4 Wheel Drive Too Much Traction

[r0306]

Our robot is using a 4 wheel arcade drive using the wheels given in the kit of parts. However, we are experiencing a hard time turning the robot due to friction opposing the wheels. How should we address this problem?

[AllenGregoryIV]

Can you post a picture of your robot, or give more details. It's hard to find a universal solution. You can search and find lots of threads on this but a picture will get you more useful advice.

[Mk.32]

6 wheel dropped center drive :cool:

Or you can switch the back or front to omins which is what we did rookie season and it worked wonderfully.

And a photo would help :P

[r0306]

Can you post a picture of your robot, or give more details. It's hard to find a universal solution. You can search and find lots of threads on this but a picture will get you more useful advice.

6 wheel dropped center drive :cool:

Or you can switch the back or front to omins which is what we did rookie season and it worked wonderfully.

And a photo would help :P

Sorry, should've posted that earlier.

http://sphotos-b.xx.fbcdn.net/hphotos-prn1/859989_153240221498409_1413115399_o.jpg

[AllenGregoryIV]

That's an interesting setup.
Going to 6 wheel dropped center will help but your current problem is that your drive train is geared way to high. Those look like CIMpleboxes when paired with the kit belts and pulleys don't provide enough of a reduction for the motors. (EDIT: on second glance those aren't the kit pulleys so you might be running 1:1 which is even faster)

I would recommend adding a center wheel that is about 3/16" lower than the current wheels and moving to sprockets and chain. 12 to 26 or 12 to 32 would probably be good. Though those are just guesses. You can use a design calculator to figure out your actual speeds; something around 8-12ft/s should be what you are aiming for with a 6 CIM drive. You would want to power the center wheel off of both of your gearboxes so that all three wheels will be linked and spinning at the same speed. That will also allow you to maintain all your driver train power to the ground even when some wheels are lifted off the ground.

AndyMark and VEX both sell sprockets and chain that would allow this fix for around $100.

Omni wheels will work but they cure the symptom not the cause of the problem.

[dricks]

The problem is that the combination of the length of your drive train and the friction of your wheels on the driving surface is not allowing you to turn effectively. To solve this you can shorten the steering base. Which is what is being suggested by the recommendations for six wheel drive with a raised center wheel. You could also move your four wheels closer together, but this may give you stability problems.

Another option is to reduce the friction by changing 2 or all 4 wheels to wheels with less friction. You could replace two of your wheels with slicker wheels or Omni wheels. Obviously this option surrenders a little traction.

Changing from arcade drive to tank drive( controlling each side of the drive train with a different joy stick) would not solve the problem but it might give your driver a little more control.

If your speed controllers are set to coast, you may be able to improve control slightly by setting them to brake.

You have to evaluate what will work best for you with the time, space and material constraints your team has.

Good Luck!

[JesseK]

There are several things wrong with this drive train setup, and there are multiple ways to remedy it. The root cause is pointed to by David, and the result is that your entire robot is current limited while turning (pulls more than 120A while turning @ 12V...which causes battery voltage to drop*). I can confirm Mr. Gregory's gear ratios because last year that's exactly what we ran (12:32) using higher traction 6" wheels and no issue with turning (though our wheelbase was MUCH wider than it was long...).

Start by examining the attached photo that was created with this tool (http://www.chiefdelphi.com/media/papers/2469) (not to toot my own horn -- but there are a couple of tools on CD that should come on a KOP CD or something ... namely, JVN's to start with (http://www.chiefdelphi.com/media/papers/2755)).

In no particular order, you could do one or more of the following:
Use omni wheels Use smaller wheels Offset your center of gravity Swap Gearboxes to Toughbox Mini's/Nano's (hard to tell exactly what gearbox you have -- looks like CIMple boxes since the CIM shafts stick out) Make a 'lessons learned' so your team never designs a wheel base that is longer than it is wide Use 12:32 or 14:38 sprocket reduction, in picture 2 Change the cogs on the gearbox to 16-tooth cogs with the same tooth profile

*One day I'll get around to updating my sheet to model this, now that I understand it better

[Andrew Lawrence]

I know this is a little off topic, but is your robot made out of aluminum? It doesn't look like aluminum to me. :confused:

[thmeans06]

Find your center of mass and (looks like it is toward the right side in the picture) and shrink your wheelbase by moving either your front or back wheels closer to the center mass.

