Coefficient of Friction Testing [Archive]

JVN

11-08-2012, 22:47

I just published a post (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html) which highlights my method for calculating the coefficient of friction for robotic drivetrain systems.

In my opinion... it is not sufficient to test the coefficient of friction of a single piece of tread, or a single wheel on carpet -- we need to test whole systems, and determine THEIR coefficient of friction. Only by testing the actual configuration of the robot you're going to use (i.e. the correct number of wheels, the correct tread contact size, the correct weight etc) will you get the correct coefficient of friction.

See full post here (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html).

So this brings up my next question...
Who actually bothers to do this?

Does anyone? Is it just me?
Does your team have their own method?
Does your team just trust the manufacturer's specs?
Do you even care, or do you just say "this is grippy as heck" and not bother?

Please share.

Originally posted here (http://blog.iamjvn.com/2012/08/do-you-test.html).

davidthefat

11-08-2012, 22:57

Our team did calculate the COF on the wheels as I was having an argument with one of the mentors regarding the number of motors to use. Personally, I just used one wheel for the test.

JVN

11-08-2012, 23:00

Our team did calculate the COF on the wheels as I was having an argument with one of the mentors regarding the number of motors to use. Personally, I just used one wheel for the test.

Which wheels? What was your result?

davidthefat

11-08-2012, 23:04

Which wheels? What was your result?

These wheels were 29635T31 from McMasterCarr. The argument was whether or not these would slip; if they did slip, a second motor in the gearbox would be useless. I do not have the results as they were written in someone's journal; I do not believe he/she still has it. Sorry for the relatively pointless post.

sanddrag

12-08-2012, 01:35

Years ago we did some friction testing on 968, but it was far from scientific. Our primary goal was to determine how much of a role wheel width played on friction with regards to maximum pushing force. In the past, 968 had run rather skinny wheels (as skinny as 5/8" tread width). We made up two sets of wheels with tread (and I don't remember if it was wedgetop or roughtop). One set was 1 inch wide, and the other set was 1.5 inches wide. We installed the narrower wheels and filled a plastic container with batteries so it was quite heavy and experimented with having the robot push this container on carpet. We then added more weight until the robot could no longer move it. We then swapped out the wheels for 1.5" wide ones and gave it another go. The robot pushed it with no problem, and it took adding something like another 40lbs to the container to make it un-moveable again.

While not very scientific, we did determine that wheel width can play a substantial role in friction.

Beyond that, the only thing I really bother to do each year is press some wheels against the ground while simulating turning them and think to myself "yeah, this grips well." I know, not very 'engineer' of me. If the need arises in the coming season, perhaps I'll have some students do the test you describe.

ttldomination

12-08-2012, 06:37

Wow, never thought I'd see this brought up.

Some years ago, we actually had to find the COF between a rough-ish sponge and a piece of wooden board (relatively smooth), for a physics lab.

My group promptly used the method you've detailed, and why not, it's so dang easy. However, our results were not what we expected. We saw huge variances in the "sliding" angles. We're talking +/- 20 degrees here.

We never did figure out what were causing the variances. The teacher provided a whole host of possible reasons, none of which really did the problem any justice.

The most common/successful method was probably to weigh down the rough surface and then use a force gauge that provided a force vs. time reading. Then, from rest, slowly pull. At a certain point, the rough surface+weights will start moving. At that same point, your force will peak, and that becomes the maximum static friction (Fs). From there, it's a simple Fs = Fn * mu calculation. Granted this did require more technical equipment, it also seemed to provide much more consistent results.

- Sunny G.

JamesTerm

12-08-2012, 08:59

Billfred

12-08-2012, 09:58

Does anyone? Is it just me?
Does your team have their own method?
Does your team just trust the manufacturer's specs?
Do you even care, or do you just say "this is grippy as heck" and not bother?

None of the teams I've been on have ever tested it themselves, trusting the manufacturer (or CD, in the case of roughtop/wedgetop) instead.

Nothing against those that do, we just inevitably run into bigger problems that eat up our time during the time it's a relevant question. (After all, the system is different next year.)

DonRotolo

12-08-2012, 11:05

John,

We don't calculate the CoF. If we need to know, we test relative to some other standard (i.e., another robot) by pushing it along carpet and measuring force - sometimes with 'calibrated legs' only.

Your method is valid for static friction, but dynamic friction is also important IMHO, since once you start moving the value is usually quite different from static friction (= Sticktion). To measure that in your test setup, you lift to just below your normal angle, set the robot sliding, then reduce the angle to where it stops. This was very important to us in Lunacy, for example, where the wheels were generally not at rest relative to the surface.
We never did figure out what were causing the variances.Probably the 'interlocking' of the soft, rough sponge with the smooth-but-not-as-smooth-as-you-think wood. Get a particularly strong splinter of wood, and it grips more than if you only get weak (or short) ones.

Ian Curtis

12-08-2012, 12:05

John,

We don't calculate the CoF. If we need to know, we test relative to some other standard (i.e., another robot) by pushing it along carpet and measuring force - sometimes with 'calibrated legs' only.

Your method is valid for static friction, but dynamic friction is also important IMHO, since once you start moving the value is usually quite different from static friction (= Sticktion). To measure that in your test setup, you lift to just below your normal angle, set the robot sliding, then reduce the angle to where it stops. This was very important to us in Lunacy, for example, where the wheels were generally not at rest relative to the surface.
Probably the 'interlocking' of the soft, rough sponge with the smooth-but-not-as-smooth-as-you-think wood. Get a particularly strong splinter of wood, and it grips more than if you only get weak (or short) ones.

