RS775 Gearboxes

[Kevin Sevcik]

Originally Posted by rahilm

~390 if I did my math correctly; ~(voltage)*(stall current)/4 = (12)*(130)/4 = 390 (not factoring in motor inefficiency)

Also, the 2010 FPs had roughly 190w. This year's have 290w. (Both figures from the FIRST motor curves)




The reduction we're going with is a 512:1 (256:1 with banebots gearbox, and then 2:1 with sprocket). This gives us ~576 lb-ft of torque if you go with 65% gearbox efficiency, which is more than enough to lift it, while we still have 25RPM, which is 150deg/s.

DON'T DO THIS. That is, unless you've carefully considered the following.

Banebots suggests a maximum torque output of 85 ft-lbs for the P80 CIM gearbox. If your mechanism is actually going to require anything close to 576 ft-lbs of torque, you'd destroy a P80 gearbox rather quickly. And P80s are tougher than the P60, which is smaller with an aluminum ring gear. BB doesn't list the max recommended torque for the P60s, but it's going to be less. At least half, possibly a quarter. So I strongly urge you to look at the maximum safe output of your gearbox and make sure you plan your gear reductions accordingly. You may want to decrease the reduction in your gearbox and add an additional reduction stage of sprockets. 256:1 gear reductions provide surprisingly little torque once the gears have failed.

[Chris is me]

I was under the impression that the P60s were counterintuitively stronger than the P80s. If the P60s can output less than 85 ft-lbs of torque, they're pretty much completely useless to me.

[Tom Line]

Originally Posted by Chris is me

I was under the impression that the P60s were counterintuitively stronger than the P80s. If the P60s can output less than 85 ft-lbs of torque, they're pretty much completely useless to me.

Remember - if you have a 15 lb end effector at 4 feet from your pivot, that's 60 ft-lbs. If you then have a 2:1 sprocket and chain reduction leading into your P-60, then it will never seen more than 30 ft-lbs of torque statically. Dynamically is dependent on the speed you move it of course.

[Chris is me]

Yeah - this is what happens when you do math in inch pounds and are too tired to remember to convert. The gearboxes should be fine for me.

[rahilm]

Originally Posted by Kevin Sevcik

DON'T DO THIS. That is, unless you've carefully considered the following.

Banebots suggests a maximum torque output of 85 ft-lbs for the P80 CIM gearbox. If your mechanism is actually going to require anything close to 576 ft-lbs of torque, you'd destroy a P80 gearbox rather quickly. And P80s are tougher than the P60, which is smaller with an aluminum ring gear. BB doesn't list the max recommended torque for the P60s, but it's going to be less. At least half, possibly a quarter. So I strongly urge you to look at the maximum safe output of your gearbox and make sure you plan your gear reductions accordingly. You may want to decrease the reduction in your gearbox and add an additional reduction stage of sprockets. 256:1 gear reductions provide surprisingly little torque once the gears have failed.

Thanks for letting us know. We will probably switch to a 128:1 planetary and 4:1 sprocket in that case. At most we would need 130 lb-ft of torque to turn the arm. With 4:1 that means the gearbox should only need to put out 32.5 lb-ft of torque. That should be fine I think, but I'll ask Banebots if they have that info on the P60s.

[Chris is me]

Be sure not to just model for torque but for speed as well. An arm that rotates 180 degrees in 1 second is not very useful.

[rahilm]

Originally Posted by Chris is me

Be sure not to just model for torque but for speed as well. An arm that rotates 180 degrees in 1 second is not very useful.

Based on our prototyping, watching some old 2007 videos, and current consensus on design, we've found that speed close to optimal. Of course, it's only week 2, and there's a lot more testing to do, so our plans may change quickly.

[Ian Curtis]

Originally Posted by rahilm

Based on our prototyping, watching some old 2007 videos, and current consensus on design, we've found that speed close to optimal. Of course, it's only week 2, and there's a lot more testing to do, so our plans may change quickly.

Sanity check: Strap a five pound weight on the end of a 2x4 and swing that from a pivot point about five feet from the end 180 degrees in a second.

Then try to precisely place a tube on a rack with that rotation rate.

I haven't done it, so maybe I'm talking out of my butt here... but that seems pretty darn fast!

[rahilm]

Originally Posted by iCurtis

Sanity check: Strap a five pound weight on the end of a 2x4 and swing that from a pivot point about five feet from the end 180 degrees in a second.

Then try to precisely place a tube on a rack with that rotation rate.

I haven't done it, so maybe I'm talking out of my butt here... but that seems pretty darn fast!

I guess I haven't been clear enough. We probably will be going with something slower for human control (as I said, week 2 just started, lots of testing to do still), I don't know of many people who would be able to control something that fast.

[pfreivald]

What's wrong with 180 degrees in one second in this game (specifically)? You have half (or all) of the field to drive before you are in position to score, after all!

[Chris is me]

Originally Posted by pfreivald

What's wrong with 180 degrees in one second in this game (specifically)? You have half (or all) of the field to drive before you are in position to score, after all!

It's far too fast to be easy to control, in my opinion. Most arms have about 90-100 degrees of travel. I"m thinking 45 degrees per second is much more reasonable.

[rahilm]

Originally Posted by Chris is me

It's far too fast to be easy to control, in my opinion. Most arms have about 90-100 degrees of travel. I"m thinking 45 degrees per second is much more reasonable.

With some degree of automation (PID, specifically) and the proper control systems, you could go with a much faster speed without losing much control.

[Ian Curtis]

Originally Posted by rahilm

With some degree of automation (PID, specifically) and the proper control systems, you could go with a much faster speed without losing much control.

Because I trust software about as far as I can throw it, and since it has almost no mass, I can't throw it very far.

I know there are a lot of great programmers in FIRST. I also know that teams are always pushed right up to a deadline, and the software teams get very little time to test. I also know a slow arm is a lot easier to stop, and if your arm is going 180 degrees/second and you miss a limit switch for some reason, something is going to break. There are a lot of great teams in 2007 that didn't actuate at anything close to 180 degrees/second. There probably are some that did too, but I'm a lot more comfortable with a slow arm that I can fall back on a human to control, than needing a relatively complex piece of software to control.

Why would you use an encoder? It is simpler to use a potentiometer since your are limited to probably less than one revolution, no?

[pfreivald]

A. Trimpots drift. Encoders don't.
B. The same idea could be replaced with "why do this in the age of the limit switch?" and the objection is the same (though with slightly different hardware and software, without quite so much elegance).
C. With 20 minutes of LabView training, I was able to bring an arm up to whatever height I wanted to by pushing the appropriate button.
D. If you haven't made things autonomous, then you can't score in autonomous...

[Jason Law]

Banebot just emailed me back:


Hello,

The new CIM-U-LATOR gearbox and P60 mount for the RS775s will be available for order Wednesday.

Thank you,

BaneBots Sales


Jason Law