pic: Small CIM in wheel swerve

[AdamHeard]

Originally Posted by Andrew Schreiber

The last sentence in R39 is what makes these illegal despite being the appropriate size wire. They are not RATED for 40A. (I've been trying to find a way around this for a while). I'd be especially hesitant with the splicing.


As promised: http://www.ebay.com/itm/200661012580

As with most things: Light, Cheap, Legal. Pick 2.

Anyone know the weight on these?

[Andrew Schreiber]

Originally Posted by AdamHeard

Anyone know the weight on these?

Heavy. Compared to the 30A ones or the Mercotacs... It's been a while since I last held one, but I'd guess almost a pound.

[Dunngeon]

Originally Posted by Andrew Schreiber

There's a few out there, I don't have any links handy right now though. I'll see if I can scrounge up a part number when I get home.

Out of curiosity, why would you ignore the fact that 30A breakers are totally a thing and also require your drive to be on a 40A breaker?

With a swerve you're already limiting yourself on drive power compared to other drivetypes. Going from 40A to 30A is reducing that even further. It's not that I wouldn't use them, just that I'd prefer to have 40A slip rings in most cases.

[bkahl]

Originally Posted by Dunngeon

With a swerve you're already limiting yourself on drive power compared to other drivetypes.

Please elaborate.

[VioletElizabeth]

On the subject of weight being off--the other thing whole robot models ignore, besides nuts and bolts, is wiring, and sometimes pneumatics as well. (I have never been able to find a CAD model for a compressor.) 6 gauge wire is heavy, and smaller gauge adds up. Also, nuts and bolts are steel, which is more than 3x as dense as aluminum. (.30 lbs/in^3 vs. .098 lbs/in^3 Rachel knows these numbers by now ) We took off a handful (literally, I held them, maybe 10?) of bolts and t-nuts, that we had added as spares in our 80-20, and I would estimate it lost us a pound.

The most accurate way I found was to just weigh the base in real life, and assume that weight for that assembly. All other assemblies were a lot closer to reality.

[InFlight]

Originally Posted by Andrew Schreiber

The last sentence in R39 is what makes these illegal despite being the appropriate size wire. They are not RATED for 40A. (I've been trying to find a way around this for a while). I'd be especially hesitant with the splicing.


As promised: http://www.ebay.com/itm/200661012580

As with most things: Light, Cheap, Legal. Pick 2.

The detailed specs (link below) has these rated at 35 Amps and 50 Amps Maximum. It also has graphite contacts, so the voltage drop @ 12V would be ugly.

http://vi.raptor.ebaydesc.com/ws/eBayISAPI.dll?ViewItemDescV4&item=200661012580&cat egory=121837&pm=1&ds=0&t=1431397986744

Moog has high current contacts, but the price of four of those would be in the four figures. Not much out there that meets all the FRC needs in a small inexpensive package.

[asid61]

Originally Posted by VioletElizabeth

On the subject of weight being off--the other thing whole robot models ignore, besides nuts and bolts, is wiring, and sometimes pneumatics as well. (I have never been able to find a CAD model for a compressor.) 6 gauge wire is heavy, and smaller gauge adds up. Also, nuts and bolts are steel, which is more than 3x as dense as aluminum. (.30 lbs/in^3 vs. .098 lbs/in^3 Rachel knows these numbers by now ) We took off a handful (literally, I held them, maybe 10?) of bolts and t-nuts, that we had added as spares in our 80-20, and I would estimate it lost us a pound.

The most accurate way I found was to just weigh the base in real life, and assume that weight for that assembly. All other assemblies were a lot closer to reality.

Generally I find that on a 5lb swerve module, which is relatively screw-dense, adding screws and nutsonly adds between 0.25-0.4lbs.

[Dunngeon]

Originally Posted by bkahl

Please elaborate.

Alright, so first let me lay down some qualifiers.

I'm going to assume that the amperage never exceeds either 40A or 30A on the snap breakers, and that power scales linearly to acceleration. Further, there's no mechanical loss of power due to friction ect.

In it's simplest form, a DC motor is really just a square loop of current (multiple loops) inside of a Magnetic Field (referred to as B-field from this point on). Now the equation for Force when there is a line of current and a magnetic field is
(Equation 1)

What's important is that the Force is proportional to current flowing through the wire, assuming that the length of the loop and B field stay the same. Since the current flows in opposite directions on each side of the loop, the force produced by each side sums.


Torque is equal to r x F (radius cross force) (I'm skipping directions for the vectors because I don't have the time to draw everything out) or r*F.

Now, Power in rotational motion is Prot = T*omega (Torque*Rotational Velocity)

I'm going to pull from this webpage. You can go there to see how this next equations/graph were derived.

Power for a DC motor with respect to speed and torque end up being quadratics.



and they produce this graph



This curve is for a Green Maxon Motor, but the CIM shares a very similar power curve (as do many DC motors).

We know that Power is dependent upon both Torque and Omega. We also know that Torque is equal to rI*integral dlxB.

So, assuming that B, dl (or essentially L), r, and omega remain the same, power is affected by changes in the magnitude of current.

