www.robowars.org

RoboWars Australia Forum Index -> Technical Chat

Drive train calculations.
Goto page Previous  1, 2

Post new topic   Reply to topic
  Author    Thread
Daniel
Experienced Roboteer


Joined: 30 Aug 2005
Posts: 2729
Location: Gold Coast


 Reply with quote  

quote:
Originally posted by DumHed:
Assume a square bot with 4 wheels, one at each corner, and only the rear wheels driven.
If it has enough grip and power on takeoff, it will do a "wheelie" due to geometric weight transfer - or in other words, the lever action between the axle line and the plane of thrust (being the floor).
Having the front wheels driven doesn't change the weight transfer - they'll still be in the air if the rears have enough grip, and all weight will be transferred to the rear.
This obviously means the rears get more traction, and can put more power to the ground, but only under the right conditions.


Then you will have to calculate the down force on your rear driven wheels at each point of the acceleration so you can calculate the amount of maximum power able to be transfered. You arn't making things more difficult or point out mistakes. Your just making it sound more complicated then it really is.


But as we continue to point out you only get skidding at the change in acceleration. Technically know as Jerk (no stupid coments please Aaron).
Once the car starts to take off and the weight has moved the acceleration is constant and therefore doesn't effect the initial skid. It is only the instentanous point when power is applied that we are concerned with. Will the instentanous torque cause the wheels to slip or not. What happens after the car or robot starts to move is not a concern. Wether the front end begins to lift or not has no effect of this gear ratio equation.

In fact I have only even seen one 4 wheeled robot lift its front wheels during a change in acceleration.

Post Thu Feb 09, 2006 7:04 pm 
 View user's profile Send private message Send e-mail
prong
Experienced Roboteer


Joined: 19 Jun 2004
Posts: 839


 Reply with quote  

I thought that the weight transfer is instant?

so even before anyhting has moved the weight transfer is there

Post Thu Feb 09, 2006 7:14 pm 
 View user's profile Send private message Send e-mail
Knightrous
Site Admin


Joined: 15 Jun 2004
Posts: 8511
Location: NSW


 Reply with quote  


quote:
no stupid coments please Aaron


You've already made them for me...
_________________
https://www.halfdonethings.com/

Post Thu Feb 09, 2006 7:19 pm 
 View user's profile Send private message
DumHed
Experienced Roboteer


Joined: 29 Jun 2004
Posts: 1219
Location: Sydney


 Reply with quote  

Brett: yes I know serious drag cars have a lot of work in the suspension, and that going really fast in a straight line isn't simple or easy.
What I'm saying though is that the suspenion is more about geometry than springs.
The weight doesn't get transferred because the car pitches back on its soft suspension, it's transferred because the suspension is designed to load up the rear tyres as much as possible to make maximum use of the available weight to improve traction.

The springs and shocks are there to dampen the movements and prevent oscillation (wheel hop / axle tramp, etc) occurring when the limits of grip are close to being exceeded.

Daniel: I'm not saying that anything you've said is wrong, but I think that a bunch of calculations with some guesstimated values and a lot of "ideal" circumstances are not going to be that relevant to real world robotic combat.

There are plenty of reasons why your available grip could suddenly be halved, doubled, or you could have an imbalance left to right.
On a solid chassis with no suspension and four wheels, even a very slight (1mm or less) variation in floor height (or chassis alignment) can have a big impact on the amount of weight on each wheel.
Putting a sheel of cardboard under one wheel will put more weight on that wheel and the one in the opposite corner, and reduce the weight on the other two wheels by the same amount.
On the race car when we're measuring corner weights we have to have it level to within the thickness of a couple of sheets of paper to have it spot on - and that's a full size car with pneumatic tyres and suspension!

This is the reason I went with the pseudo three wheel setup in stealth. It keeps the driving wheels on the ground at all times.

