Intermittent motion using differentials

[arieben]

I was messing around with differentials and found a way to use a linkage to achieve intermittent motion. however, the standard gear diffs ended up having too much backlash so i built this version.

In this example, the output stops for 1/5th of a rotation of the input, and makes a full revolution for every 5 rotations of the input. It can be modified to suit any gear ratio, the original one I made with the lego diff pieces was 3:1.

The input goes directly to a 2L crankshaft attached to the rocker, which is 6L from the attachment point to the pin it's mounted on. Since what matters is the distance between centers of rotation, this is a 5:1 ratio (the same as the input to the differential, 8t to 40t gear. If i were to do it with a gear ratio of 1:3, then the rocker would be 4L,.... etc). The rocker extends one more stud and attaches to the differential 3L from the axle it's mounted on. this gives it a final 2:1 ratio to cancel out the averaging effect of the differential (B = A + 2C). I dunno, it's easier to make sense of in my head, sorry if my math doesn't make sense.

Unfortunately, as I show in the end of the video, I still have the problem of backlash. most of it stems from the fact that the differential is mounted on an axle and can shift around a bit. it doesn't help that it's built out of multiple pieces as well. If you had the space, you could probably use turntables and a 5x7 frame and surround them with idler gears in order to remove most the backlash from the mechanism. this is simply a proof of concept. I figure it could be useful though, particularly because it can handle a lot of torque, and you guys could probably find better ways to optimize it. 

[schraubedrin]

That's a cool concept!

Could be used in some kinematic sculpture.

 

Is the backlash still a problem with a load on the output?

[arieben]

6 hours ago, schraubedrin said:

That's a cool concept!

Could be used in some kinematic sculpture.

 

Is the backlash still a problem with a load on the output?

Thanks, I was thinking it could probably be useful for GBC if refined further. it seems most Geneva mechanisms in lego are finicky and not terribly adaptable. 

Aside from a load that would bend the axles, what you see at the end of the video is the extent of the backlash. I'm pretty much applying a maximum amount of force to it.

i think tomorrow I'll try to draft up a version with turntables and 5x7 frames.. 

Actually, the small turntables are probably better for the task. However, I have precisely zero of those. So unless anyone wants to have a crack at it, I think that'll be on hold for a while until I can get some. 

* never mind about the turntables. I just tested the large ones with idler gears and they still have the same problem of shifting around. Are there any good purely lego bearings? Everything I try seems so wobbly because of the built in tolerances. Added friction is not a huge issue because the mechanism is pretty heavily geared down. 

Edited August 22, 2020 by arieben

[pleegwat]

7 hours ago, arieben said:

Are there any good purely lego bearings?

Wheel hubs?

[amorti]

The new large turntables with only gear teeth on the outside are a fair bit tighter than the older type. But there's always going to be an amount of play in Lego.

Do you suppose the new 5-gear differentials would have less play?

[1963maniac]

Can you supply a digital file or more pictures on how to build it? I'd like to try it. Thanks!!

[BusterHaus]

11 hours ago, arieben said:

most Geneva mechanisms in lego are finicky and not terribly adaptable. 

A strong Geneva mechanism with minimal slack in Lego is almost a Holy Grail.  The ones that I've seen are either huge, have a lot of play, or only work as a proof of concept and could not support high torque.  Your solution looks to be more locked in place than those, and looks like it could be refined to be stronger.  I'll second @1963maniac, could you please post some pictures of the assembly?

[Void_S]

Fantastic! A pure genus idea to combine the input with custom-made adder and pair or levers.

[arieben]

9 hours ago, amorti said:

The new large turntables with only gear teeth on the outside are a fair bit tighter than the older type. But there's always going to be an amount of play in Lego.

Do you suppose the new 5-gear differentials would have less play?

I doubt it, considering they still use lego gears.. i tried a 4 gear differential (with thick gears, even) and there was just too much slack.

6 hours ago, BusterHaus said:

A strong Geneva mechanism with minimal slack in Lego is almost a Holy Grail.  The ones that I've seen are either huge, have a lot of play, or only work as a proof of concept and could not support high torque.  Your solution looks to be more locked in place than those, and looks like it could be refined to be stronger.  I'll second @1963maniac, could you please post some pictures of the assembly?

 

9 hours ago, 1963maniac said:

Can you supply a digital file or more pictures on how to build it? I'd like to try it. Thanks!!

https://imgur.com/a/SQjZ0Co 

might learn LDD soon so i don't have to disassemble things.. :P but it worked out for this because i was able to re-engineer it to be a bit more efficient and robust.

and thanks, @Void_S :)

I am going to get to work on bearings. I have some ideas. Please if any of you know pieces that are round, 1 stud wide, and have a pinhole (not an axle hole) in the center AND have mounting points for the differential casing, let me know. For now I'll dig through my collection and see what I can find.

the idea is to use wheels with rubber tires (probably the little roller ones) and said round piece as a bearing for it.

Edited August 22, 2020 by arieben

[aeh5040]

That's extremely interesting and innovative!  I too have wondered for a long time how to do such things robustly.

What I would really like is a mechanism that achieves something along the lines of:

no rotation for 5/6 of the time period, then exactly 2 rotations taking up the remaining 1/6 of the time period, repeated.  All with good torque and minimal backlash.  (The exact numbers don't matter to me, but I want something roughly like this).

Timed shifter mechanisms don't work, because the number of rotations cannot be controlled precisely.

Cardan gears can't achieve anywhere near this much or the period with no rotation (I think).

