I have been fascinated by Continuously Variable Transmissions (CVT) for many years now. I made this Lego CVT back in 2011 and since then have set out on a mission to improve on CVTs.
For quite a long time I have been considering how to design a CVT that does not work due to friction, which has no risk of slipping. Most available CVTs use friction as the core of the design, be it belts, rollers or even spheres. Thus I set out on a mission to create one that has good old solid mesh of gear tooth to gear tooth contact. From my many Google searches, this has not been done (except by BitRaptor with his/her Edyson CVT) . I could be wrong, in which case this would be an alternate approach
By having a Continuously Variable Transmission (CVT) design that transmit torque via gear mesh, the CVT would be able to transmit high torques without risk of slipping. I came up with two methods of accomplishing this. The first method and less satisfactory design is a modification of the Plate and Wheel style (like in my Lego CVT). The plate is modified to have radial slots on its top surface. The wheel is a gear that have adaptable teeth. The second more successful design is to have two "gears" that can move their teeth in the radial direction. Both methods are able to achieve a ratio range of about 1:2 to 2:1, this can be easily changed by having a bigger plate or having multiples of this system in series
This design is done in Solidworks 2013 with motion simulation to evaluate the performance of the system. No physical model has been fabricated and this is (for now) just a proof of concept. Minimal consideration was given to cost and ease of manufacturing/maintenance. In order to speed up simulation, I have choose to omit all connecting pins. The 10 tooth variant was an earlier variant that does not perform as well as the 20 tooth variant
I will go over the more successful design in better detail first before i quickly cover the less successful design Now that picture annotations are working again, I will can include information in the pictures
At first I thought of having a system based on the dual cone method but instead of a roller or belt connecting them both, I will use a set of teeth on each cone that will be able to slide along the cones surface in the tip to base direction but will rotate with the cone in the theta direction. This was meet with a dead end as I could not figure out how to hold the teeth in place and control its spread. Maybe using a device like that of a camera shuttle would work.
Building on the idea mentioned above, I came up with the idea of having the teeth slide in the radial direction of a plate like the chuck of a lathe, A pair of this will allow us to have a continuously meshing CVT To control the movement of all the teeth on a plate together, an umbrella like mechanism was used.
Step 2: The gear teeth
The gear teeth are the main component of this design and will need to satisfy these requirements
- Have a surface to interface with the teeth of next gear
- Slide radially along the gear
- Allow connection with rod that controls its movement
Once we have identified the requirements we can get to work on the teeth.
The dimension of the tooth is arbitrary and in this case selected to be 15mm x 4mm x 15.5mm (L x W x H) The design of the tooth is quite straightforward, basically create a profile of it, extrude and cut out the slot and hole for the tooth rod (tooth rod detail in a later step)
*note, the contact surface does not need to span the whole component. There are advantages and disadvantages to having it span the whole component in length
Step 3: The gear plate
This plate holds all the gear teeth in place and may be used to receive torque from the input or transmit torque to the output. The plate will need to satisfy these requirements
- Have slots the only allow the gear teeth to move radially
- Allow the gear teeth's contact surface to protrude from the plate's surface (front)
- Allow the tooth rod to extend out of the back
- Allow the actuator hub (covered in later step) to rest in the plate when teeth fully extended
The plate in this case is 140mm diameter, 20mm thick.
To satisfy the first 3 requirements, We draw a profile of the slot for the teeth in the center of the plate and extrude cut it until its close to the edge. Using a circular pattern feature we repeat this 20 times (or as many as needed) Next, cut out a circle in the center for the actuator hub to rest in.
Notes* To hold the plate in place, maybe a huge bearing to hold the entire plate to the case or rollers to support it from multiple points
Step 4: Tooth rod and actuator hub
The tooth rod is simple a rod with a hole at each end. Its requirements are:
- Connect the gear teeth with the actuator hub
- Be able to pivot about both ends
In this case the overall length is 54mm and it is 2mm thick and 6mm tall.
These were accomplished by simply making a flat strip with a hole at each end.
The actuator hub is where you connect you actuators to control the "spread" of the gear teeth. The requirements are:
- Allow the tooth rod to pivot at the connection
- Have a hole in the center that goes over the axle
- Be able to rest in the gear plate when teeth fully extended
Here the hub is 30mm in diameter and 10mm thick
Similar to to the gear plate the profile of the slot was cut out and the pin extruded, then both were circular patterned 20 times (or as much as needed) Notes* For the manufacturing of this part, I was thinking the pins will drop into a semicircular hole from the top after being fixed to the tooth rod and a cap will hold the pins in place (might be a little hard to picture)
Step 5: Additional parts used for testing/simulation
Some additional parts used:
The axles and brackets hold everything at the proper position. The tubes together with more brackets ensure the actuator hubs move together. (Optional)
Note* Notice the rate the teeth move is not linear to the rate the hub moves. More on this in problems and improvements.
Step 6: Assembly
- Insert each gear tooth into the slots of the gear plate.
