The rover uses a novel transmission design whereby all motors are mounted to the chassis, and drive lateral shafts via bevel gears. The shafts then transmit the drive power to the wheel axels via timing belts. But between the two sets of shafts the suspension arms are mounted. As the suspension arms swivel on the prop shaft the distance between the drive pulleys is always constant. This gives the following benefits
Unsprung mass is kept low (as the only weight at the end of the arms is the wheels)
Strength is localised (all motors mount to an aluminium channel section, which keeps the structure strong without leading to a proliferation of aluminium mounting points)
Mechanical strength and simplicity (the swing arms can be made very robust as they are single pieces, not intricate assemblies like whishbones)
Distances are constant (enables shaft to shaft transmissions such as chains and belts, which are far simpler than variable geometry alternatives like universal joints)
Packaging is efficient (as travel is constrained to one plane, everything can be packaged tightly. Timing belts are narrow and the motors can be arranged longitudinally keeping width down)
The previous iteration of the rover highlighted issues with the 3D printer style GT2 belts in that slipping was frequent at high torque. To maximise power delivery I sought to eliminate that this year. 3 proposals were considered
Timing belts (but with improved tensioning
Ladder chain drives
Simplex (bike chain)
A key driver was sustainability and affordability. Ladder change could have been a good solution but was limited to a few suppliers, and was not the cheapest option. I was worried this situation could deteriorate further in the future.
Timing belts were discounted early on due to prior experience. More aggressive teeth patterns were considered but none were as readily available as the GT2, printer-style belts
It was decided to go with a bike chain approach as this is the srongest of the 3 and readily available. The one issue was the sprockets, as they were hard to find in the 5T size I needed, so after some initial tests I realised I could 3D print these and secure them with a lock wire keyway.
These were very loud and a bit clunky, but worked well! However an issue was soon identified. Whereas the failure mode of the previous timing belts had been belt slipping, and despite this approach eliminating this issue, the new weak point was the torque experienced by the retention of the sprocket on the shaft. However whereas the belts just slipped a few teeth, this ripped the sprocket in two along the 3D printed layers -yikes!
I decided I preferred a more frequent failure mode that just reduced performance, over a less frequent one that was catastrophic failure - after all, unlike the ORT, PiWars is not an off-road competition. So back to timing belts I went!
The issue now with timing belts was that with a centre to centre distance defined by the competition rules, to maintain tension in standard sized belts I needed some rather specific pulleys. Last time I had used 20T pulleys, but I wanted to try wider pulleys as this will both increase the number of teeth, and the mechanical advantage and such reduce the force at each tooth, and so reduce slipping (I hope!) 40T is twice the diameter, so half the torque, and with twice as many teeth that's 4 times less force per tooth! However with standard belt sizes, the closest match for 40T was 37T - which being a non standard size 38T would do. However 38T pulleys aren't cheap and wanting to get the rover rolling ASAP I didn't have time for an AliExpress order!
So back to 3D printing I went!
I'm very happy with how these little fellas turned out - they are working well in initial tests and now all thats needed is to install them and test them under load!
Related Posts
