your answer (and signing up to post it!) and please have a +1 from me! I've considered variations on this before, but it's subject to the same problems that plague the project as a whole. With each disc being only 2mm in width, making something like this that will mate correctly and not jam under such tight tolerances is crazy.. $\endgroup$ Apr 16, 2016 at 18:15
  • 1
    $\begingroup$ Thanks! I figured this wouldn't be practical for you. Anyway, this method can be used to engage a clutch as well or any sort of a ratchet. Maybe I missed it, but I don't think you've specified the specifications as for the rotation speed and engagement speed etc. - these might help the community to better understand your needs and find an optimal answer :) $\endgroup$
    – AutoDude
    Apr 16, 2016 at 19:14

One possible idea:

  1. Lock the discs on the shaft using either wax or hotglue. Making the shaft out of plastic may make the glue bond better.

  2. Wrap a few turns of resistance wire around the shaft next to each disc.

  3. Run normal cable from the resistance wire back to control electronics through the center of the shaft.

  4. Apply power to melt the glue/wax on each disc you want to be free of the shaft rotation. It will take some time to melt it, and you will want to adjust the power so that it will not overheat.

  • $\begingroup$ Certainly creative but not a very pragmatic solution, right? I don't see it being very reliable or long-term. $\endgroup$ Apr 15, 2016 at 14:52
  • 2
    $\begingroup$ Hot-melt glue would clearly be impractical, but similar mechanisms have been created using ferrofluids activated by magnetic fields or electrorheological fluids that are activated by electric fields. $\endgroup$
    – Dave Tweed
    Apr 15, 2016 at 15:43

Band brakes around each disk.

What you do to the disk itself depends on the use case:

if it's unloaded, like an indicator like a digit in a mechanical computer, friction couple it to the shaft (which must then be capable of driving 80 times the friction torque of each disk).

If it has to transmit power, the band brake can hold the outer ring of an epicyclic gear. That'll allow efficient slippage and transfer of the motion to something else (in the 2mm gap between disks).

Each band brake is just a length of spring steel like a watch spring, wrapped round part or even most of the disk depending on the torque required ... one end is fixed, the other is either tightened or slackened by some actuator ... the keyboard from a mechanical typewriter?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.