- Next add a tooth rod to each tooth. (should be much quicker with mate references defined)
- Add the actuator hub and axle
- Join all the tooth rods to the actuator hub
- Use the brackets to hold the axles apart at desired spacing (adjust by tuning the brackets)
- ensure the gears mesh and the gear plate does not slide along the axle (use distance mate)
- Connect both actuator hubs together (Skip if you want to control each hub individually)
Feel free to head over to my other instructable Steerable Continuous Track, the second step gives more information on the use of mate reference
Step 7: Running the simulation and data processing
To evaluate the performance of the design, I ran a series of simulations using Solidworks' Motion Simulation add-in. Two methods of controlling the actuator hub (and hence teeth spread) were employed.
The first method simply links both actuator hubs together rigidly and we move both hubs together to "shift gears", this method is easier to simulate and may also be easier and cheaper to implement in a real life model. In the second method, both actuator hubs are controlled independently of each other, this allows better flexibility and control of the teeth spread and gear ratio. This may not be difficult to implement but would require an electronic control system.
Here I will cover the first method, the second will be in the next step/page.
1)First off, the usual: Activate the motion analysis add-in and create a new motion study
2)Next create contacts (use contact groups), group all the teeth of one "gear" together and all the teeth in another
"gear" in the contacting group.(You can select everything under contact, but it would be awfully inefficient) Set the rotating motor at the drive axle and choose a reasonable speed for testing (i used 50rpm)
3)Start simple by letting it run at the lowest possible ratio, then at 1:1 ratio and finally at highest possible ratio. You may need to set a linear motor running at 0mm/s to ensure the hubs do not move.
4)Next, you will need to decide how you want to run your simulation, two good set ups i used are 2-5-2 and 2-10-2. 2-5-2 means running at lowest ratio for 2 seconds, gradually shift to the highest ratio over 5 seconds and let run at highest ratio for 2 more seconds. 2-10-2 is similar except it shifts over 10 seconds. Use a linear motor to move the hub. Since both are linked together, only one linear motor is needed, I used a constant velocity motion. (some simple calculations may be required to determine what velocity to use)
5)Plot a graph of the output angular velocity over time and export the data for processing. I processed the data and cleaned up the noise using Matlab. I wrote a custom .m file to visualize the performance.
I will attach the .m file, feel free to view it if you are interested. (The algorithms I used are of my own design as I am unable to recall the proper methods my lecturer taught me)
*Note that the relationship between the displacement of the teeth radially and the displacement of the hub along the axle is non-linear, thus a constant velocity hub will not yield a linear change in gear ratio Go to the next step/page to see how I controlled both hubs independently to ensure the gears mesh properly at all times
Step 8: Simulation with independently controlled actuator hubs
As mentioned in the previous step/page, the gear shift rate is not linear. And you may notice the teeth may lose contact (de-mesh? can someone tell me the right term). To overcome this, we will need to control each hub independently.
note* I believe this is not the best way to do this but its the only method i could think off
1)Similar to the linked hub method covered in the previous step/page, turn on the add-in and define the contacts.
2)This time you would want to "attach" the linear motor to the teeth itself and make sure the hub and plate do not rotate (either fix the plate or define a parallel mate for a teeth and a fixed face). Next you would want to record the linear velocity data for the hub and save it. Do this for opening and closing the teeth spread as both are different.
3)Now set up linear motors for the hubs as in the previous step/page. But this time make sure the velocity is not constant and follows the data you collected earlier. You should run it without rotating the drive axle first to test it.
4)Once you are able to move the teeth spread as intended, add in a motor at the drive shaft and you did it!
5) Similar to the previous step/page, plot and save the data then run it through some processing (same .m file can be used)
Step 9: Problems and Possible improvements
Here are some problems and possible improvements I have identified
Strength of teeth
One of the advantages of this design is that it should be able to transmit very high torques because it has no risk of slipping as it uses gear teeth. However, the design as it is will have its teeth stripped off if made from conventional materials should it be faced with torque large enough to cause slipping in modern CVTs.
With the large number of parts and moving components, manufacturing and maintaining it will not be the only nightmare. Wear and tear on the parts will be a very big problem, there will also be many points of possible failure.
Gear de-meshing (if actuator hubs are rigidly linked)
As you might be able to see from the videos, the teeth slide in/out faster closer to the center of the plate, this cause the teeth to move away from each other. Here I "solved" this problem by using teeth with very very long contact surface.Alternatively and more preferably, we can control the two actuator hubs independently and electronically to allow better meshing (minimize the teeth sliding against each other). *Note, I wrote this before figuring out how to control the hubs independently the way i wanted in solidworks motion analysis.
Step 10: Design 2, Modified disc and wheel
This design has a wheel modified as a gear with stacks of retractable plates that act as the gear teeth. Here each stack/tooth consist of 5x 1mm plates. All the plates are able to fully retract into the wheel.
The disc is modified to include slots for the teeth to catch on. Each slot is also modified to include a groove that help the plates retract as the wheel slide along the radial direction of the disc.
This design does not seem too promising, thus I did not run it through the motion analysis and took a couple days off thinking about CVTs before coming up with the design detailed in previous steps.
Step 11: Thank you!
I hope you can learn something from this instrucble and be inspired to work on your next project.
Like everyone, I am still learning and would appreciate constructive feedback. If you have any questions/clarifications, please feel free to post it in the comments, I will try my best to help and other readers will also be able to help too. Thank you! Check out the playlist for a couple of extra videos