Fun question: Is dynamic friction a constant with respect to velocity? I know a Physics 101 textbook will tell you it is, but I wonder if you ask a tribology (http://en.wikipedia.org/wiki/Tribology) (isn't that a fun word) expert about the dynamics of tread on carpet what the answer is. Or if anyone has ever tested it?

That is, do you have the same coefficient of friction with a tire sliding at 1 ft/s as you do with a tire sliding at 10 ft/s.

JVN

12-08-2012, 12:40

The most common/successful method was probably to weigh down the rough surface and then use a force gauge that provided a force vs. time reading. Then, from rest, slowly pull. At a certain point, the rough surface+weights will start moving. At that same point, your force will peak, and that becomes the maximum static friction (Fs). From there, it's a simple Fs = Fn * mu calculation. Granted this did require more technical equipment, it also seemed to provide much more consistent results.

- Sunny G.

This test has flaws also; the "puller" is usually not able to provide a pull of constant velocity. If there is any acceleration or change in acceleration, the waters get muddy quickly, and the data can be flawed.

We found that the tilt test is much more consistent for FRC applications as described.

-John

Siri

12-08-2012, 14:32

We don't do this routinely (being both somewhat limited and somewhat naive in wheel choices), but with the bridge this year we did several true experiments, including using the full robot and bridge with various wheels and other surface texture changes. We used both tipping and dragging (as our CG meant tipping sometimes preceded slipping). With all our off-season drivetrains, we're setting up weighted testbed chassis and switching in wheels/treads on carpet, polcarbonate and polypropylene (approximation for polyethylene).

Some years ago, we actually had to find the COF between a rough-ish sponge and a piece of wooden board (relatively smooth), for a physics lab.The situation you describe is not strictly Coulombian, because of inconsistency both in approximating contact area as proportional to normal force and in the angle(s) of the contact planes(s) in general. (Leaving the sponge's contact with the wood and the wood's own imperfections to prevent constant behavior in both time and space).

Weighing down the surface (higher but constant normal force) gets you past some of that initial "noise". However, beware both the "pull slowly" difficultly JVN describes and the failure of Columbian models when contact area isn't proportional to normal force (below saturation) and/or frictional force isn't proportional to normal force (independent of contact area).

JVN

12-08-2012, 14:39

JamesTerm

12-08-2012, 15:48

full post here (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html).

Thanks for sharing this link... are you trying to solve for static friction only? What about kinetic friction?

JVN

12-08-2012, 17:39

DonRotolo

12-08-2012, 18:39

Data? No. That's lost to me.

The first question we wanted to answer was if 12 wheels were any better than 6. One test was to have a 6-wheel robot and a 12-wheel robot (both same wheelbase and weight) push against each other, to see who would win. At first, the loser was universally the one which spun its wheels first. With extremely careful and skilled control, the 12 could win but the 6 could not. Repeatable with difficulty, but repeated enough times that we went with 12.

The second question was related to whether traction control - to prevent wheel spin - would provide an advantage.

We first needed to find the amount of wheel slippage that would be acceptable, and for Lunacy, IIRC we found that grip vs wheel velocity relative to the surface was somewhat linear - higher speeds giving less grip. We did this by powering up the wheels with a PID loop to a certain speed, and measuring the resulting pull with a scale (think fish scale, but of a higher quality, up to about 40 Lbs).

We then needed to measure the effect of a slip limiter. Using a 5th wheel for odometry, we used a PID loop to limit PWM input to Jaguars such that the slip was some percentage of the robot speed. (I think we settled on 12%, but I can't say why). We then went drag racing, timing the robot over a fixed distance, the driver simply giving full throttle and letting the system throttle it back.

Our final result was that a simple rate-of-change limiter in software was nearly as effective, avoiding a lot of hardware, and that's what we competed with.

Chris Fultz

12-08-2012, 21:36

we did a test with a kit robot and various wheel widths and diameters.
when we got done, we could not make sense of the data.

we made a test robot and used a pull scale. The robot was set in a frame, with carpet on the floor. We set the control system to ramp up the motor power at a predetermined rate. we ramped up until the wheels broke free and recorded the pulling force at that break point.

I think this test could easily be repeated with various types of wheel treads, using the bridge material instead of carpet.

R.C.

12-08-2012, 21:39

Siri

13-08-2012, 07:39

Do you have any results you can share?

-JohnGenerally the published results were fairly reflective of our results, but we ended up with more noise than I was expecting, for which we couldn't find any real trend or explanation. We also found that the blue nitrile wore better and handled dust, etc better than the roughtop. (No numbers--a reflection that's reminded me to belatedly add wear testing to our summer procedure.) Colsons are somewhat better on the bridge, but by the time we did the switch we were using our real robot (CG around ~15") and we always tipped before falling* -- at well over the bridge's maximum angle. We haven't done much wedgetop testing yet--on the summer list.

*EDIT: Do'h. Absolutely need to point out that this test was conducted with our pivots at 45deg angles. Wow. My bad.
/...\
\ ../

IKE

13-08-2012, 08:21

Why do you want to know the COF of your robot?*

*Are you sure you are asking the right question?

Very cool post. I have used a similar method for estimating COF of material on material.
The tilt test is also a pretty good way of testing CG location, but it can get tricky.