If we limit the CIM to 40Amps, it will produce a peak power. However, if the CIM is limited to 30Amps, it will produce a lower peak power because it has less current flowing through the loop that creates the torque.

However, our power systems aren't limited to only 40 amps, you can exceed the amperage on our breakers by quite a bit. I have no idea exactly how much a 30A vs 40A breaker will affect overall power output, but it's enough that we have noticed acceleration differences.


So barring any of the current limitations discussed above, what else puts a swerve down on power compared to other FRC drive-trains (primarily 4/6/8wd to keep it simple)? Here's the main one that comes to mind.

In a 6wd all the wheels are chained together, and one gearbox of (usually) multiple motors powers all the wheels on that side. So that when your robot is being lifted/losing normal force (under defense ect.) you don't lose the power. Since (most) swerves now days have an individual module per wheel (and a single motor), this benefit is lost. When the front is off the ground, the power being produced by those wheels is simply lost. Whereas in a 6wd, that power is still being put to the ground by the rear 1-2 wheels on each side.

Obviously, this ignores things like slippage of the wheels, drivetrain ineffiencies, and other trade-offs that make it a bit more complicated. Power also doesn't scale linearly to more acceleration iirc, so keep that in mind as well.

Edit:

Let's say each swerve module has 1 CIM, this cim produces a power P.
When all 4 wheels are on the ground, the drivetrain is producing 4P.
When 1-2 wheels are off the ground, the drivetrain is producing 2-3P.

With a 6wd, assuming each side has 2 motors, the drivebase has a power of 4P. When 2 wheels are off the ground, the base is still producing power 4P because the wheels are connected and at least one is still touching the ground. (again, this is simplified a bit)

[asid61]

Originally Posted by Dunngeon

*snip*
So barring any of the current limitations discussed above, what else puts a swerve down on power compared to other FRC drive-trains (primarily 4/6/8wd to keep it simple)? Here's the main one that comes to mind.

In a 6wd all the wheels are chained together, and one gearbox of (usually) multiple motors powers all the wheels on that side. So that when your robot is being lifted/losing normal force (under defense ect.) you don't lose the power. Since (most) swerves now days have an individual module per wheel (and a single motor), this benefit is lost. When the front is off the ground, the power being produced by those wheels is simply lost. Whereas in a 6wd, that power is still being put to the ground by the rear 1-2 wheels on each side.

Obviously, this ignores things like slippage of the wheels, drivetrain ineffiencies, and other trade-offs that make it a bit more complicated. Power also doesn't scale linearly to more acceleration iirc, so keep that in mind as well.

Edit:

Let's say each swerve module has 1 CIM, this cim produces a power P.
When all 4 wheels are on the ground, the drivetrain is producing 4P.
When 1-2 wheels are off the ground, the drivetrain is producing 2-3P.

With a 6wd, assuming each side has 2 motors, the drivebase has a power of 4P. When 2 wheels are off the ground, the base is still producing power 4P because the wheels are connected and at least one is still touching the ground. (again, this is simplified a bit)

The current limitations are still the same as those on the 40a breakers on the PDB, because it's not like the slip rings are about to explode the moment you exceed 40a (probably ). The spec sheet probably has the exact curves.

The thing about 4 wheels on the ground makes perfect sense, but unless you're planning on having a pushing match halfway on a table it's largely irrelevant. Just use sailcloth for bumber material and slide away perpendicular to the opponent's robot; presumably that doesn't take all 4 motors to do, and once you're on the ground you have plenty of power to flee.

[InFlight]

You can skip the whole lesson on Maxwells equations and simply use the quite linear DC motor equations:

Current = ((Stall Current – Free Current) / Stall Torque) x Torque Load + Free Current

Torque Load = (Current – Free Current) x Stall Torque / (Stall Current – Free Current)

Speed = - (Free Speed / Stall Torque ) x Torque Load + Free Speed

Power = 3.14 x Speed x Torque Load / 30

The efficiency is simply power out over power in,

Eff = Power / (Current x Voltage)

For a CIM motor:
Free Speed: 5,310 rpm (+/- 10%)
Free Current: 2.7A
Maximum Power: 337 W (at 2,655 rpm, 172 oz-in, and 68A)
Stall Torque: 2.42 N-m (343.4 oz-in)
Stall Current: 133A

You run peak current at the highest torque load.

As an aside, it's no mystery why six CIM drives pop their main breakers with regularity.

[Gdeaver]

If you are discussing power input and swerve don't forget the steering motors. In a pin and hard action, they will draw considerable current in addition to the drive motors. Do not worry about tripping breakers. The Roborio will invoke brown out protection long before a breaker trips. For this year no team should have had problems with brown out and drive. If we go back to a 2014 game teams better look at where on the power curve they design and look at total robot power budgeting. We developed a CVT swerve module last summer. There was no need for it this year. We will be revisiting it again this summer. With a CVT the total gearing reduction increases as the load increases keeping the drive power input in a safer zone. Read the Roborio brown out doc and enforce power management in future robot design process. Of course if we have a 2014 like game next year why would you use a minicim swerve.