You will get skidding (wheelspin) under change of acceleration *or* change of grip level.
If you're accelerating at x ms/s you're exerting a constant tractive force on the floor. If the grip level drops to a point where that force can no longer be applied, you have slip - and that tends to worsen as the wheel speed increases due to reduced motor load.
That grip level can be changed greatly by weight shift, uneven terrain, or the weight of another robot.
_________________
The Engine Whisperer - fixer of things

Post Thu Feb 09, 2006 10:28 pm 
 View user's profile Send private message Visit poster's website MSN Messenger ICQ Number
Daniel
Experienced Roboteer


Joined: 30 Aug 2005
Posts: 2729
Location: Gold Coast


 Reply with quote  

no calculation anywhere in the world is perfect. Not even the ones used by car manufactures.
Any change in grip due to weight distrabution, spots on the floor, tyre wear, ect... will only change the gear ratio I have calculated by a very small amount , 0.1:1 style of thing, and since no would ever be even able to make the correct gear ratio it does not matter about these small features. These sorts of calculations are a guide line only so builders will have a rough idea of where they need to start when designing their own drive train. Taking into account every little factor you can think off and those you can't is a waste of time when the results will still be rounded of to the same gear ratio as the origonal, simpler calculation.
I would be surprised if Brett and crew didn't have to do some rounding and beleive in ideal situations when designing the IBC. Resistors and capacitors only come is certain sizes, just like gears and sprockets. If engineers took things as literally as you do then we wouldn't design anything and you would have to walk everywhere because know one would be able to design your car with every little factor accounted for.

I'm surprised you havn't mentioned floor angle, thermal expansion, cross winds or that damn butterfly that causes tornados.

Post Fri Feb 10, 2006 8:09 am 
 View user's profile Send private message Send e-mail
Spockie-Tech
Site Admin


Joined: 31 May 2004
Posts: 3160
Location: Melbourne, Australia


 Reply with quote  

Ah, the traditional clash between Engineer and Techie.. Wink

Back when I was a lowly Techie, We used to joke that fresh-out-of-school Electronic Engineers would turn a resistor into a 4 page quadratic equation withour raising a sweat, but would then burn their fingers trying to solder it onto the board. Laughing

Of course some theory to back things up is good, so you understand *why* it works the way you *know* it does.

Otherwise you're just doing a "monkey-see, monkey-do" type of operation. Theory on its own results in you knowing why it *should* work, even although the Techie knows (From Experience) that it *wont* (even if not why).

IMO, The best engineers start out as tech's and pickup the theory to fill in the gaps and reasons. Soo. Back to the point.

Daniels attempt to model the traction/power of a robot, while always going to be less than "reality" (only reality is a perfect simulation of the real world) can nonetheless still be developed to be useful.

There's no point simulating a hypothethetical concrete plant next door blowing dust all over your freshly VHT'd track, nor what the tornado-causing butterfly is up to.. (he'd get caught in a car radiator in a *real* simulation anyway.. Wink )

Developing simulator models that leave out the "unimportant" factors to reduce the problem to a more dealable with size is the *art* of simulation. The trick lies ies in deciding what factors can be ignored, or "assumed to be X" without invalidating the usefulness of the model.

The best way to test a sim model is to have it make some predicitions, change some parameters, run the numbers again. Then head for reality and then see if the sim results and the real life results reflect approximately the same change in resulst for the same change in parameters.

Developing a model *too far* without this step is quite likely to end up leaving you in a reality dimension where your spinning bar has 4.83e-10 Kilojoules of Energy, but you suffocate and die while it spins up because you forgot to include the air (and consequent resistance).

So while his initial equations might have a few holes in them, I see nothing wrong with using them as a starting point. Just structure them into some easily used programming code that can have more factors included, and gradually your CAD model will turn from a square-box with 4-wheels on the corners into something useable as you tweak it to suit your reality-results.
_________________
Great minds discuss ideas. Average minds discuss events. Small minds discuss people

Post Fri Feb 10, 2006 8:39 am 
 View user's profile Send private message Send e-mail Visit poster's website MSN Messenger
DumHed
Experienced Roboteer


Joined: 29 Jun 2004
Posts: 1219
Location: Sydney


 Reply with quote  

I agree that the basic premise of modelling the traction / power of a robot could be handy, but there are so many variables that *do* become signifigant that in my opinion you're better off spending your time building something rather than simulating something.

My point about weight shift is that depending on the robot's design, any traction calculations could very easily be out by a huge margin.
Well positioned drive wheels could increase their traction 1.5 times under acceleration, but depending on the robot's design, there could be other trade offs (front wedge lifting, front wheels losing drive / steering, etc).
Have a look at SpeedBump. Due to slightly uneven motor / wheel heights, and a lot of weight shift onto the edges of the body, it often has a wheel or two spinning in the air, and doesn't get much pushing power. As soon as SpeedBump is wedged under another bot though, it has huge amount if grip and power, and the wheels that were getting poor traction are suddenly very useful.

