For a project that I am working on, I need a mechanism that automatically switches a rotating axle clock-wise then counter-clock-wise, then back to clock-wise et cetera. All solutions that I could find on-line, and that I have used myself in other projects, realize this through a rotating crank with rod and a piston or slider. The problem with such solutions is that the clock-wise movement does not last exactly as long as the counter clock-wise movement. That is, they are asymmetrical. In this page, I introduce a symmetrical periodic auto-reverse switch, which is based on a sliding worm-wheel.
If you are only interested in the building instructions, you can download a .pdf here.
Asymmetrical periodic auto-reverse switch
Let me start by showing what I mean with the asymmetrical solutions one can find on-line. The video below shows a variant of such a design. It is a bigger and less compact version of this design by 2-in-1at Rebrickable.com
On its right hand side, there is the crank ( the big gear ) with rod. On the left is a slider which in turn drags a gear-selector from left to right. The gear box transforms the input to the clock-wise or counter-clockwise movement. This is also a standard design that one can find on the internet. A difference is that the gear selector is quite a while in ‘neutral’ while it is being dragged from the left to the right. This is a feature that I need for the project that I am working on, but it does not change the unsymmetrical character of the design.
Now that you have seen the basic idea, have a look at the output axle with the yellow part in the video below. You can see that its direction changes, but that after having changed direction twice, it does not end up in the same position as where it started. This means that the switch turns longer in one direction than in the other.
With the insect walker project, I ran into the same problem. It is problem 1, on the insect walker description page, where I also explain why the asymmetry occurs.
I decided to try a different solution.
Worm gear based design
The inspiration for the basic idea of my solution comes from Sariel’s description of the worm gear in his book The Unofficial LEGO® Technic Builder’s Guide ( 2nd edition ). The basic idea is as follows. Worm gears ( with opening type II ) can slide over an axle. When the axle rotates and the worm gear is meshed with a non-moving gear (it may for example be connected to an up-gearing gear box), the worm will be pushed along the axle. When the worm gear can not slide any further, it will then exert force on the the non-moving gear (which may then make the gear box run). One can use this effect to make a switch flip. If that flip causes the axle to turn in another direction, the worm gear will slide in the opposite direction. This may require some force, but since the worm gear has a very low ratio ( a full turn on its axis will result in turning a meshed regular gear only one tooth ), it can exert a lot of force.
The figure below shows most of the moving parts that make the switching happen.
The yellow lift arms are part of the switch. It is difficult to see, but it is slightly tilted to the right ( that is clock-wise ) [ 1 ]. This cause the input to the system to go through the black axle [ 2 ] , which in turn makes the worm gear slide to the left [ 3 ]. The worm gear in the picture has just reached the end of its move to the left and is about to exert force on the 8-teeth gear, which eventually will make the axle driving the witch [ 4 ] turn to the left ( counter clock-wise ), means that the switch is flipped.
When the switch is flipped, there is a fraction of a second that it is in ‘neutral’ and the input is not driving anything. The shock absorber ( not in the picture above ) however makes sure that the flip is completed, and after that makes sure that the switch is pushed towards the side so that its gears mesh well with the receiving gears. It means that the worm gear not only needs to move the switch, but also to overcome the pressure of the shock absorber. If the shock absorber does not apply enough pressure, the gears may slip. The slipping point also depends on the resistance on the output of the system. The more torque it needs to deliver, the more pressure on the switch is needed. I have not tested what is the maximum pressure that the worm gear can overcome.
How well does it work?
See for yourself. To make the videos on this page, I have run the mechanism for minutes, and it has not skipped a beat. One can also see that the switch indeed is symmetrical. After a full cycle the yellow lift arms attached to the output return to the same position as where they started.
Adjusting and even making the system asymmetrical
I have not experimented with it, but this design seems quite adjustable to your needs by replacing the gears in one, two or three different places.
Starting at the front, i.e. the output side, array of gears. This can be replaced to create different speed/torque output, while the switch speed remains the same. If you only change this array, then it needs to be symmetrical so that the outer left and outer right axles ( the ones connected to the system’s input ) turn at the same speed.
Next, the tan gears driving the worm gear can be replaced to achieve a higher or lower switch speed. This means that the axles delivering to the output will switch direction sooner or later. Replacing the tan gears also need to be symmetrical if it is the only set of gears that is being replaced. Replacing these gears may involve changing the structural design as well to make space. For example when one replaces the 20 tooth gear with a 36 tooth gear.
Thirdly, the gears at the back, that connect to the switch, can be replaced with a different set. Since these gears drive both the worm gear and the output at the same time, I think the left and right side can be a-symmetrical in principle. So turning in one direction can have both a different switch time and a different output speed/torque than turning in another direction. However, the differences can not be too big. After all, in case of such a-symmetry, notice that what is down gearing in one direction, will mean up gearing when the system turns in the other direction. Depending on the strength of the engine the up gearing can be too much.
Lastly, I assume that in principle there is another way of achieving a-symmetrical output, which is to change both the array of gears at the front and the tan gears that drive the worm gear. However, they should be changed so that the a-symmetry is the same between these two sets. Meaning the gear ratio from left to right ( which is 1 in the current design ) should be the same for the tan gears as for the front gears.