AdamHeard

13-08-2012, 08:25

This is something 973 was planning on testing internally during the fall on a couple of our projects. We're very experienced with roughtop/wedgetop style tread and like their performance, but are interested in how colsons and the popular variants of pneumatic wheels compare.

We plan to do the incline plane test with carpet and polycarb sometime in November. Ideally we'll publish results.

Lil' Lavery

13-08-2012, 10:04

Why do you want to know the COF of your robot?*

*Are you sure you are asking the right question?

I'm going to echo this question. Just for further iterative designs to objectively improve your CoF? Otherwise I'm struggling to see why having CoF data that late in the design process (ie, when you already have your drive base built) would be useful.

JVN

13-08-2012, 10:23

I'm going to echo this question. Just for further iterative designs to objectively improve your CoF? Otherwise I'm struggling to see why having CoF data that late in the design process (ie, when you already have your drive base built) would be useful.

One would test drivetrain configurations (size wheels, location of wheels, qty of wheels, robot weight, CoG location) so one has better understanding of those particular configurations, and their CoF, to better understand their potential pushing force.

You don't need to complete your robot to get this data. You can get it early in the design process... or even in pre-season.

-John

JVN

13-08-2012, 10:33

*Are you sure you are asking the right question?

I was pretty sure... until your post. Am I missing something?

-John

kramarczyk

13-08-2012, 12:58

The last time I did qualitative traction testing was in 2009. See this post. (http://www.chiefdelphi.com/forums/showpost.php?p=855679&postcount=21)

Since then I have not needed to build a traction limited drivetrain, so the actual number was not essential. We did some qualitative testing for the polycarbonate on the bridge this year. We were looking for the best material that we had access to, so A>B testing was adquate.

Admittidly, I would like to have a library of exactly the information the OP is asking about, but it is less important than foundation work that needs to be done with/for the team.

JamesCH95

13-08-2012, 15:00

The more I read up on various stories the more I believe that most FRC wheel act like car tires. It matters what temperature they're at and what contact pressure they see. Therefore it is important to test the drivetrain as a whole unit, not just individual wheels. I

f some wheels in the setup are more heavily weighted than others they will have a lower CoF than identical wheels loaded less, i.e. a drop-center 6wd, where the center two wheels generally bear 90%+ of the robot's weight. In a configuration like this I believe that it would be most efficient to use wide wheels in the middle and narrower wheels on the outside corners.

Additionally, if one were doing testing, it would be interesting to heat up the wheels by spinning them for a few moments, then re-trying the static friction test.

@DonRotolo- your slippage value of 12% is very similar to high performance traction control systems in race cars; IIRC typically 8-10% slippage resulted in maximum straight-line acceleration. This may be because the surface of the wheel/tire is getting 'ripped up' a little bit, causing it to engage the driving surface positively, but I am no tribologist.

DonRotolo

13-08-2012, 20:46

@DonRotolo- your slippage value of 12% is very similar to high performance traction control systems in race cars;Despite finding our values empirically, they correlate well with the automotive values found in the Bosch Handbook. (I work for a car company)
but I am no tribologist.Yeah, I don't know much about tribbles (http://en.wikipedia.org/wiki/Tribble) either.

IKE

14-08-2012, 08:06

I was pretty sure... until your post. Am I missing something?

-John
Let me prefice by saying this is a really good test, and a pretty accurrate method, but as this thread is starting to expose, like any acquired taste, there are various levels of CoF snobbery.

************************************************** ************
I am only slightly messing with you. As Don and James have described, there really isn't a CoF number, but more of a CoF curve that depends on a lot of parameters. As having the most accurrate CoF is generally not an award at a Robotics competitoin (though it would make a great science fair project), one must assume that you are trying to get a CoF to do a calculation. If you are trying to do a calculation, the inherent question you need to ask yourself is how accurrate do I need to be? 1%, 2%, 5%, 10%? As the CoFs I have seen for non-lunacy wheels tend to range from 0.8 to 1.3 with most being in the 0.9 to 1.1 range, I would assume that a test with +/-10% accuracy is probably not sufficient.

Back to how accurrate does it need to be, within the measuring technique you showcase, If your angle accuracy is accurate to within +/-1 degree, at 45 degrees, you will come up with a CoF of 0.966 to 1.036. Or a total error band of about 7%. +/-2 degrees will get you 0.93 to 1.07. In terms of FRC wheels, that would be pneumatics fully inflated (according to AM wbsite) to fresh wedgetop. My phone has an inclinometer app on it that measures within 0.1 degrees, but on a flat surface, with me holding it steady, it varies +/-0.7 degrees.

As you said in your post, you don't believe a patch, or a single wheel will give you represenative data. This tells me that you believe that the weight and weight distribution must have an effect. If this is true, then a tilt table will shift the normal force distribution on those wheels and thus the tilt table will give you a different result than flat ground pushing. this is especially true when the board reaches high angles which just so happen to correspond to high tractions. Also, small flex in the board can throw off the angles as you will get a different contact patch than you were expecting.

Couple this information with the Slip % variation Don and James are talking with (on street tires it is often around 0.95 peak at 8-12% slip, and then dropping down to 0.8-ish past 20% slip), and you will find a lot of neat variation that adds up.

We used a similar method on the polycarbonate seeing which wheels slid first. We didn't need to know the exact number, just which had more traction.

