Using 2 CIMs to lift 250lbs

[Ether]

Originally Posted by MrForbes

We're planning to lift one 150 lb robot with one CIM, using 64:1 reduction, and a 1" radius spool.

We expect it to take around 5 seconds to raise the robot a bit more than one foot.

0.2     linear ft/sec
150	pounds load
30      ft_lb/sec power required
40.7	Watts

1 CIM can easily provide 40 Watts output power, with lots of margin.

A wide range of gear ratios will work.

[Ninja_Bait]

Originally Posted by JamesCH95

How do you know? I'm curious to hear how you determined that. A calculated 30A is really gentle on breakers, even if you were pulling 60A/breaker for 1s you wouldn't be close to tripping them.

Well to be fair, this is my gut instinct/murphy's law speaking. Mathematically, yes, this is a sound design. But I am anticipating worse losses from friction in our elevator and our gearbox, there's a solid risk that our motor pinions will jump teeth, our elevator rigging is a likely weak point, the design we have for our robot 2 holder puts us at a mechanical disadvantage, and at the end of the match our battery won't be giving us 100%. Plus you have to consider that lifting a robot is a dynamic scenario, while the math we use is steady-state. I expect that the compromises we had to make in our actual assembly will quickly eat up any margin we have.

Anyway, I've probably just seen too many robots go wrong in my career...

EDIT: The point is that you can't just say "oh motor gearbox doo doo dah jvn calculator" and spit out a working system. You have to think outside of the (gear)box, as it were.

EXTRA EDIT: I haven't told my students about any of these fears yet; I'm hoping it just works and then they will believe anything I tell them from here on forward.

[Fields]

Originally Posted by Ninja_Bait

We're doing two CIMs with a 51:1 reduction and a Ø3" drum. In the magic mathematical universe (assuming 80% efficiency from the 3 stage Toughbox), we pull off a double climb (300 lbs) in one second without blowing our 40A breakers - but we are right at the limit!



Work in rotation is torque times the number of rotations in radians. It is a bit of a trick there, since radians is a dimensionless quantity.

Dang, nice catch there.

What I should have said was:
Work = Torque x Linear Travel Dist. / Radius of the wheel

Revolutions = 2pi radians
Revs required = LTD / (2pi x Radius)
Work - Torque x Revs x 2pi (2pi cancels out leaving LTD / Radius)

[MrForbes]

We did the math last year on our rope climber, which was a very rare one that actually climbed the rope (the 1" field rope). We had an incredible amount of friction, so we had to add another motor, even though the math said it would work fine, and hardly load the motor at all.

Yes, you do have to take the real world into account.

[Ninja_Bait]

Originally Posted by MrForbes

We did the math last year on our rope climber, which was a very rare one that actually climbed the rope (the 1" field rope). We had an incredible amount of friction, so we had to add another motor, even though the math said it would work fine, and hardly load the motor at all.

Yes, you do have to take the real world into account.

I'm curious, were you able to pinpoint a source of the friction, and how are you avoiding it this year?

[JamesCH95]

Originally Posted by Ninja_Bait

Well to be fair, this is my gut instinct/murphy's law speaking. Mathematically, yes, this is a sound design. But I am anticipating worse losses from friction in our elevator and our gearbox, there's a solid risk that our motor pinions will jump teeth, our elevator rigging is a likely weak point, the design we have for our robot 2 holder puts us at a mechanical disadvantage, and at the end of the match our battery won't be giving us 100%. Plus you have to consider that lifting a robot is a dynamic scenario, while the math we use is steady-state. I expect that the compromises we had to make in our actual assembly will quickly eat up any margin we have.

Anyway, I've probably just seen too many robots go wrong in my career...

EDIT: The point is that you can't just say "oh motor gearbox doo doo dah jvn calculator" and spit out a working system. You have to think outside of the (gear)box, as it were.

EXTRA EDIT: I haven't told my students about any of these fears yet; I'm hoping it just works and then they will believe anything I tell them from here on forward.

Originally Posted by MrForbes

We did the math last year on our rope climber, which was a very rare one that actually climbed the rope (the 1" field rope). We had an incredible amount of friction, so we had to add another motor, even though the math said it would work fine, and hardly load the motor at all.

Yes, you do have to take the real world into account.

No one is suggesting that the 'real world' be ignored.

However, when I see a design that looks quite reasonable and effective but is being called 'right on the edge' it piques my curiosity. Guessing at failure modes like 'the CIM pinion will skip' is even more interesting.

[Ninja_Bait]

It was a totally fair question and I realized I shouldn't have said that the design was right on the edge in the same way that we were discussing the original question. It's also true that I'm making many worst case assumptions that haven't been proved out yet.