In the end, as you say, you can't even get the exact gear ratio needed, so you have to compromise.
The average first builder is not going to be constructing their own gearbox or other reduction drive to a calculated ratio when there are plenty of pre made things that work very well (drills, scooter drives, etc).

Anyway, I'm all for people simulating things and calculating how they should work, but in the end there's only one way to find out Smile
_________________
The Engine Whisperer - fixer of things

Post Fri Feb 10, 2006 9:02 am 
 View user's profile Send private message Visit poster's website MSN Messenger ICQ Number
Nexus
Experienced Roboteer


Joined: 17 Jun 2004
Posts: 903


 Reply with quote  

Might as well have a rant.
Interesting topic, Can see many similar issues with working with tracks with suspension.
The final challenge I have to work out with Pod is getting the suspension setup to basically handle the forces that get applied to the rear of the body on acceleration and through most of the drive.
It would have been easier to not have suspenion but it’s a fun challenge to get this thing working properly and will be very satisfying when it does.

On acceleration a lot of weight gets shifted to the rear and puts a lot of load on the back.
When tanks have their drive sprockets on the front it tends to lift up the front of the body, when the drive sprocket is on the rear it tends to pull the back down. Either way the weight gets put backwards and tanks generally have their engine at the rear regardless of which sprocket is doing the drive. This is more evident as you increase power.
In a lot of ways it seems similar (but probably isnt) to speed boats in that the front raises and you are moving forward with just a small amount of contact with the ground with increased pressure and that’s always at the rear.

Weight distribution becomes a big factor as well, in tanks for example when the turret is removed and a cabin is installed for training purposes a counterweight is installed on the front to compensate.

Theory is useful if applied, if it is not applied correctly then its just a book placed next to your robot as a drink coaster. Everyone am sure has tried something that has failed and then thought about it and realised it was always going to fail but they couldn’t see the obvious because they where motivated strickly by the theory of one component. Theory can fail sometimes because it doesn’t consider all other variables.

A common saying is (The whole is greater than the sum of the parts). Focusing on one component can be a failing as you are ultimately building a system with relationships between the components, without considering all components and their effects on each other then all you will end up doing is learning from your mistakes, which is actually a good practical way to learn things because you will remember the lessons and apply them next time, hopefully anyway.

Being able to visualise things can also be very helpful, as a funny haired old dude once said, “Imagination is more important than Knowledge”. Think about what that means.
_________________
Bots that do not destroy you, only make you stronger.

Post Fri Feb 10, 2006 12:48 pm 
 View user's profile Send private message
DumHed
Experienced Roboteer


Joined: 29 Jun 2004
Posts: 1219
Location: Sydney


 Reply with quote  

exactly Smile

You can often come up with a "mental simulation" taking in a lot more variables than you can in a mathematical simulation.
Such is the beauty of neural net processing Smile
_________________
The Engine Whisperer - fixer of things

Post Fri Feb 10, 2006 12:54 pm 
 View user's profile Send private message Visit poster's website MSN Messenger ICQ Number
Philip
Experienced Roboteer


Joined: 18 Jun 2004
Posts: 3842
Location: Queensland near Brisbane


 Reply with quote  

Daniel's formula seems good to me if you want to calculate maximum acceleration for a given surface. The coefficient of friction might be hard to predict.

I have also read that some builders will always gear their bots to have enough torque to always be able to break traction and skid even when they have the weight of another bot on top of their own. The rationale is they will never stall the motors and cook their electrics. They limit current by gearing.
_________________
So even the rain that falls isn't actually going to fill our dams and our river systems

Post Sat Feb 11, 2006 8:45 am 
 View user's profile Send private message
DumHed
Experienced Roboteer


Joined: 29 Jun 2004
Posts: 1219
Location: Sydney


 Reply with quote  

yeah that's another point I was going to make.

As long as you still have enough top speed, it's worth having the power to spin the wheels so you don't stall the motors.
_________________
The Engine Whisperer - fixer of things

Post Sat Feb 11, 2006 10:20 am 
 View user's profile Send private message Visit poster's website MSN Messenger ICQ Number
  Display posts from previous:      

Forum Jump:
Jump to:  

Post new topic   Reply to topic
Page 2 of 2

Goto page Previous  1, 2

Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

 

Last Thread | Next Thread  >
Powered by phpBB: © 2001 phpBB Group
millenniumFalcon Template By Vereor.