Jan. 20th, 2022
Pixar Lamp video part 4: initial jumps, remaking bearings.
This is what happens when you ignore friction.
So today we finish building and move onto jumping — for our Pixar Lamp project to re-create the intro. For today and for the first "video," we're simply going to have the lamp jump up and down not across the screen like in the animation, we have some ideas for this but not yet. Before we get to jumping we have to do a few things.
Attach the bungees or some equivalent form of "spring" to the vertical motion of the lamp.
Fix the bearings so they aren't so "sticky" on the rail.
Find a power source for the lamp, calibrate, and start working on a trajectory.
We also want a better base the lamp can jump up and down on because the current one is interfering with all the stuff that's on the floor (like the carabiners and the steel cable). We're going to try to use what we already have lying around.
The bungees we had were way too short and also way too strong. So what we did was grab two springs and tie nylon para cord in-between them and the lamp. We did the classic "burn the end of the rope" trick to stop everything from fraying.
To solve the friction in the bearings we removed two of the four bearings from the back of the lamp, we thought this might help with the friction. The lamp now has an extra, yet small. angular degree of freedom and we hope this will not make a difference. We will certainly find out though. If that's the case we will add the bearings back in. We added some WD-40 as lubricant and this made the system a lot better.
Here's a video of it all bouncing up and down by me exciting the spring by hand. It feels like removing the other two bearings might cause issues so we're going to try to put them back in now, add more WD-40 and see what happens. The best compromise seems to be three bearings, the two in the video above and a third to constrain rotation, all of them lubricated with WD-40!
To fix the floor we ideally want a platform that goes over all the junk we needed to anchor to the floor and is made of a relatively rubber-y material to aid in traction. For now all we could find was a book so that is what we will use temporarily as we test the system.
For the power supply we will probably use our Dark Power Pro because we don't think this system will need a whole lot of power to get this lamp jumping due to the bungee.
So let's explain the control strategy here. The Pixar lamp itself is running an impedance controller which essentially means each joint is acting like a spring set about some calibration position (standing position) for the lamp. With the lamp acting as a spring and connected to the bungee spring, we are going to bounce the lamp up and down and read the joint velocities and position from the lamp into a CSV file. When we do this we have to be careful to feel we are around the system's natural resonance frequency because essentially we are trying to inject a frequency into the system that excites both the spring in the lamp and the spring in the bungee to make the lamp cyclically jump up and down.
So that's what we've been trying to do for most of the evening, playing with trajectories and trying to get the lamp to jump. Here are a few problems.
First, the friction in the rail system even with the WD-40 is absolutely massive, it's very much impeding the lamp-spring system damping the resonant peak so far we might not be able to get the peak super high like we want.
Second, there's another resonance which is the excitation mode of the steel cable itself. That's fairly low so there's time where we see that getting excited and then the whole lamp tries to shake itself to pieces.
Third, the power supply crapped out on us one time when currents went super high.
In previous videos on the makerspace page we can see the lamp works well with the original linear rail and a bungee so now we have a few options.
go back to the original system and built a better frame like we were originally going to
use something more powerful than a bungee (cheat more) by belaying the lamp by hand
come up with a better bearing that allows the lamp to slide super smoothly on the cable
We don't want to manually belay the lamp because that's just cheating. The linear rail is boring because we already did that, we want to do something more interesting. So maybe we just need a better bearing.
Taking a look at camera cable dollies, we see an interesting bearing design where there are three wheels and the string weaves in-between the three wheels and the wheels act as bearings. We're going to take inspiration from this to make a better bearing for our Pixar Lamp so it's time to walk to the machine shop. We'll machine a block that has holes for the bearings that will place the linear rail slightly farther from the lamp while also making operation a lot more smooth.
Because we don't know exactly what positions of bearings we want in this thing, we're going to add in some adjustability to the system and drill multiple holes as places where the "v-groove ball bearings" we have lying around can go.
So we used the watejet to cut some blanks, the drill press to drill out some holes, M4 taps for the screws for the bearings, M5 taps for connecting everything to the lamp itself. It took us a few hours to machine this. We designed in multiple M4 hole locations so we could change the patterns of the pullies on the block.
In conclusion, we tried everything and even then it only barely worked. The issue was when we tensioned the cable the string became so tight that the friction against the pullies was so high they didn't turn.
I think it's important to leave these "failure" style posts up here because learning from mistakes when building things is half of engineering. What we've learned today is that linear motion is really hard and there's a reason rails are designed the way they are.
We think we need to take a step back and probably re-think this, this cable system is way too high in friction and the performance difference between them and the linear rails is huge. If we got bearings that could slide on the tensioned cable we might have a change but it still isn't stiff enough to prevent the lamp from rocking back and forth.
We're going to think about all we've tried to do so far and all we've learned and go back to the drawing board for a few days and come back with some new ideas to make this work the way we want it to.
#updates #pixar #back-to-the-drawing-board