npicard's small mechanisms

[npicard]

I am working on a very large-scale project, but I keep coming up with small mechanisms along the way that may be useful to others. I plan on using this thread to collect them.

Of course, I stand on the shoulders of giants, and most of my mechanisms are merely refinements of the work of others. I'll credit where I can.

As I'm new to this forum, I might take this space to briefly introduce myself. I built a lot of Mindstorms as a teenager in the RCX days. Then I had my dark ages from around 2007 until 2019. I decided to get back into it and purchased the 42100 Liebherr 9800. I'm mostly just interested in functional mechanisms.

Two-way to one-way converter

a.k.a. one-direction output

This is is a mechanism that takes rotation in either direction as input (green) and outputs rotation in only one direction, no matter what the input direction was. In this configuration, the red axle only outputs counter-clockwise rotation. Swap red and grey for clockwise output.

Inspiration came from this page: https://www.instructables.com/member/Jorbs3210/

I have seen and tested many of these mechanisms, but none that I found met my three biggest constraints:

• works in any orientation (doesn't rely on gravity)
• very low-friction (so, no rubber bands, no friction pins for idler gears, and not too many gears in total)
• if the input is in neutral and the output is being driven, it won't bind up

This final constraint is the downfall of most other solutions. If you rotate the output axle, gears will bind together and the whole thing will lock up. In my solution, the white 3L liftarm touches the studs of the grey 1x2 plate during operation. This prevents the gears from over-engaging, meaning they can't bind no matter which axle is driven. This has a nice bonus effect of keeping friction low.

Rotational driving ring selector

This mechanism takes constant rotation as input (green) and, based on whether it is clockwise or counter-clockwise, selects between two states of a driving ring. The benefit of this system is if the input is connected to a powertrain that is also doing other things, so it can't be "turned off" when the correct gear is chosen.

When you turn the input clockwise, the blue worm gear travels forward and eventually hits the light grey stop. Then it begins to turn the z8 gear, which turns the white 2L liftarm, which flips the red changeover catch.

This is based off of Sariel's pneumatic autovalve: http://sariel.pl/2008/12/pneumatic-autovalve/

@Keegan Pilling suggested Sariel's solution. I just worked out the details.

Low-friction driving ring axle

In the previous mechanism, I found that the white driving ring axle (that keeps it engaged with clutch gears with a "click") is too high-friction. The input needs to be fairly high torque. Unfortunately, the zero-friction 3L axle connector doesn't keep the driving ring in place at all. I was seeking a solution that was a balance between the two, providing just enough friction to keep the driving ring engaged with the clutch gear, and no more.

@2GodBDGlory provided the idea of using a 2L axle connector with a 3L 3.18 bar. I found it paired well with a 3/4 pin.

I hope these are helpful for someone! I'll post more as they occur to me.

Edited September 30, 2021 by npicard

[Hrafn]

These are clever!  Thanks for sharing.

[1963maniac]

These are good ideas! Keep up the good work. Thanks for sharing!!

[kolbjha]

Nice idea for a topic, and good ideas so far! I think it's good that you credit others when relevant, both being fair and also gives an insight to how your designs evolved.

I really like your two-way to one-way rotation solution, very simple and compact, yet effective! I have already tested it, and it works very well. Already have an idea for use 

Keep up your good work 

[MangaNOID]

23 hours ago, npicard said:

Two-way to one-way converter

This is very nice! 

with the low friction driving ring you can ‘flex’ out the two middle tabs a little to loosen Or eliminate the friction. Is this the same as what you have posted or does your solution have more too it?

[npicard]

22 hours ago, MangaNOID said:

This is very nice! 

with the low friction driving ring you can ‘flex’ out the two middle tabs a little to loosen Or eliminate the friction. Is this the same as what you have posted or does your solution have more too it?

Thank you! That would achieve basically the same result. However, my final build will have dozens of driving rings and I want my techniques to be universally applicable. I am not willing to modify or damage any parts to make it work. Of course, that's just me. To each their own! 

[AFOLegofan66]

Nice idea and very useful…thanks. 

[MinusAndy]

These are great. I’ve used something similar to your worm gear clutch ring mechanism to auto disengage the clutch when the worm reaches the end of its travel.

More importantly. What’s your big project???

[npicard]

On 10/3/2021 at 7:36 AM, MinusAndy said:

These are great. I’ve used something similar to your worm gear clutch ring mechanism to auto disengage the clutch when the worm reaches the end of its travel.

I find components with axle holes but no friction incredibly interesting and useful. I'm always glad to hear of mechanisms using them and I'd love to see yours.