[Andrew Schreiber]

Originally Posted by Dunngeon

With a swerve you're already limiting yourself on drive power compared to other drivetypes. Going from 40A to 30A is reducing that even further. It's not that I wouldn't use them, just that I'd prefer to have 40A slip rings in most cases.

Ah so it is just the "moar powah" claim.

I thought there might be more to it. I'm ok trading some pushing power for simpler modules. (As a programmer I find designing with bevel gears to be far more difficult than slip rings) If you're pushing with a swerve you're probably doing something wrong.

[Dunngeon]

Originally Posted by Andrew Schreiber

Ah so it is just the "moar powah" claim.

I thought there might be more to it. I'm ok trading some pushing power for simpler modules. (As a programmer I find designing with bevel gears to be far more difficult than slip rings) If you're pushing with a swerve you're probably doing something wrong.

Yeah, at its core it's about the power, but not exclusively for pushing. With a swerve, I'd rather avoid contact as you said. That seems to require that a robot be nimble, of which quickness/low-end acceleration seems to play a large part (2014, 2013). Maybe I'm overestimating the importance because I've never fielded a swerve before. Also, my opnions are based upon swerve in games like 2013,2014, where there was a need for quick direction changes and occasional pushing to get to a ball/traverse the field. For games like 2012 and 2015, sacrificing 10A wouldn't matter.

Out of curiousity, what else are you thinking when you said there's "more to it"?

Originally Posted by asid61

The thing about 4 wheels on the ground makes perfect sense, but unless you're planning on having a pushing match halfway on a table it's largely irrelevant. Just use sailcloth for bumber material and slide away perpendicular to the opponent's robot; presumably that doesn't take all 4 motors to do, and once you're on the ground you have plenty of power to flee.

I used an extreme case to make the point about the power loss, it rarely occurs to that extent. Generally it's just a loss of a few pounds of normal force on a few wheels.

[Andrew Schreiber]

Originally Posted by Dunngeon

Yeah, at its core it's about the power, but not exclusively for pushing. With a swerve, I'd rather avoid contact as you said. That seems to require that a robot be nimble, of which quickness/low-end acceleration seems to play a large part (2014, 2013). Maybe I'm overestimating the importance because I've never fielded a swerve before. Also, my opnions are based upon swerve in games like 2013,2014, where there was a need for quick direction changes and occasional pushing to get to a ball/traverse the field. For games like 2012 and 2015, sacrificing 10A wouldn't matter.

Out of curiousity, what else are you thinking when you said there's "more to it"?

30A breakers don't pop immediately. (http://www.snapaction.net/pdf/MX5%20Spec%20Sheet.pdf) So, a 30A breaker pulling 80A (which according to tests I've heard about is about the max you can get to a CIM given other loads on the system and the battery) will last 0.8 - 1.8 seconds. A 40A breaker will give you about twice that time. But that's assuming you're practically stalling the CIM. And it completely ignores the fact that, more than likely, the RoboRio will cut your motor power due to voltage drop.

Acceleration is also a function of code/driving/operator interface. All too often I see swerves driven like tanks. Go to point, pivot wheels, go sideways, pivot wheels again, go back to going forward. If you watch how teams who have mastered swerve (ok, I'm a 16 fan boy, I'll admit it) they tend to be much more fluid and rely less on raw power to accelerate from 0 in that direction. 254 does similar things when it comes to tank drives, it's why they seem to get away with gearing so much higher than other people.

Basically, while being able to pour more power for longer may seem like a good solution to the acceleration problem there's a fair bit more to it and utilizing it will mean your motors run cooler, your batteries last longer, and you can STILL out swerve most people.

That all being said, if I could easily find a 40A rated slip ring for comparable weight/cost of the 30A I'd choose it every time. But if I given the choice between a more complicated module (bevel), more complicated code (limit rotations), and limited continuous current (30A breaker) I'd choose limited current every time. But I'm a software engineer, I have a certain set of criteria I apply to decisions. Other folks with different backgrounds might make different decisions.

[GeeTwo]

Originally Posted by Andrew Schreiber

All too often I see swerves driven like tanks. Go to point, pivot wheels, go sideways, pivot wheels again, go back to going forward.

I strongly suspect than many teams program both holonomic and redundant (e.g. swerve) drive controllers like tank, and if so, then of course they're driven that way. By "like tank", I mean that the primary drive is either two-tread style or arcade style, with options like strafing, translating in odd directions, translating odd directions while rotating, and rotating around a fixed point being either special cases or left for the driver to figure out.

When 3946 did mecanum (Aerial Assist, not our best tactical call), we had all of the translation in a single joystick, and rotation on the left and right "triggers" of the xBox controller. This did encourage a bit of combination maneuver - you could move towards the ball or goal while simultaneously rotating. Even I was able to do both at the same time within a couple of minutes when I did a demo at my office department's spring picnic. I'd personally like to have done a joystick with a long (+/- 60 to 90 degrees) proportional twist axis, but I wasn't mentoring programming that year, and I understand that we didn't have any joysticks up to the task, in any event.

Does anyone have any insight on what 254 (or any other more "fluidly" driving team) uses to control their swerves or holonomic drives?