Ideal pushing gear-ratios?

[Oblarg]

So, I've been doing a bit of back-of-the-envelop calculations, and it seems that for any drive geared for a top speed of under ~12 feet-per-second, a dead pushing match (i.e. comparable to pushing against a wall) is going to be traction-limited rather than motor-limited.

If this is the case, why is the spread of gear ratios offered on most FRC shifters so big? If you're already traction-limited in a pushing match, you're not going to be getting much utility out of gearing your robot to push harder. A 2.56:1 shifter spread will result in a drive that's set for 16 ft/s in high-gear having a low-gear which seems wastefully slow - does it actually result in better performance than, say, 2:1, or perhaps even less?

I feel that I'm missing something here. Note that for all my calculations I've been assuming a robot weight of ~150 lbs, and a wheel COF of ~1 - I am aware that the former is a bit on the heavy side, and the latter is a bit lower than what you'll actually get with plaction tread.

[AdamHeard]

All things being equal (wheels, etc...) a low gear at 6 fps versus 10 fps will perform substantially better.

It's torque required (at the motor) is lower, so there is less current draw (and voltage drop), and you're operating at a higher percent of free speed. You could push all match long, and at an appreciable percent of your free speed. A faster drive will not be able to push in the same manner, or with much speed as you're very near stall on the motors.

I have not recently done these calculations, but have many times in the past. The above is a generalization, but the trend I convey is correct. The specific numbers and crossover points could be different though.

[KrazyCarl92]

Originally Posted by Oblarg

So, I've been doing a bit of back-of-the-envelop calculations, and it seems that for any drive geared for a top speed of under ~12 feet-per-second, a dead pushing match (i.e. comparable to pushing against a wall) is going to be traction-limited rather than motor-limited.

If this is the case, why is the spread of gear ratios offered on most FRC shifters so big? If you're already traction-limited in a pushing match, you're not going to be getting much utility out of gearing your robot to push harder. A 2.56:1 shifter spread will result in a drive that's set for 16 ft/s in high-gear having a low-gear which seems wastefully slow - does it actually result in better performance than, say, 2:1, or perhaps even less?

I feel that I'm missing something here. Note that for all my calculations I've been assuming a robot weight of ~150 lbs, and a wheel COF of ~1 - I am aware that the former is a bit on the heavy side, and the latter is a bit lower than what you'll actually get with plaction tread.

You're missing current draw. For a drive train with 4 CIMs and 12 fps gearing, the CIMs will be pulling somewhere in the ball park of 80 amps a piece in a pushing match (highly dependent on wheel CoF and efficiency). According to the 40 A breaker spec. sheet, this will hold for 1.5-3.9 seconds before tripping breakers. (Tripping breakers means you stop moving, which is usually not so good in a match)
http://www.snapaction.net/pdf/MX5%20Spec%20Sheet.pdf

If you want a 4 CIM drive train to pull 40 Amps per motor in a pushing match (which should hold indefinitely), the gearing will usually correspond to around 5-7 fps. If you want a top speed in high gear in the 13-18 fps range, then you're looking at a 2.56:1 spread.

Also, voltage drops due to wiring and other components become more significant as current increases. V = IR. Less current means more juice for your motors.

[AlecS]

Like Adam said, there is definitely a benefit to gearing for lower speeds when pushing. Most noticeable, will be the lower current draw. For instance, when geared at 5fps, you could slip the wheels (at your specs) without even exceeding 40a continuous breaker. At 10-12 fps, you can still mathematically slip the wheels, however you would be drawing far more current, and it would not be sustainable for long periods. Also, when you account for real world batteries, and other electrical loads on the robot, it would be debatable whether you could actually sustain wheel slippage, due to the excessive current draw.

This does not however, invalidate the idea of gearing for a lower spread (200% ish). IMO, 5-6 fps is painfully slow, as I would prefer our drivers avoid the pushing matches altogether.

[Oblarg]

Thanks a bunch for all the informative replies, everyone.

[AdamHeard]

Originally Posted by Oblarg

Thanks a bunch for all the informative replies, everyone.

No problem.

It's also worth noting that you need to figure out which camp someone is in when they discuss low gears.

Some view that low gears should be a bit faster, and used for maneuvering at times (I know 33 is in this camp, and with their solid autoshift code they have a valid argument).

Another argument is that high gear is the default gear for ALL motion, and that low gear is only used for pushing, climbing, etc... Sometimes for precise in place turns for aiming. This camp would obviously prefer a slower low gear than the first.

Although I am in the second camp, I can not definitively say it is the way to go. It's important to figure out which camp someones if preaching their information from so you can interpret it accordingly. Too many people post that X FPS is TOO SLOW/TOO FAST as if it is a hard line rule

[Ether]

Here's a very simple spreadsheet for a drivetrain with one CIM per wheel.

User inputs:

Cell A1 is the (lossless) vehicle free speed in feet/sec at the CIM's free speed of 5310 RPM.

Cell A2 is CIM current.

Cell A3 is the estimated drivetrain torque efficiency.

Output:

Cell A5 is the available motive pushing force at each wheel in lbf. Compare this to the weight on the wheel times the static coefficient of friction of the wheel to determine if the vehicle is traction limited or torque limited for the given current.

[Oblarg]

Originally Posted by Ether

Here's a very simple spreadsheet for a drivetrain with one CIM per wheel.

User inputs:

Cell A1 is the (lossless) vehicle free speed in feet/sec at the CIM's free speed of 5310 RPM.

Cell A2 is CIM current.

Cell A3 is the estimated drivetrain torque efficiency.

Output:

Cell A5 is the available motive pushing force at each wheel in lbf. Compare this to the weight on the wheel times the static coefficient of friction of the wheel to determine if the vehicle is traction limited or torque limited for the given current.

Anyone here have any data on the amount of internal frictional losses FRC drives actually have? I've always wondered about this, but never really gotten around to testing it. Just wondering how accurate the 80% efficiency figure in the spreadsheet actually is/

[Ether]

Originally Posted by Oblarg

Anyone here have any data on the amount of internal frictional losses FRC drives actually have?

It probably varies considerably. Even identical drivetrains may have significantly different friction because the friction is affected by lubrication, chain/belt tension, alignment (both toe-in and camber) of wheels and sprockets and pulleys, gear tooth tolerances, etc etc.

Just wondering how accurate the 80% efficiency figure in the spreadsheet actually is

It's just a ballpark placeholder value, same as in JVN spreadsheet.

I've always wondered about this, but never really gotten around to testing it.

See the last sentence on Page2 of the PDF Drivetrain Acceleration paper and supporting docs here.