Two seasons ago, the students put together an elevator with pneumatic damping, which worked pretty well, but the fixed orifice meant that the damping force wasn’t really the right shape, as a function of velocity.
So I’m looking around for damping technologies that might work better. Plan “A” is further developing the pneumatic dampers, but I’m also wondering about electromagnetic damping. The easiest way to get that is to short the leads of a CIM.
So if we did that, had a shorted CIM, connected to nothing, would that be legal? It’s on the allowed motors list. It’s not a custom circuit, since it’s a COTS part, so rule R614 (24v) doesn’t apply. It’s not controlled by something other than the RoboRIO (since it’s not controlled at all).
Try that reasoning again, mate. COTS parts are absolutely considered Custom Circuits in FRC, per the definition in R614. Any electrical system part that isn’t core control system or actuator is Custom Circuit–this does, in fact, include wire.
The question is whether that CIM would count as an actuator or not, given that it isn’t connected to the robot electrical system. If it did, then you would be OK, as actuators are excluded. If it did not, then you would need to ensure that even at full-speed backdrive it never generated more than 24v.
But there’s an easier way to make this legal. Put it on a legal speed controller, under robot system power. Set speed controller to brake mode. Don’t command the speed controller to do anything at all in the code. Now it’s definitely an actuator.
If you use permanent magnets and either aluminum or copper you can build an eddy current brake…
You may find that “brake mode” for modern motors/drives is a LOT of damping. You could legally tune/reduce this damping with excessively long wires to the motor of the minimum allowable size.
Or the better option, design the appropriate controller for the system dynamics you want. Using a model of the system, you can explicitly solve for PID constants that give you the desired damping using a variety of different methods. An ideal P only velocity controller targeting 0 RPM will behave like a damper, and is a lot easier to tune than a mechanical damper. You also have the advantage that you can turn on or off the damper, such as using one controller to go to position, while using a separate controller with your desired damping/springiness to maintain position.
yeah, i agree that’s a good control strategy. the damper notion is intended as a kind of “soft hard stop” so the mechanism can be driven at full power without worrying about it self-destructing during over-excursion.
Shorting the wires together should handle that trivially. You’re talking about a resistance measured in milliohms. It would take an incredible amount of EMF to cause a 24 V drop across a few inches of 14 gauge wire.
If you are worried about the carriage slamming occasionally, I don’t see this being necessary as nice and beefy hardstops, while loud on impact, don’t tend to get damage things when built correctly. 4414’s 2023 robot sounds like it has solid hardstops and they get ran into quite hard, but yet the hold up.
If you are worried about the elevator slamming on every cycle, it looks like you are trying to fix a controls problem with hardware. Decent elevator controllers (aka reasonably tuned PID or better) will decelerate the elevator before hitting the hardstop, which won’t create significant impact forces.
There are many dimensions within the problem space. The mechanism could be made more robust to impact, at the cost of weight, and/or by accepting a lower safety factor. Or it could be run within the control envelope, at the cost of speed, and/or accepting a lower control margin.
The earlier student work was aimed at exploring these tradeoffs in the direction of lower mass and higher speed.
One mistake I’ve made many times as a designer is building a mechanism with hardstops that aren’t strong enough to withstand the full energy of the system. Nowadays we make that a requirement in our design review.
It might seem difficult to design strong stops within weight and packaging constraints, but spending design time on figuring those things out is easily justifiable. There are great examples of good hard stops on robots every year, but I usually look at 254’s stuff because they’re so good at optimizing for weight and stiffness.
Good endstops and general stiffness have other benefits too, like the ability to zero the mechanism using motor current; can’t do that with soft stops.
Its much less of a tradeoff than you think is is. Good control never slows down your system compared to bad control, and will actually significantly speed up your system dynamics by making use of the motor’s available power and not overshooting. Beefy hardstops aren’t as heavy as you think, because they are loaded in pure compression/shear and not bending, meaning a little metal goes a long way in making durable hardstops.
Another way to think of it: you can almost certainly do better by applying an intelligent damping voltage to your existing motor control signal, than you can do by adding an additional “dumb” motor with an electromechanically-fixed control signal. The extra motor is a distraction - if you wanted, you could simulate exactly the effects of the additional motor as a linear drag term in code without the need for more hardware.