****************************************

I may be a bit sensitive on this subject as I am coming off of a 24 hour race with a bunch of CoF super snobs. At around the 12 hour mark, there was a 1% difference in lap count between 5th and 14th place. There was only about 3% difference between 1st and 20th.

kramarczyk

14-08-2012, 13:31

Back to how accurrate does it need to be, within the measuring technique you showcase, If your angle accuracy is accurate to within +/-1 degree, at 45 degrees, you will come up with a CoF of 0.966 to 1.036. Or a total error band of about 7%. +/-2 degrees will get you 0.93 to 1.07. In terms of FRC wheels, that would be pneumatics fully inflated (according to AM wbsite) to fresh wedgetop. My phone has an inclinometer app on it that measures within 0.1 degrees, but on a flat surface, with me holding it steady, it varies +/-0.7 degrees.

Do you guys actually measure the angle and then do the trig? Wouldn't measuring the rise over run with a carpenters square provide better resolution. Rise / run = tan theta = CoF

An 1/8" difference in rise is pretty easy to measure and works out to ~ 1% change in CoF.
12/18 = 0.66667, 12.125/18 = 0.67361, 1.03% difference
17/12 = 1.41667, 17.125/12 = 1.42708, 0.73% difference

JamesCH95

14-08-2012, 13:36

Do you guys actually measure the angle and then do the trig? Wouldn't measuring the rise over run with a carpenters square provide better resolution. Rise / run = tan theta = CoF

An 1/8" difference in rise is pretty easy to measure and works out to ~ 1% change in CoF.
12/18 = 0.66667, 12.125/18 = 0.67361, 1.03% difference
17/12 = 1.41667, 17.125/12 = 1.42708, 0.73% difference

How accurately can you measure the rise and run with respect to gravity? Using a carpenter's square is good, but you have to ensure that it is square with gravity also.

Ether

14-08-2012, 13:58

Rise/Run requires 2 measurements.

Do it with just one: measure the height "h" of the ramp.

Then CoF=h/sqrt(L2-h2) ... where "L" is the (fixed) length of the ramp.

Ether

14-08-2012, 13:59

How accurately can you measure the rise and run with respect to gravity? Using a carpenter's square is good, but you have to ensure that it is square with gravity also.

Just do it on a level surface.

JamesCH95

14-08-2012, 14:19

Just do it on a level surface.

One still needs to measure the level surface to see that it is, in fact, level.

Ether

14-08-2012, 14:38

One still needs to measure the level surface to see that it is, in fact, level.

I don't see the point you are trying to make, in the context in which you originally posted.

You'd have to do that anyway, if you were measuring the angle between the ramp and the floor.

JamesTerm

14-08-2012, 15:00

Fun question: Is dynamic friction a constant with respect to velocity? I know a Physics 101 textbook will tell you it is, but I wonder if you ask a tribology (http://en.wikipedia.org/wiki/Tribology) (isn't that a fun word) expert about the dynamics of tread on carpet Iwhat the answer is. Or if anyone has ever tested it?

That is, do you have the same coefficient of friction with a tire sliding at 1 ft/s as you do with a tire sliding at 10 ft/s.

I find this interesting... I've never heard it called dynamic friction (just static or kinetic)... so I googled around... it appears kinetic and dynamic are synonymous, but now I wonder if dynamic means in terms of velocity. Kinetic friction is very clear... its either static or kinetic values depending on if object is moving, but is there some cutoff point where the value changes from one state to another or is it a blend. This is one reason why I find the test methods interesting. In code I measure the value 0 velocity against a theshold of around 1E-05.

Ether

14-08-2012, 15:42

is there some cutoff point where the value changes from one state to another or is it a blend

search for "Stribeck effect" or "Stribeck friction"

JamesCH95

14-08-2012, 15:43

I don't see the point you are trying to make, in the context in which you originally posted.

You'd have to do that anyway, if you were measuring the angle between the ramp and the floor.

In my original post my point is that unless you measure actual rise (movement parallel to the force of gravity) and actual run (movement perpendicular to the force of gravity) the angle you calculate will be inaccurate by however off the reference surface is.

Furthermore, one has to measure either the angle of the reference surface (the surface that the carpenter's square is placed on) and then measure rise/run, or just measure the slope directly. The former contains three sources of error, the latter has only one source of error. I would prefer to measure just one thing, the angle of inclination, instead of the angle of a reference surface and two distances.

Determining if these errors are significant is the responsibility of the experimenter.

Ether

14-08-2012, 16:13

just measure the slope directly [with] only one source of error.

How do you propose to "just measure the slope directly" with only one source of error?

EricVanWyk

14-08-2012, 16:22

How do you propose to "just measure the slope directly" with only one source of error?

A plumb-bob and a protractor should be a good start.

JamesCH95

14-08-2012, 16:24

How do you propose to "just measure the slope directly" with only one source of error?

I would use a digital angle finder (http://www.amazon.com/Wixey-WR300-Digital-Angle-Gauge/dp/B001PTGBRQ/ref=sr_1_2?ie=UTF8&qid=1344975811&sr=8-2&keywords=digital+angle+finder).

JVN

14-08-2012, 16:25

A plumb-bob and a protractor should be a good start.

We have a device called an inclinometer which is these two things in one package.

http://www.priceit.in/wp-content/uploads/2011/01/Inclinometer-2.jpg

kramarczyk

14-08-2012, 17:04

John,

Just for reference, what is the resolution on that inclinometer?