Geneva mechanisms or intermittent chain drives in principle do it, but it is very hard to make them strong or precise enough. 

The feedback mechanism I used in the Pendragon does do this, but it also has problems of torque and robustness and backlash.

Edited August 23, 2020 by aeh5040

[arieben]

10 hours ago, aeh5040 said:

That's extremely interesting and innovative!  I too have wondered for a long time how to do such things robustly.

What I would really like is a mechanism that achieves something along the lines of:

no rotation for 5/6 of the time period, then exactly 2 rotations taking up the remaining 1/6 of the time period, repeated.  All with good torque and minimal backlash.  (The exact numbers don't matter to me, but I want something roughly like this).

Timed shifter mechanisms don't work, because the number of rotations cannot be controlled precisely.

Cardan gears can't achieve anywhere near this much or the period with no rotation (I think).

Geneva mechanisms or intermittent chain drives in principle do it, but it is very hard to make them strong or precise enough. 

The feedback mechanism I used in the Pendragon does do this, but it also has problems of torque and robustness and backlash.

I remember that pendragon machine! Haha, it went way over my head the first time I saw it. It still does a bit, I might have to study it a little.

I think as long as the math is done correctly what you're wishing for can be done with some combination of adders and differentials. If I can get a way to make the differential without slack, I may just have to try that. I'd be interested in a mechanism like that myself, anyway.

Here's what I have so far: http://imgur.com/a/3UX7MBx this bearing seems to work really well. By using 3L bars with the blue pins, I'm able to attach the frame solidly to the blue wheel piece. The knob wheels are held nicely in place by, again, the umbrella stand. So far, it seems to have very little play overall. 

Edited August 24, 2020 by arieben

[Horologist]

23 hours ago, aeh5040 said:

What I would really like is a mechanism that achieves something along the lines of:

no rotation for 5/6 of the time period, then exactly 2 rotations taking up the remaining 1/6 of the time period, repeated.  All with good torque and minimal backlash.  (The exact numbers don't matter to me, but I want something roughly like this).

The Russian mathematician Chebyshev made several linkage mechanisms that achieve something similar: see https://en.tcheb.ru/12 (The whole website is full of interesting mechanisms as well).

[iLego]

11 hours ago, Horologist said:

Russian mathematician Chebyshev made several linkage mechanisms that achieve something similar: see https://en.tcheb.ru/12 (The whole website is full of interesting mechanisms as well).

Thanks for sharing this. Lots of interesting ideas there.

[Horologist]

8 hours ago, iLego said:

Thanks for sharing this. Lots of interesting ideas there.

You're welcome . I found it while doing research on walking mechanisms.

[arieben]

New update:

I have rebuilt the mechanism to be as strong as possible including those new bearings (which work wonderfully, actually). The good news is, it's possible to eliminate the backlash from the linkages as well as the axle or turntable the differential was mounted on. 

The bad news, however, is abundant. As you can see in the video:

-The bearings force the construction to be 6 studs longer and many units wider.

- The main concern, backlash, is still present. Especially with the increased length due to the bearings and diff casing, the axles are so long that they just bend. Combined with small but unavoidable backlash of the knob gears, it's quite significant.

So, in conclusion.. I think it's a lego problem. I don't believe the mechanism is a problem, since the axles bend before any of it does.. but either way, they do.

Edited August 25, 2020 by arieben

[BusterHaus]

Saw this today. The linkage solution for intermittent motion is gaining popularity. 

https://twitter.com/05_okome/status/1299310973191254020

@05_okome: 無限に見てられる https://t.co/Q6CNCx4Qlg

 

[arieben]

On 8/28/2020 at 8:41 AM, BusterHaus said:

Saw this today. The linkage solution for intermittent motion is gaining popularity. 

https://twitter.com/05_okome/status/1299310973191254020

@05_okome: 無限に見てられる https://t.co/Q6CNCx4Qlg

 

Neat.. i wonder how strong. it is definitely compact though that's for sure.

i'm still on the quest for a solid intermittent motion device. so far i think @Doug72's solution is the best, since it has the shock absorber right at the very end of the mechanism.

perhaps ours could be combined somehow? differential plus the shock absorber thing. i'll have to study it more in depth.

[doug72]

6 hours ago, arieben said:

Neat.. i wonder how strong. it is definitely compact though that's for sure.

i'm still on the quest for a solid intermittent motion device. so far i think @Doug72's solution is the best, since it has the shock absorber right at the very end of the mechanism.

perhaps ours could be combined somehow? differential plus the shock absorber thing. i'll have to study it more in depth.

Don,t recall making a post about this !

[Horologist]

On 8/28/2020 at 8:41 AM, BusterHaus said:

Saw this today. The linkage solution for intermittent motion is gaining popularity. 

https://twitter.com/05_okome/status/1299310973191254020

@05_okome: 無限に見てられる https://t.co/Q6CNCx4Qlg

 

Technically, the intermittent motion is generated through the eccentrically driven gear, the linkage is there only to transfer the output of the gear to a stationary point on the structure.

Edited August 31, 2020 by Horologist

[arieben]

19 hours ago, Doug72 said:

Don,t recall making a post about this !

Right - should have linked the thread. I didn't realize how old it was! 

 

I will try my hand at making a combination of the two. It seems that using a mechanism at the output is the only way to really resolve the issue of backlash.

[doug72]

Thanks for reminding me, forgotten about building that, never got the Ballkirk  wheel fully completed.

Might  rebuild the mechanism and see if can make more compact.