Shooter Problem

[Ether]

Originally Posted by MrForbes

This is what we're planning to try tomorrow. And we'll set it up so we can play with the speeds, by using chain drive and computer speed control, eventually with a PID loop of some type.

Try P-only and crank the gain up to the sky*. If you've got enough inertia in the wheel, and you run the loop at 10ms, that's all the tuning you should need.


* this will mimic a bang-bang controller.

[Woolly]

Originally Posted by Ether

Try P-only and crank the gain up to the sky*. If you've got enough inertia in the wheel, and you run the loop at 10ms, that's all the tuning you should need.


* this will mimic a bang-bang controller.

Yeah, I'd say that you might want to go with a bang-bang controller.

If you get flack from your mechanical/fabricators for "slamming" the motor, do what I did, add a linear cushion within ~3% of your target speed.
Ex. For a shooter wheel capable of 10,000RPM when you're over 300 RPM lower than your setpoint, you go 100% power, when you're 150 RPM Lower, 75%, Dead on 50%, 150RPM Higher 25%, 300RPM Higher 0%.

This will mimic the PID you're trying to use, and unless you really know what you're doing when PID tuning, will be half the headache.

[Ether]

Originally Posted by Woolly

If you get flack from your mechanical/fabricators for "slamming" the motor...

Seriously? Are you saying there was some visible or audible manifestation?

If so, check the following and fix if necessary:

- make sure the motor controller jumper is in the "coast" position (not "brake").

- make sure you are running the control loop fast enough (10 ms is good)

- make sure your sensor is capable of giving an up-to-date speed reading faster than your control loop iteration rate

[Woolly]

Originally Posted by Ether

Seriously? Are you saying there was some visible or audible manifestation?

If so, check the following and fix if necessary:

- make sure the motor controller jumper is in the "coast" position (not "brake").

- make sure you are running the control loop fast enough (10 ms is good)

- make sure your sensor is capable of giving an up-to-date speed reading faster than your control loop iteration rate

I didn't know what I know now when I made that control loop last year, and I just dropped it in autonomous and teleop. Obviously a case of it not running fast enough. Still, even our poor bang-bang controller was better than a PID for this application.