JVN

14-08-2012, 17:05

John,

Just for reference, what is the resolution on that inclinometer?

Not very good. I think ours is +/- 1-degree.

-John

Tom Line

14-08-2012, 17:05

I just published a post (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html) which highlights my method for calculating the coefficient of friction for robotic drivetrain systems.

See full post here (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html).

So this brings up my next question...
Who actually bothers to do this?

Does anyone? Is it just me?
Does your team have their own method?
Does your team just trust the manufacturer's specs?
Do you even care, or do you just say "this is grippy as heck" and not bother?

Please share.

Originally posted here (http://blog.iamjvn.com/2012/08/do-you-test.html).

John, I believe you MUST check as a full robot, and with varying weights on the wheels. The tread we use does not act as an ideal surface - it has sub-surface interaction and interlocks with the carpet pile. In addition, on a six wheel drive the center of gravity is rarely in the center. So how, exactly, are you riding on your wheels?

A number of people have created spreadsheets detailing 'can my robot turn'. I don't believe those tell you what you actually want to know. You've seen robots on the field that can turn. Then you've seen robots like 254 and 1114 that can FLY. I think everyone knows the difference I'm talking about.

Something that I want to do later this year (during Beta Test) is to get our precision scales from the academy's physics class. Attach carpet to them, and put them on one of our marble measuring tables, then set the robot on them. I'd like to get the normal force each wheel exerts.

Between 1718's 3 nearly fully assembled robots, it should be possible to create a 'model' of drivetrain geometry and weight distribution that produces a robot that turns well rather than one that 'just turns'.

I've got 'turning' stuck in my head because we've fought with it on a couple robots and had to go back and change wheel types and treads to get acceptable performance. Frankly, I have never ONE seen one of our 4 wheel, 6 wheel, or 8 wheel drive robots lose traction, except when getting pushed completed sideways by a robot with treads and a low gear. I think how the friction affects robot turning is the more important application of the data you folks are talkign about. (Completely apart from the fact that I believe if you're in a pushing matches often, you're likely playing the game wrong).

Ether

14-08-2012, 17:19

I would use a digital angle finder (http://www.amazon.com/Wixey-WR300-Digital-Angle-Gauge/dp/B001PTGBRQ/ref=sr_1_2?ie=UTF8&qid=1344975811&sr=8-2&keywords=digital+angle+finder).

For purposes of this specific dialog, can you see that Step1 is analogous to setting up your ramp on a level surface?

Ether

14-08-2012, 17:26

John,

Just for reference, what is the resolution on that inclinometer?

From the picture, there are markings every 2.5 degrees. I suppose a very careful observer could eyeball it to 1 degree.

A 40" long ramp at a angle of 45 degrees changes height by approx 1/2" for 1 degree change in angle.

My tape measure is marked in 1/32" intervals.

Ether

14-08-2012, 17:34

A plumb-bob and a protractor should be a good start.

It would be fun to see a YouTube video of students making that measurement to the required accuracy for useful results.

JamesCH95

15-08-2012, 08:30

For purposes of this specific dialog, can you see that Step1 is analogous to setting up your ramp on a level surface?

That particular angle gauge isn't exactly appropriate then. There are digital angle gauges with a "level" function that does not require calibration on a particular surface.

Jared Russell

15-08-2012, 09:43

If you want to know the CoF of your robot, it's generally to answer one or both of the following questions:

1) How well can my robot push/resist pushing?
2) How well will my robot turn/handle?

#2 seems decidedly more complicated than #1 to me. Then again, #2 depends on a lot more factors than just CoF. To measure #1, wouldn't a swath of carpet and a stationary scale/load cell to push against be sufficient?

It would be swell if a team brought such a rig to the Championship some year. If they actually give the Hall of Fame teams more than a cocktail table next year, maybe that would be a good use of the space...

Joe Ross

15-08-2012, 10:19

It would be swell if a team brought such a rig to the Championship some year. If they actually give the Hall of Fame teams more than a cocktail table next year, maybe that would be a good use of the space...

10 years ago, 494 brought a dyno to championships.

Clinton Bolinger

15-08-2012, 10:27

10 years ago, 494 brought a dyno to championships.

I am pretty sure they still have the Robot Dyno in their facility.

-Clinton-

Ether

15-08-2012, 10:29

That particular angle gauge isn't exactly appropriate then. There are digital angle gauges with a "level" function that does not require calibration on a particular surface.

They do not require calibration by the user because they are calibrated at the factory. That is a source of error, analogous to using a bubble level to locate a level surface for test set-up.

In most modern buildings, it is easy to find a flat level location using a bubble level. Once that location is selected for the test setup, multiple "tilt" tests can be conducted without the need to measure the gravity reference each time.

Whether or not the factory calibration of a device with a built-in level is inferior or superior to the use of a mechanical bubble level is an open question at the moment.

The digital readout of an angle sensor is certainly more convenient1, but that's not the point of contention here.

1It's also more expensive if you don't already happen to have one on hand

JamesCH95

15-08-2012, 11:23

They do not require calibration by the user because they are calibrated at the factory. That is a source of error, analogous to using a bubble level to locate a level surface for test set-up.

In most modern buildings, it is easy to find a flat level location using a bubble level. Once that location is selected for the test setup, multiple "tilt" tests can be conducted without the need to measure the gravity reference each time.

