Quote:
Originally posted by M. Krass
If I've made some grossly wrong assumption here, please correct me. Thanks.
|
You're on the right track, but you've made a few wrong assumptions. I've done some work (and a lot of math) thinking of a way to use a differential in reverse as a gearbox.
In the end, I came up with several conclusions:
1. A differential is merely a planetary gearbox, designed differently.
2. A differential is less efficient due to the bevel gears needed.
With a differential, the housing is the input, with the two sides being the outputs. A differential allows either the two sides to have the same speed and torque. It also allows torque and speed to shift to one output for the sake of turning. One of the outputs could have twice the input speed, while the other has none.
Unfortunately, with such a design, as Paul Copioli said, the power of the second ("control") motor is basically lost.
In using a differential as a gearbox, you basically put yourself back at the Thunderchicken's problems, because it behaves like a planetary transmission. You cannot have two motors with dissimilar torque and speed characteristics hooked up without experiencing motor fighting. The motor with higher torque will stall the other motor and will win every time. The result is that your robot won't go anywhere. The only way to make such a design work is using worm gears. I then determined that the Thunderchicken's way of using the worm gears would be more efficient.
If you have two motors with similar torque and speed characteristics (or geared to match), you can, essentially, create a continuously variable gearbox similar to the Thunderchicken's CCT. Using one motor to spin the housing, and the other as a control on one of the side outputs, you can simulate a load on that output. Thus, it will behave normally as a differential, achieving a variable gear ratio anywhere from 1:1 to 1:2. The maximum torque is that of a single motor, while the maximum achievable speed is twice that of a single motor, without all the quirks.