[Kevin Sevcik]

I know this is a little off topic, but is your robot made out of aluminum? It doesn't look like aluminum to me. :confused:That's pretty obviously bent and welded sheet steel. I know we're all about the lightness in FRC, but has it actually reached the point where people can't recognize steel when they see it?

[Taylor]

In no particular order, you could do one or more of the following:
[list=1] Make a 'lessons learned' so your team never designs a wheel base that is longer than it is wide

Can you please explain your thought on this? It's common practice for teams to have robots longer than they are wide - 48 had a (in)famously long robot last year, and they were multiple regional winners. I agree that wider bases have less scrub, but never longer than wide?

[JesseK]

Can you please explain your thought on this? It's common practice for teams to have robots longer than they are wide - 48 had a (in)famously long robot last year, and they were multiple regional winners. I agree that wider bases have less scrub, but never longer than wide?

If the team doesn't understand the fundamentals of the forces of scrub in a skid steer drive train, then it's probably best they steer clear of longer wheel bases (w/o dropped centers) until they do. Team 25 won Einstein last year, and they typically have the same setup (8WD, all touching the ground, all long configurations from year to year). The difference is, they understand what works, what doesn't, and why.

[Racer26]

I think we're going to see alot of robots with this approximate design (chute with indexer, dropping into a linear shooter at a ~30deg angle).

This particular robot seems much heavier than it needs to be though... Why two gearboxes to a side unless you were putting mecanums on it? The whole chassis appears to be steel tube and bent sheet steel, which, might be more viable this year than ever before, with the reduced perimeter, but still.

[Lil' Lavery]

Can you please explain your thought on this? It's common practice for teams to have robots longer than they are wide - 48 had a (in)famously long robot last year, and they were multiple regional winners. I agree that wider bases have less scrub, but never longer than wide?

Wheel base, not robot/frame. I believe 48 had a dropped center, so their wheel base was still wider than it was long.

[Andrew Lawrence]

That's pretty obviously bent and welded sheet steel. I know we're all about the lightness in FRC, but has it actually reached the point where people can't recognize steel when they see it?

I recognized it as sheet steel right away, I was just curious because I was talking to someone on 253 earlier and they said they were using aluminum.

[Lil' Lavery]

I'm curious as to how those pneumatic wheels/hubs are fixed to the output shafts on the CIMs. Care to share the details?

[Ether]

I agree that wider bases have less scrub, but never longer than wide?

Yes, under certain conditions wheelbase can be longer than trackwidth

For the robot pictured, if you make the following simplifying assumptions:

1) all four wheels identical and driven independently with the same torque

2) coefficient of friction of the wheels the same in X and Y directions

3) Center of Mass lies on the the longitudinal axis, somewhere aft of the geometric center of the 4 wheels

... then the analysis becomes straightforward:

http://www.chiefdelphi.com/forums/showthread.php?t=99089


JesseK's advice is good though:
If the team doesn't understand the fundamentals of the forces of scrub in a skid steer drive train, then it's probably best they steer clear of longer wheel bases

[theawesome1730]

I know this is a little off topic, but is your robot made out of aluminum? It doesn't look like aluminum to me. :confused:

That without a doubt is steel

[Racer26]

I'm curious as to how those pneumatic wheels/hubs are fixed to the output shafts on the CIMs. Care to share the details?

If I had to guess, I'd say those are AM hubs with the 8mm bore /2mm keyway in them, and a snapring just inside them, bolted to the wheel (ie. the snap ring is on the end of the motor shaft, inside the body of the wheel.

[jblay]

The easiest fix is to put tape on your wheels and leave it there so that the tape is what contacts the ground, the won't be able to push anyone but you will be able to turn, we did this a few years ago wrapping our wheels in duct tape at the competition when we realized it was becoming a really big issue

[billbo911]

Running through the JVN calculator, your current drive train has an overall reduction of 4.67:1. It yields a top speed of 24.13 ft/sec. You need to at least twice that amount of reduction.

My I suggest swapping out the belt drive to chain with the following sprockets.

12 tooth on the transmissions and a 30 tooth on the wheels.

That will increase your reduction to 11.67:1, drop your top speed to 9.65 ft./sec, and give you enough torque to turn.

The sprockets you want are available from both AndyMarh and VexPro.

[Madison]

Running through the JVN calculator, your current drive train has an overall reduction of 4.67:1. It yields a top speed of 24.13 ft/sec. You need to at least twice that amount of reduction.

My I suggest swapping out the belt drive to chain with the following sprockets.

12 tooth on the transmissions and a 30 tooth on the wheels.