Whether or not the factory calibration of a device with a built-in level is inferior or superior to the use of a mechanical bubble level is an open question at the moment.

The digital readout of an angle sensor is certainly more convenient1, but that's not the point of contention here.

1It's also more expensive if you don't already happen to have one on hand

The electronic angle finder/level that I've used extensively, and that I really like, was calibrated from the factory to +/- 0.2deg of level and +/- 0.1deg angle change from level. McMaster PN 21465A81. Just turn it on and directly measure...

Your average carpenter level is only good to about +/- 1.5deg of level. More precise machinist bubble levels are available of course

You're right, a digital angle finder/level might be considered expensive, but it's the sort of tool that will last for many years of reliable service. It is substantially easier to use and more accurate than a non-digital angle finder. It is especially useful when making complex mill setups or doing complex tube notching and cutting.

EricVanWyk

15-08-2012, 11:52

Good news everyone! I have obtained a magic inclinometer with infinite precision and perfect repeatability! It is pretty sweet.

What is the accuracy of the inclined plane test now? What sources of error still exist?

Ether

15-08-2012, 11:54

EricVanWyk

15-08-2012, 13:50

Curvature of the Earth :)

Maybe curvature of the gravitational field, but at that point we are back to complaining that the weight of the people moving in the stands are throwing off the accelerometer measurements.

I'm just trying to understand out how much of the spread is due to mismeasuring the angle, and how much is from other sources. As soon as it starts moving, it goes into kinetic friction. So, if you jostle the setup it will start moving before it 'should'. My untested hypothesis is that this thread is focusing too much on the angle, and not enough on things like how quickly you raise the ramp, or how steady your hand is.

So - is your money better spent on a better inclinometer, or keeping the freshmen away from the red bull?

Ether

15-08-2012, 13:59

Maybe curvature of the gravitational field, but at that point we are back to complaining that the weight of the people moving in the stands are throwing off the accelerometer measurements.

I'm just trying to understand out how much of the spread is due to mismeasuring the angle, and how much is from other sources. As soon as it starts moving, it goes into kinetic friction. So, if you jostle the setup it will start moving before it 'should'. My untested hypothesis is that this thread is focusing too much on the angle, and not enough on things like how quickly you raise the ramp, or how steady your hand is.

I think even under laboratory conditions with expensive test equipment and experienced technicians you would still see a lot of variation regardless of the measurement methods used. Friction is finicky.

Richard Wallace

15-08-2012, 14:00

So - is your money better spent on a better inclinometer, or keeping the freshmen away from the red bull?Hand being drawn toward the spotlight button ...:]

DonRotolo

15-08-2012, 21:17

is there some cutoff point where the value changes from one state to another or is it a blend.
Mu versus relative velocity is an asymptotic curve approaching zero as velocity increases, but there's a small bump up just after you start to slip (very much depends on the surface and wheel material, so not always).

In general, static friction is higher than kinetic friction. Usually.

As for experimental error: Yes, by having the gravity vector point through something other than the bottom(ish) of the robot, you're changing things and that is usually significant.

A better method if you need actual numbers (instead of comparative effects) would be a calibrated pulling device that drags the robot across the test plate.

KrazyCarl92

20-08-2012, 11:06

I had been toying with the idea of making a simple test bed for finding coefficients of firction earlier this summer while doing some drivetrain CAD.
I was thinking a piece of plywood with carpet, polycarbonate, and HDPE covers that would tilt with a window motor and some sort of sensor to detect angle (pot, encoder, gyro, accelerometer, etc.). Thought it might be a good preseason project with basic motor control, sensor use, and automation.

Then I ran into that dilemma about testing a single wheel/piece of tread vs. testing an entire robot. I think the main problem for most teams (including ours) in testing their actual robot drivetrain configuration is that by the time there is a robot with that particular drive configuration to test, it's already well past the point in the season to make use of that information for design purposes.

Maybe the drivetrain skeleton is finished by week two or so...would you really have the time or resources to restart from scratch with your gearbox based on that information at that point in the season? In some cases it might be as simple as a sprocket size change, but this isn't a one size fits all approach. I can see where a team that uses the same or similar drive configurations from year to year could easily make use of this kind of test bed prior to a point of no return in their design. Even then, the exact CoG, wheel position, weight, etc. might not match those of this year's robot.

Does anyone have thoughts on how a test bed like this could be used in season to produce useful results prior to the drive train design phase of the season (like gathering information in days 3-5 of build)? My question deals more with the utility and practicality of such an apparatus in the actual season rather than the theory, design, or production of the apparatus.

Chris is me

20-08-2012, 12:15

Does anyone have thoughts on how a test bed like this could be used in season to produce useful results prior to the drive train design phase of the season (like gathering information in days 3-5 of build)? My question deals more with the utility and practicality of such an apparatus in the actual season rather than the theory, design, or production of the apparatus.

I think there are a few options:

1. Previous robots - if you do a particular style of drivetrain twice in a row, take an old robot, maybe remove some parts from it, redistribute weight with bricks / batteries to approximate various CGs.

2. Kitbot - This test might not be as functional since the Kitbot frame may be less rigid than a welded tube frame, but for many teams testing with a kitbot is a good approximation.

It's important to note for both of these tests, you don't even need to have gearing established yet, since the wheels will be locked. The key things to simulate in these tests from what I can gather are weight, CG, frame rigidity, wheel type, and your contact polygon, so aim to prototype with those in mind.