As far as I know, there's only the 2L worm gear, the red z8 gear, and the changeover catch. Does anyone know of any other "loose" axle connections?

Quote

More importantly. What’s your big project???

I'm making a fully mechanical computer. It's my high-level design is close to final, but there are a lot of details to work out. To illustrate a small piece of it, if you combine my rotational driving ring selector and my two-way-to-one-way, you get a logic gate. The final result should have around 100 logic gates, along with a lot of other "connective tissue." I've been at it for four months and I'm currently estimating it will take me 2-3 years.

[v01p]

On 10/4/2021 at 11:56 AM, npicard said:

I'm making a fully mechanical computer. It's my high-level design is close to final, but there are a lot of details to work out. To illustrate a small piece of it, if you combine my rotational driving ring selector and my two-way-to-one-way, you get a logic gate. The final result should have around 100 logic gates, along with a lot of other "connective tissue." I've been at it for four months and I'm currently estimating it will take me 2-3 years.

Wow, this sounds phenomenal! And thank you for sharing the small mechanisms -- they're incredibly helpful. Really excited to follow your progress on such an awesome feat.

[npicard]

Thanks for all the nice messages, everyone. If anyone gets any use from my mechanisms, I'm happy.

Here is what I mentioned a few days ago, and AND gate.

If both green inputs are rotated clockwise (representing "1"), the pink output rotates clockwise as well. But if either (or both) of the green inputs are rotated counterclockwise (representing "0"), then the pink output will rotate counterclockwise.

I have gone through about ten revisions to get it this small, reliable, and low-friction. Unfortunately, it still has high-friction moments when the green inputs are neutral and the pink output is driven (i.e. the mechanism is run in reverse). And it can always be smaller. :)

[v01p]

Very cool and elegant! Apologies if this is a dumb question -- what's the reason for the "output needs to be driven" constraint? In practice, would the inputs ever be neutral?

[npicard]

1 hour ago, v01p said:

Very cool and elegant! Apologies if this is a dumb question -- what's the reason for the "output needs to be driven" constraint? In practice, would the inputs ever be neutral?

It's actually a very good question. I suspect most people won't care at all.

In my application, a demultiplexer will send power to only one logic gate out of a parallel bank of logic gates, then their outputs will all be joined together, so that the input to the demux has logic applied to it and is then output to one axle. When the logic gates are all joined directly, the one selected by the demux will drive the final output, but will also drive all the other logic gates backwards. Their inputs will be neutral since they were not selected by the demux.

If none of this makes sense, I suppose you'll have to wait for my final design to be revealed! Suffice to say that dozens of these will be driven backwards at one time. Friction is at an absolute premium.

[v01p]

1 hour ago, npicard said:

If none of this makes sense, I suppose you'll have to wait for my final design to be revealed!

Thanks for explaining. I tried to draw a diagram and my head exploded. It makes enough sense that I think I get why gates need to be driven backwards. It didn't occur to me that to make this work mechanically you need a neutral state in addition to the counter-clockwise 0 state. This is far more complicated than I imagined. Eagerly anticipating the final reveal :) In the meantime, loving the systemic bottoms-up building blocks. Looking forward to the demux.

[MinusAndy]

On 10/4/2021 at 7:56 PM, npicard said:

 

I'm making a fully mechanical computer. It's my high-level design is close to final, but there are a lot of details to work out. To illustrate a small piece of it, if you combine my rotational driving ring selector and my two-way-to-one-way, you get a logic gate. The final result should have around 100 logic gates, along with a lot of other "connective tissue." I've been at it for four months and I'm currently estimating it will take me 2-3 years.

Wow! That is an incredible project, and a huge build. Do you have any more wip pics?

Next time I access the inside of my Moc I’ll take pics of the sliding worm switch.

[npicard]

15 hours ago, MinusAndy said:

Wow! That is an incredible project, and a huge build. Do you have any more wip pics?

Next time I access the inside of my Moc I’ll take pics of the sliding worm switch.

Here's a WIP: a functional 3-bit adder. It takes 3 booleans (represented by the red axles rotating either clockwise or counterclockwise, on the far right side in the pic) and outputs a 2-digit binary number representing the sum of those three booleans (near left side).

Unfortunately, you can't see how it works due to it being too densely built. Ultimately, I'll probably end up doing a series of videos explaining the principles behind it.

[1963maniac]

Looking forward to that. Thanks!!

[MinusAndy]

That looks great. Very dense!

I had to google boolean.

As promised, here’s how I’ve used a sliding worm to perform two functions.

