Flywheel Consistency

[lpieroni]

@sigilwig44

This is actually what we do! It seems to me to make sense, if you just ran it for one wheel, chances are one will be spinning quite a lot faster then the other, unless your gearboxes are somehow built perfectly identical…

One thing you can do is figure out which flywheel has more friction, then make that the “master” flywheel. It gets a control loop to reach a certain velocity. Then, the “slave” flywheel, the other one, gets a control loop to match its velocity to the master flywheel. The only difference is, instead of using

error = targetVel - currentVel;

you would use

error = masterVel - slaveVel;

[sigilwig44]

@lpieroni

The Ki value is what you multiply your integral by. Generally, your integral gets really large, so something like .01 is a good place to start. Another thing to add if you notice your flywheel doesn’t decelerate, and the integral is really big, is limiting the integral windup. Otherwise, your integral eventually can get to be large enough to maintain 127 motor power, regardless of how far beyond the target velocity (or position) you are.
The C constant is the power the motors receive when error is 0. Because it requires force to maintain the position, and P, PI, PD, and PID gravitate toward 0 power when error is 0, you need to adjust what motor power it is. I suppose with a PI loop, you might just need to tune your I really well, and you don’t need C then. However, I haven’t tried this. I use the letter C to refer to it because this is a constant, and it should only change when you change the target velocity. It is experimentally found, and that is one of the benefits of TBH; it finds what motor value it settles at eventually, although it takes longer.

Ok so I think I know what to do except for one thing: what should I set C to? How do I do that?

[Nehalem]

@sigilwig44

Ok so I think I know what to do except for one thing: what should I set C to? How do I do that?

Well, C is what you’ll be using to maintain 0 error after the PID loop reaches 0 error. This is something you’ll have to tune on your own. Spend some quality time with the debugger window and robot to figure out what C needs to be in order to maintain 0 error.

You’ll set C just like any other variable by using an assignment operator “=”.

[sigilwig44]

Alright, so it seems I’ve been able to figure it out without doing any other things, it’s been just messing around with variables, and we seem to have it pretty good, about 18/24 balls, which I’m happy with… Thanks guys for your help