I'll go ahead and throw this on the "list of things I wish we did". We've approximated our way into a great drivetrain the past 3 years.

JamesTerm

20-08-2012, 14:09

Don,
Do you have any of the data from the tests you've done, or any more info on how you take dynamic friction into account?

-John

I'm going to throw an idea out there and I hope people will respond with why it would or would not work. As a starting point for explanation this stems from the spreadsheet for the cell "Drivetrain Efficiency", where I propose kinetic friction = %100 - "Drivetrain Effiency". What I've observed in my real and simulated tests is that kinetic friction will reduce the overall top speed that would have happened, and this is easy to measure by use of encoders. So perhaps it could work like this:

In reference to this (assuming you use CIM)
http://www2.usfirst.org/2005comp/Specs/CIM.pdf

1. Try to use the peak efficiency voltage to measure 19.8 amps by setting victors to a constant speed percentage. Measure with amp meter what you have.
2. Read back the rps from the encoder
3. Using the stall torque of 45 Oz-In converted to newtons use spread sheet equations to determine the max speed for the encoder
4. Similar to linearization of victors, get robot running at a steady state of the known speed in step 1... you may record each iteration of the encoders for later analyses

It should fall short of the optimal speed just as the drive train efficiency also lowers the max speed in the spread sheet. This would be because some of the 45 Oz-in torque is used to fight the friction.

My studies also show that Cof impact a delay in acceleration but a quicker response in deceleration (assuming victors are in coast). So far these timings are very small probably 5ms worst case when robot has light payload. I need to do further research for heavy payload.

In regards to dynamic friction and the stribeck curve... I think the steady state should have the most valuable coefficient, and it should be possible to conduct same test at different velocities.

I do not quite yet understand the link between kinetic and dynamic, so for now I propose the kinetic is the overall average one-number that represents the friction but only for one given velocity... just as it is for drive train efficiency cell in the spread sheet.

It should be noted that I believe a similar test to these steps are being done here:
http://www.chiefdelphi.com/forums/showthread.php?threadid=107889

JamesTerm

30-08-2012, 21:26

Hi everyone... today I wanted to share some data in hopes that someone can help explain why our robot has the right side slower than the left. Here is a quick graph:

http://www.termstech.com/files/RateGraphs/TestDirections.gif

And here is the actual numbers it represents:
http://www.termstech.com/articles/TestDirections.txt

What this test does is simply goes full throttle forward, and then full throttle backwards. The green 'v' is voltage left and right (both left and right sides are interleaved within the pixels). The 'Y' grey is displacement in meters. The "p" magenta is the desired speed (p for predicted). The cyan 'e' encoder is the actual velocity recorded. Both 'p' and 'e' are linear velocity measured in meters per second. That leaves the yellow 'eo' which is the PID influence. Each line of text (or two pixel columns) represent 10ms iterations.

At first I thought the speed differences were due to CoF, but looking closer at the right side it does something contrary to that theory. Like CoF it will reduce the overall top speed... clearly when voltage reaches 1.0 the left side always has a faster speed. I wanted to make sure that there was not significant motor bias of one direction to another (e.g. motor brushing placement). So here's the part where it gets interesting... when it decelerates and I try to slow down both sides... the right side all of a sudden goes faster than the left. If it were CoF... this would not be the case in fact just the opposite as high CoF would act like a mild brake and make it decelerate faster. So that leaves two possible answers I can think of:

1. The mass on the right side is much heavier
2. There is not as much current going on the right side given the same amount of voltage.

Also note how the PID tries to increase the voltage but still the right side is going faster... the speeds want to have a 300ms lag from its natural momentum.

I believe it's a lack of current as I have observed it getting worse when the battery gets lower voltage readings (e.g. 11.2)... on several robots.

Does this sound right? Anyone have an explanation why the current distribution becomes unbalanced?

Jared Russell

31-08-2012, 08:55

Possible explanations:

1. With manufacturing tolerances and different wear patterns, it is possible that the right CIMs are "slower but torquier" and the left CIMs are "faster but less torquey". There could also be mismatch and therefore non-equal current draw within an individual gearbox, if there are more than one motor per gearbox.

2. Different retarding forces on each side, including both "coulomb" type stick-slip friction (~proportional to acceleration) and viscous damping (~proportional to speed).

3. Different lengths of wire or quality of crimps between the sides resulting in more or less current reaching the actual motors.

4. Unequal mass.

5. Damaged CIMs. A toasted motor continues to run, but draws more current and produces less mechanical power.

I don't know how much time you want to sink into this project, but a some relatively simple experiments you could run would be swapping CIMs around, measuring wire/crimp resistance with a multimeter, bench testing each CIM/gearbox, swapping whole transmissions left/right, and measuring and balancing weight on each set of wheels.

JamesTerm

01-09-2012, 00:57

Thanks for these items for checklist... I'll want to keep these in mind for our new off-season drive we'll be experimenting on. I guess after seeing this same problem on 4 different robots... it's time to really put some effort into fixing this mechanically.

For now, I have however fixed this problem in software by simply changing the overall scale of voltage prior to applying the polynomial equation to linearize the victors. It looks something like this:

LeftVoltage=(LeftVelocity+m_ErrorOffset_Left)/ (MAX_SPEED + m_TankRobotProps.LeftMaxSpeedOffset);
RightVoltage=(RightVelocity+m_ErrorOffset_Right)/ (MAX_SPEED + m_TankRobotProps.RightMaxSpeedOffset);

Chris is me

01-09-2012, 11:09

Have you calibrated your speed controllers? Improperly calibrated controllers can cause a lot of drift.