Input is via the red axle, the worm climbs along the dark grey gear due to friction added to it using the dark grey 8t gear mounted on a friction pin. 
the worm has a shuttle mechanism attached to it which has around 3 studs of linear movement. When it reaches either end of its limit the worm then turns the grey gear and the output from that is used to disengage the    power from the input. The limitation on this is that the shuttle mechanism can only move something that requires less effort than is needed to turn the grey gear. 

 

[npicard]

3 hours ago, MinusAndy said:

As promised, here’s how I’ve used a sliding worm to perform two functions.

Input is via the red axle, the worm climbs along the dark grey gear due to friction added to it using the dark grey 8t gear mounted on a friction pin. 
the worm has a shuttle mechanism attached to it which has around 3 studs of linear movement. When it reaches either end of its limit the worm then turns the grey gear and the output from that is used to disengage the    power from the input. The limitation on this is that the shuttle mechanism can only move something that requires less effort than is needed to turn the grey gear. 

Nice! But how is it that the shuttle doesn't get caught on the teeth of the z24 gear? It always seems to be a problem for my builds, so whenever I've needed a shuttle structure, I've resorted to the 1L worm gear. (A poor solution, since it has friction.)

It almost looks like the worm gear and the z24 gear are under-engaged, and the z24 gear and the z8 gear are over-engaged. Is that the case? And is it deliberate? If it's not the case, how did you solve the catching problem?

Thanks for sharing.

[MinusAndy]

I think it’s the angle of the photo. I know what you mean about catching but this one doesn’t seem to do it. I feel like the 24 tooth may be slightly high and that’s what’s stopping it from snagging on the shuttle ends. I’ll post a video when I can work out how to power it and film with my phone at the same time!

[Hrafn]

This takes me back to my freshman year analog circuits class, and not in a good way.  Flip-flops and muxes and XAND, oh my!

Which is all to say that I am very impressed by your ambition.  Many years ago I saw someone implement individual logic gates with Lego pneumatics, but I have not seen this approach before.

[2GodBDGlory]

16 hours ago, MinusAndy said:

I think it’s the angle of the photo. I know what you mean about catching but this one doesn’t seem to do it. I feel like the 24 tooth may be slightly high and that’s what’s stopping it from snagging on the shuttle ends. I’ll post a video when I can work out how to power it and film with my phone at the same time!

Could a 24T crown gear work any better?

[MinusAndy]

Ooh good shout. I’ll try that when I muster up the enthusiasm to delve that deep into the innards again!

[sm1995]

This is one of the most exciting builds I’ve seen here in a while! 

Please keep the updates coming :)

[npicard]

For this (small?) mechanism, I had the following design spec: there should be eight output axles, and they should be powered one at a time, so only one is turning at a given time.

(For those following the larger project, this is the clock, part of a computer processor that provides timing for all the other parts. It's like the conductor of the orchestra.)

Over the course of 8 or so prototypes, I came up with a system which had two pairs of Chiron shifters connected to two driving rings each. The idea was that I'd power the left side and right side alternately. However, I realized that I could just fold the two pairs of shifters into one.

I've never seen four driving rings placed in a square around one Chiron shifter before. I'd be interested to hear if anyone has used this in a build before!

In my final build, power goes into the two blue axles (simultaneously, in the same direction). The outputs are the eight pink gears. Every time the bottom shifter turns 180 degrees, the mechanism switches from powering the North and West output axles to powering the South and East axles. Meanwhile, every quarter turn of the top shifter changes which of the four gears is powered, either on the left or the right, depending on which is powered.

To get all the switching to line up, the bottom shifter should make a quarter turn for every half turn of the top shifter. This way North, South, East, and West each get a turn in the output axles on the left, before power is shifted to the right. After a full cycle, all eight gears are given an equal turn being powered.

Of course, the mechanism needs connective tissue for structure, to provide input power, to turn the shifters, to keep them aligned on 90 degree increments, and to gear down the motor. I used a PU XL motor geared down by 1:64 for the power and 1:1536 for the shifting. Just for completeness, this means each output gear turns 24 times before switching and the whole cycle takes a little less than 5 minutes.

In researching for this build, I was really surprised that I couldn't find any sequential gearboxes that "wrapped back around." As in, 1-2-3-4-1-2-3-4... They all seem to go sequentially up, then back down.

Hopefully it goes without saying that the eight outputs in this mechanism could be connected to different gearing ratios and then unified afterwards. This would create a sequential 8-speed gearbox that wraps around from 8th gear back to 1st gear. As far as I can tell, this is a totally novel mechanism. If I'm wrong and it's been done before, I would really appreciate seeing another approach!