The CIM pinion skipping is because we were forced to put it at the very end of the motor's output shaft, and I don't feel that good about the tolerances in our mounting plate. I think it's a totally valid concern in our case. Oh, also we're not using a planetary gearbox so it's not as stable. That was a resource driven constraint.

We are about to test it this afternoon so I'll come back and let you know how good my gut's guess was.

[MrForbes]

Originally Posted by Ninja_Bait

I'm curious, were you able to pinpoint a source of the friction, and how are you avoiding it this year?

Yes, it was due to using the field rope. This year we are using a little tiny string, which has a straight shot. Hopefully we will have enough code in the robot to test it today...

[Jon Stratis]

Last year we ran a CIM on a 20:1 reduction on a 1" spool to climb. Our only issue happened when the strap traveled to far to the side and got bound up on itself, jamming the whole thing. oops.

We're going with the same 20:1 reduction this year, with a much better spool that won't have that same problem. That gives us a calculated 1 foot travel time of under 1 second. The calculator says it'll have a current of about 50A, which the breaker datasheet says will take 5 seconds before tripping... that'll be enough to let us get up and engage our ratchet to lock us in place. If we have any issues, we can easily swap out one of the stages to bring us up to a 25:1 reduction, which brings the current down just under 40A, according to the calculator. I don't expect to need that, though.

This is going to be the third time we've climbed with roughly same reduction and same spool size, so we've got a lot of confidence in it!

[Ninja_Bait]

Alright, the proof is in the pudding: video link

This is our robot. Again, single CIM, 51:1 reduction, 3" drum. We'll be adding a second motor later so we can do a double lift.

The predicted speed was 12 inches in just under 1 second, you can see that it takes probably twice that. The robot is underweight in this video. I'm running the motor right off of the battery and the battery is definitely on the low side of 12v. There's some jumpiness because I can't hold the connector in place. Mmmmaybeeee with a juiced up battery and real motor control we could get 1 second but I really doubt it.

On the other hand, it's a lot smoother than I expected, so take that, Mr. Murphy!

My conclusion: doing the math right helps a lot. Having margin helps even more.

[JamesCH95]

Video doesn't load.

If you're just holding a wire to a battery connector (which it sounds like) you're introducing a huge contact resistance AND damaging your battery's Anderson connector with arcing. We use a PDB hardwired to motors and the battery for this sort of testing because 'hold the wires in place' yields piss-poor performance.

[Ninja_Bait]

Originally Posted by JamesCH95

Video doesn't load.

If you're just holding a wire to a battery connector (which it sounds like) you're introducing a huge contact resistance AND damaging your battery's Anderson connector with arcing. We use a PDB hardwired to motors and the battery for this sort of testing because 'hold the wires in place' yields piss-poor performance.

Huh, I'm going to have to blame your browser for that one... I just double checked that anyone should be able to at least download the video.

What you're saying about the wiring situation sounds true. Even so, I would be cautious about expecting perfect performance from the finished robot. I guess it makes me a downer, but I find that it's best to be prepared for things to go wrong, even if everything goes right.

EDIT/Extra thought: I don't think I've ever been in a situation where the electrical performance was the issue more than the mechanical design. Again, maybe just me.

[Cothron Theiss]

Originally Posted by Ninja_Bait

Huh, I'm going to have to blame your browser for that one... I just double checked that anyone should be able to at least download the video.

What you're saying about the wiring situation sounds true. Even so, I would be cautious about expecting perfect performance from the finished robot. I guess it makes me a downer, but I find that it's best to be prepared for things to go wrong, even if everything goes right.

EDIT/Extra thought: I don't think I've ever been in a situation where the electrical performance was the issue more than the mechanical design. Again, maybe just me.

Video wouldn't load for me either.

[JamesCH95]

Originally Posted by Ninja_Bait

Huh, I'm going to have to blame your browser for that one... I just double checked that anyone should be able to at least download the video.

What you're saying about the wiring situation sounds true. Even so, I would be cautious about expecting perfect performance from the finished robot. I guess it makes me a downer, but I find that it's best to be prepared for things to go wrong, even if everything goes right.

In 2013 we tried testing our climber with a 'touched wire' setup and were really disheartened. We then wired up to a PDB and everything worked great. I think you might be please by the performance increase.

My original point in all of this was that your design had good margin, so I wouldn't call it 'on the edge' but safe. Clearly it worked in sub-optimal conditions, which is a great result.

[Ninja_Bait]

hmm... well here's an alt link

James, you and I are in agreement that the design is good. I also agree that saying the design was "right on the edge" was hyperbolic of me. It amuses me that you're trying to convince me that the finished product is even better than I expect, when it's my team's finished product. It's a nice little ego boost and I hope you're right.

I hope our conversation has been helpful to OP and given them plenty to consider in their design.