JamesTerm

01-09-2012, 13:05

Have you calibrated your speed controllers? Improperly calibrated controllers can cause a lot of drift.

No we have not... that may indeed be the problem. I'll have a chat with the team about this... any good links or tips on how to do this would greatly be appreciated. Thanks.

JamesTerm

01-09-2012, 15:29

No we have not... that may indeed be the problem. I'll have a chat with the team about this... any good links or tips on how to do this would greatly be appreciated. Thanks.

We found this link and it seems pretty straight forward on how to calibrate it.
http://content.vexrobotics.com/docs/ifi-v884-users-manual-9-25-06.pdf

JamesTerm

11-10-2012, 18:05

I just published a post (http://blog.iamjvn.com/2012/08/coefficient-of-friction.html) which highlights my method for calculating the coefficient of friction for robotic drivetrain systems.

John, (or anyone else who has followed the math of this approach)

I found this link
http://tcdprd.autodesk.com/tcdprd/evsfiles/exports/1248534846843/evo_html_4_10/html/15/48.htm

Can this same ramp technique -using testing material (e.g. carpet) on ramp- work to determine the static CoF for wheel traction by simply locking down the gears (i.e. make them never move during this test)? Or is there an easier way. I know wheels are rated, but it would be nice to actually measure it.

p.s. I presume robot must have a low CoG for this test so it will not tip over.

JVN

12-10-2012, 00:25

John, (or anyone else who has followed the math of this approach)

I found this link
http://tcdprd.autodesk.com/tcdprd/evsfiles/exports/1248534846843/evo_html_4_10/html/15/48.htm

Can this same ramp technique -using testing material (e.g. carpet) on ramp- work to determine the static CoF for wheel traction by simply locking down the gears (i.e. make them never move during this test)? Or is there an easier way. I know wheels are rated, but it would be nice to actually measure it.

p.s. I presume robot must have a low CoG for this test so it will not tip over.

James,
Thanks for the link, but that is nothing new to me -- since I wrote it.
It is part of the pre-2012 version of the Autodesk VEX Robotics Curriculum. (New version just went up here (http://curriculum.vexrobotics.com), coincidentally).

Where did you find it linked?

Yes, locking the wheels, putting your robot on a ramp of carpet, and tilting the ramp is the test being described in this thread. That was the math I showed in my initial post.

-John

sanddrag

12-10-2012, 01:03

James,
Thanks for the link, but that is nothing new to me -- since I wrote it.
It is part of the pre-2012 version of the Autodesk VEX Robotics Curriculum. (New version just went up here (http://curriculum.vexrobotics.com), coincidentally).
I don't mean to derail this thread, but I'm surprised to see the 2012 curriculum available now. The website didn't notify me, although I had put in my e-mail address previously. The 2012 curriculum has some rather significant differences from the 2011 and 2010 curriculum. Is there another thread existing someplace where it is discussed, perhaps from an educator-to-educator point of view?

JVN

12-10-2012, 01:16

I don't mean to derail this thread, but I'm surprised to see the 2012 curriculum available now. The website didn't notify me, although I had put in my e-mail address previously. The 2012 curriculum has some rather significant differences from the 2011 and 2010 curriculum. Is there another thread existing someplace where it is discussed, perhaps from an educator-to-educator point of view?

PM me or email me john.vneun@gmail.com. We did a soft launch, in anticipation of a more "announced" launch shortly.

-John

JamesTerm

12-10-2012, 03:01

James,
Thanks for the link, but that is nothing new to me -- since I wrote it.
It is part of the pre-2012 version of the Autodesk VEX Robotics Curriculum. (New version just went up here (http://curriculum.vexrobotics.com), coincidentally).

Where did you find it linked?

Yes, locking the wheels, putting your robot on a ramp of carpet, and tilting the ramp is the test being described in this thread. That was the math I showed in my initial post.

-John

I found this link using google as I was double checking my wording (after chatting with Ether these past few years I tend to double check my wording before I post). The funny thing is that I totally missed this line in your blog:

"Make sure the wheels are "locked" so the robot cannot roll."

So all of this time I thought it was drive train performance that this was trying to find, and why I've been asking about dynamic/kinetic friction... which by the way for the good of the group, I'd like to share the method you presented to me for this:

"
I believe kinetic and dynamic friction are the same thing.
I think that the method you outlined would work if you want to calculate the kinetic friction at any given wheel speed. I believe these values shouldn't change much based on speed. The simpler test to figure out kinetic friction is similar to the static test. You put the robot on a ramp, tilt it until it starts moving, then tilt it back until it stops moving. Take the tangent of the angle at the time it stops moving to find the kinetic CoF.
"

And funny thing... yep... I thought this too was to solve for drive train performance... as I interpret moving as rolling. Doh!

Part of the reason I missed that this test was about traction the first time around was because I was so focused on being able to simulate how fast a robot will move given time, gearing, motor(s) and, voltage. I think for me personally to avoid confusion I may wish to adopt the term Coefficient of Traction.

On a side note I made an interesting discovery today that I do not have an explanation for yet. It deals with some measuring results that "appear" that it took more voltage to slow down the robot (wheels were up on boards too) than to accelerate. I'm sure there is a simple explanation... I'll just need to study a bit more on kinetic energy.