Blowing fuses/tuning drive train

Hi,

Our rookie team implemented a gear shift for the drill motors, but in high gear, we blow fuses unless we are going pretty well straight. After watching Rage at the Nationals, zooming around getting balls, we have ambitions of being able to drive in high gear. (We also pick up balls.) Can anyone give us some tips on tuning our drive train? Any other ideas?

We have 12 teeth gears on the motors and 60 teeth gears on the wheels. Our gear shift uses a single servo to switch both motors.

I saw a message somewhere about using a can of coolant to cool things down. Does anyone else have experience with that?

Thanks for any help you can give us. (We are going to the Southern California Region Summer Classic on June 29th.)

Doug Hogg
Team 980

Doug,
Whenever you are blowing fuses in turns and not when running straight it is due to high friction in your drive system to the carpet. This is almost always the case with tank treads and to a lesser extent with tank style four-wheel drive. The turning friction is exteremly high as the robot is trying to drag the treads sideways. The only way to fix this is to reduce side friction or reduce current in turns. (i.e. lower speed, slower turns, larger turning radius, etc.) It should be an easy fix in software to slow the drive when it senses a turn without shifting gears. Hope this helps.
Good Luck

easiest way around this problem is to just get a good driver who knows only to turn when you’re in low gear, or just to turn slowly/in a large radius. that’s what we did this year.

Our team published our gear shifting trannies after each season

Single Motor (drill)

duel motor with the drill/FP combination

This year we used a shift on the fly, Andy Baker is working on the white papers for that and look for it in august

*Originally posted by Al Skierkiewicz *
**Doug,
Whenever you are blowing fuses in turns and not when running straight it is due to high friction in your drive system to the carpet. This is almost always the case with tank treads and to a lesser extent with tank style four-wheel drive. The turning friction is exteremly high as the robot is trying to drag the treads sideways. The only way to fix this is to reduce side friction or reduce current in turns. (i.e. lower speed, slower turns, larger turning radius, etc.) It should be an easy fix in software to slow the drive when it senses a turn without shifting gears. Hope this helps.
Good Luck **

We also had that problem with our robot, we were drawing something like 120 amps of current we had fooled around with diffrent gear ratios but none of them seemed to work so in the end, becasue our robot had six wheels to start with we just took off two and popped on some roller casters but that would only work if you had six to start with.:smiley:

Here is the thread where i mentioned the coolant spray stuff…

http://www.chiefdelphi.com/forums/showthread.php?s=&threadid=3376&highlight=tripping

As stated earlier, drawing too much current while driving is a result of too much friction in the drive train while turning. At first, it seems like a pretty tough problem to solve since you want as much friction as possible to push things with, yet you want much less friction for turning.

While the problem seems tough at first, the solution is fairly simple. A little static moment analysis of your drive system should make things very clear.

There are three basic methods of keeping a lot of friction for pushing a having much less for turning. Two methods I like (we use them every year on our robot), and the other method I don’t care for. Here they are:

  1. Use two drive wheels in the front or rear and use caster wheels on the other end. This is very simple, but I don’t really like this method since not all of your robot weight is on driven wheels which will ultimately mean you will have less pushing force than you could potentially have.

  2. For the parts of your drive train that contact the floor, design them such that they have much more friction in one direction than they have in the perpendicular direction. That is, make your wheels / treads grip in one direction and slip in the perpendicular direction. There are two good examples of this: snow skis and ice skates. Both slide very easily back-and-forth, but are very sharp and grip well side-to-side. You can design your wheels / treads this same way - we’ve been doing is for years with great results.

Another method to achieve this (but a lot more complicated) is to use the holonomic wheels that a lot of teams have been using. This way you have low turning friction like casters yet all of your wheels are driven.

  1. The last method is to do some static moment analysis on your drive train. In short, you will find that if your wheel base is longer than your track width, you won’t turn very well (if you will turn at all). You should always have a shorter wheel base than your track width - especially if you don’t do #2 above.

Our team always does a combination of #2 and #3 above. We’ve always had great drive trains that have great traction and are easy to steer and control.

Two things to remember:

a) If you do a static moment analysis of a 4-wheel drive tank steer system, you’ll see that there are two ratios affecting how easily your robot turns. They are: 1. the ratio of the track width to wheel base (higher ratio => easier turning); and 2. the ratio of the longitudinal friction coefficient to lateral friction coefficient (higher ratio => easier turning).

b) Some turning friction is GOOD!!! If you eliminate all of your turning friction, the drive train becomes very difficult to control. This is because when you let go of the stick(s), the robot continues to turn until friction stops it. For the robot that is easiest to control, you ideally want the robot to stop turning as soon as you release the sticks. Therefore, you should never have a goal to eliminate all of the friction - the goal should be a good balance between keeping the motor current low and being able to stop the robot from turning once the sticks are released.

I hope this helps,

Chris

This is very interesting and helpful, but I’m a little confused as to what you called “track width.” What exactly did you mean by that?

Chris wrote a very good description of the problem but there is at least one more option and that is to have the drive wheels turn in the direction you want to go. We have used this on two robots and call it “crab steering”. This year, all four wheels were driven and all four turned. The effect on robot movement is to make it look like a crab running down the beach to avoid waves. The robot moves in any direction without having to point in that direction. There are down sides to this approach though…Software must get feedback for position info, turning driven wheels is more mechanically complex and driver practice is essential to know the reaction of the steering during competition. That last one is the most important because there no longer is a “front” side although the robot is still capable of backing up or moving forward in any direction. Think of it as the queen on a chess board, any direction same power and same speed in all directions, extreme manueverability.

*Originally posted by Jeff Waegelin *
**This is very interesting and helpful, but I’m a little confused as to what you called “track width.” What exactly did you mean by that? **

Sorry. I’ve been working in the auto industry so long that I think “track width” and “wheel base” are part of everyone’s household vocabulary.

Here are the simple definitions (assuming your robot has 4 wheels in the corners of an imaginary rectangle):

Track Width: Track width is the perpendicular distance between a line connecting the left side wheels and a line connecting the right side wheels (i.e., the left-to-right distance of your wheel footprint). Put another way: Coat your wheels with paint and drive it in a straight line. Your robot will make two paint lines on the ground (one from the left side wheels, and one from the right side wheels). The distance between the two lines is the track width.

Wheel Base: Wheel base is the perpendicular distance between a line connecting the front wheel contact points and a line connecting the rear wheel contact points. (i.e., the front-to-back distance of your wheel footprint).

I hope that helps.

-Chris

Thanks everyone for your very helpful ideas.

We’ll be trying to implement some of them this week.

For those on or near the west coast, hope to see you next week (June 29th) at the Southern California Summer Classic event at Cal. State University Northridge.