redesigning chip: finally taking the plunge [updates]
So for a while now we've been considering re-designing chip into a better version of the platform (one will less slop, lower leg mass/inertia, and etc). We want the design to not be so different from the current design that we waste all the manufacturing we've already done but we want to fix a few key things.
Drop gear ratios from 100:1 to more like 10:1 (increase speed and reduce mass and reflected inertia)
Make the upper and lower leg the same "size," currently the lower leg is longer than the upper leg which produces some strange effects on the control
Move the actuators to the top of the leg and belt-drive the knee joint
Make electronics 18-24V and not 12V for better performance from the motors
We're still planning to keep the NEO motors with versa-planetary gearboxes. The reason for this is because other systems would be too expensive to switch over too and we can't manufacture custom gearboxes right now (too expensive). So we need to get a little creative with how we're going to package these motors (because their form factor is garbage). Few other notes.
Hopefully removing a stage of the VP gearboxes cuts the slop in half which would help with control a lot
We also want to move the "ab-ad" or Abduction/Adduction joint to joint "one" on the leg, meaning we want Ab/Ad to occur before the knee so the leg can swing out like a normal legged robot and the axis-of-rotation stays fixed (we think this will greatly improve the workspace of the robot)
Below are some initial sketches and ideas about the design!
So we started with the lower leg as this would be the easier to kind of "spec out." We decided to move the lower-leg such that it sits *under* the upper-leg as it does it most robots. This makes the whole leg stronger as we aren't cantilevering anything off the sides.
The lower leg will be a foot, a carbon-fiber rod, a "U"-shaped mount which contains the shaft, spaces, and main pulley. The sides of the upper leg will be what house the bearings to make this system pivot about the axis.
The above image is when we just started thinking about pulley designs and belt lengths and things, we moved on from this quickly as figuring out the upper-packaging was important before sizing the belts.
We started trying to find ways to package the upper joints. We first thought of having the knee motor be at the pivot and remotely driving the shoulder and knee both with a belt as positioned in the first photo. This was mechanically complicated and provided no read advantages. This would be super cool but it's not worth the mechanical complexity. Tensioning belts would be hard too.
In the meantime tho, we did find these cool bullies we could use as belt tensioners: https://www.amazon.com/Letool-coated-1-57inches-bearing-8x40x20mm/dp/B01M9G96S6/ref=b2b_gw_d_simh_1?pd_rd_w=asLAW&pf_rd_p=7f2e36e8-1a9d-42fc-bb96-93aa1395a1d0&pf_rd_r=DNJ94QWNF8Y3VQB6VNGZ&pd_rd_r=481f3e8a-80b3-4de0-afbb-cf29da872f84&pd_rd_wg=mWWQO&pd_rd_i=B01M9G96S6&psc=1
We are planning on using 15mm HTD belt and pullies from VEX PRO as they interface well with the current motors.
For the next idea, we thought it might be cool to try a "dead-axle" design where the main axle of the upper-leg is also used as the axle of the pullies for the lower leg (remotely driven by a gear and not a versa-planetary). This did not go so well because the shaft isn't wide enough to even try something like this. The NEO probably wouldn't fit either and this would probably be heavy because of the large gears involved.
The last design we came up with was to make a module that spins about the knee joint axis and connects to the upper leg. We moved the Ab/Ad join inwards on the body and this module would connect to the Ab/Ad with the 90 deg plate we already have manufactured (which reduces work on our end).
The module would look something like the above, but the tabs on the side closest to us would be bent up to attach to the upper-leg. We would make the plates out of 5052 aluminum 0.13" thickness (the same as an FRC drive base) for both lightness and strength. The only steel component (besides the gears) would be the 90deg bracket (which we may re-make).
Notes for manufacturing:
(1) we're going to ROUND OFF ALL CORNERS because we've been hurt many times by the robot
we're going to actually shave off the 1/8" of the NEO shaft that needs shaving so that the gearboxes fit better
We're going to take this one slow, we're going to start by making just the module you see above and testing out the geometry and etc to see if something like that could work. If so, we will then move on to making the "lower-leg" and then the "upper-leg" last since that is the connection between the two. We want to precisely size the "upper-leg" so that it's effective length is the same as the lower leg.
For chip-2.5 as a whole, we're going to start by making ONE LEG and learning to control that ONE LEG seeing how much weight it can support and etc, as well as how much it weighs. We will then make the other legs before moving onto making the rest of the body and then the control of the entire platform. We're going to start by trying to make the leg jump and measure forces from the actuators and etc, etc.