That will increase your reduction to 11.67:1, drop your top speed to 9.65 ft./sec, and give you enough torque to turn.

The sprockets you want are available from both AndyMarh and VexPro.

This is the best, most straightforward advice you've received so far. Do this.

[JesseK]

Running through the JVN calculator, your current drive train has an overall reduction of 4.67:1. It yields a top speed of 24.13 ft/sec. You need to at least twice that amount of reduction

...


This is the best, most straightforward advice you've received so far. Do this.

This doesn't make any sense. One mentor makes a conclusion about turning based upon a tool that doesn't tell him anything about how/why a robot turns, then another veteran mentor backs him up? What is this, the Apple forums?

[Madison]

This doesn't make any sense. One mentor makes a conclusion about turning based upon a tool that doesn't tell him anything about how/why a robot turns, then another veteran mentor backs him up? What is this, the Apple forums?

Bill's advice regarding a ratio change mirrors the advice you provided. I respect that you gave the team options, but swapping the drive ratio is the most straightforward fix for this team at this time. They can triage *why* their drive failed to turn later; for now, it's probably more important that they get a machine working so they can continue testing other systems.

They're also likely to have chain and sprockets around from years passed. They may not have other wheel options and moving gearboxes or CoM around is a ton of work.

[Ether]

This doesn't make any sense. One mentor makes a conclusion about turning based upon a tool that doesn't tell him anything about how/why a robot turns, then another veteran mentor backs him up? What is this, the Apple forums?

I did the calculation, using what I thought to be conservative numbers:

I assumed the Center of Mass was at the Center of Geometry of the wheels

I assumed the trackwidth/wheelbase ratio was 1/2 (it appears to be a bit better than that from the picture.

I assumed the coefficient of friction of the wheels he's using is 1.2

I assumed the wheels are 6".

I assumed a total weight of 150lbs.

I assumed 10% loss of torque in the gearbox and sprockets.

Using Billbo911's 11.7:1 gear ratio, I calculated ~180 oz-in torque required at each motor.

[billbo911]

This doesn't make any sense. One mentor makes a conclusion about turning based upon a tool that doesn't tell him anything about how/why a robot turns, then another veteran mentor backs him up? What is this, the Apple forums?

Jesse,
Instead of being offended by your tone, I think I'll build on your point.

Jesse's post (http://www.chiefdelphi.com/forums/showpost.php?p=1233725&postcount=7) will lead you to determining the root cause of the problem. It will teach you how to analyze a drivetrain and design it better in the future. That said, the solution it would lead you to at this point in the season would be impracticable to implement. It is correct, but not necessarily an optimal solution for your current situation.
You really should learn to use the analyses tools he is giving for future designs.

In the mean time, what I gave you will get you going in the shortest amount of time and least amount effort, something that is needed with the amount of time you have left. And, if you apply the information I gave you to Jesse's approach, you will see that it works, too. Please give it a try once you get the drive working to your satisfaction.

[JesseK]

I did the calculation, using what I thought to be conservative numbers:

I get it, the math is doable. The point is that advice was given in the form of exact gear ratios based upon esoteric experience rather than math. On top of that, it's not the best advice for that team because we don't know that team's experience with chain.

Replacing the belt pulley on the gearbox with a 12-16T pulley has the same net effect (the robot turns) at potentially less expense -- time & money wise.

[Madison]

Replacing the belt pulley on the gearbox with a 12-16T pulley has the same net effect (the robot turns) at potentially less expense -- time & money wise.

It also requires a different tooth count on the belt, all other things being equal. The flexibility of chain is a great asset here.

[JesseK]

It also requires a different tooth count on the belt, all other things being equal. The flexibility of chain is a great asset here.

They could also slide the gearboxes more towards the center, so long as they're not welded to the frame (can't really tell from the picture if the glare is a weld joint or not). It doesn't look like the two gearboxes on the side will interfere, but it's best to judge it hands-on.

[r0306]

We solved the problem by using free spinning low traction wheels in the front and doubling up the motors in the back for more drive power. Although the robot still struggles to turn sometimes, it is a noticeable improvement to before. We can even turn in one spot. Thanks for the help everyone!

[arun4444]

As a quick fix, i'd recommend putting all four motors on the rear wheels and swapping the fronts for omnis - just a shaft, no gearbox.

its hard to tweak four independent motors and gearboxes, especially without encoders etc.

and definetly you need to gear down to about 11-12 fps max, maybe sprockets?