6 CIM Drivetrains... What's Your Experience?

[Mark McLeod]

Originally Posted by Nate Bloom

Even so, you're still limited to 40 amps per CIM...

The Snap-Action 40a Breaker Spec. overload ratings:

150 % overload = 3.9 - 47 seconds
175 % overload = 2.2 - 9.2 second
200 % overload = 1.5 - 3.9 seconds
250 % overload = 0.8 - 1.8 seconds
300 % overload = 0.5 - 1.1 seconds
400 % overload = 0.3 - 0.6 seconds
500 % overload = 0.2 - 0.3 seconds

[Ether]

Originally Posted by Nate Bloom

What's the math for a traction limited drive?

If:

- the wheels are all the same type and diameter and are chained (or belted) together on each side, and

- you are powering each side with a single gearbox with N CIMs, and

- you assume that traction force = normal force * coefficient of static friction (where coefficient is a constant), and

- the center of mass lies along the vehicle longitudinal axis,

then:

A = [(4*D/G*Istall*mu*W)/(eff*Tstall)]/N

where:

A = amps per CIM required to break traction and spin the wheels

W = weight (lbs) of vehicle

mu = coefficient of static friction

D = diameter of wheel in inches

G = the total speed reduction gear ratio from motor to wheel (including sprockets/pulleys)

eff = drivetrain torque efficiency fraction

Tstall = CIM spec stall torque (oz_in)

Istall = CIM spec stall amps

[Nate Bloom]

Just out of curiosity, how do you estimate the drive train torque efficiency?

[Richard Wallace]

Originally Posted by Nate Bloom

Just out of curiosity, how do you estimate the drive train torque efficiency?

An equally pertinent question might be, what steps can a team take to minimize drivetrain mechanical power losses? There are many factors to consider, including alignment, lubrication, fits/clearances, materials, etc.; however, the concept is simple. If it rolls straight and easily, the losses are low. Many of us learned this as Cub Scouts building pinewood race cars.

80% to 85% is a figure often "plugged in" when using drivetrain calculators (e.g., JVN) -- however, some teams build drive trains that are better than that, and quite a few build them worse.

One good test to try is measuring the free current of the completed drivetrain (wheels up).

[Ether]

Originally Posted by Nate Bloom

Just out of curiosity, how do you estimate the drive train torque efficiency?

That's the 64 Thousand Dollar Question.

[Ether]

Originally Posted by Richard Wallace

One good test to try is measuring the free current of the completed drivetrain (wheels up).

That is probably a good test to determine relative efficiency (e.g when trying to tune the alignment, chain/belt tension, etc of a drivetrain), but how would you tease the absolute efficiency from just that one piece of data?

[Richard Wallace]

Originally Posted by Ether

That is probably a good test to determine relative efficiency (e.g when trying to tune the alignment, chain/belt tension, etc of a drivetrain), but how would you tease the absolute efficiency from just that one piece of data?

I don't think you can do that. The free current test is useful for determining if 'tweaks' are moving in the right direction. It cannot capture the effects of losses that depend on loading, particularly those that become more significant at higher loads.

[AllenGregoryIV]

We ran a 6 CIM drive this year and it was okay but we geared it pretty high to be able to keep up with other robots. (5.33:1 than 26:22) with 8 versawheels. We could hold our own in pushing matches but we were always on the edge of brown/black outs. We blacked out once at Razorback and browned out in our first match at IRI other than that our driver kept it with in it's limits.

It was a fun experience and it helped us win a few matches but we will almost certainly be using 4 CIMs and Ball shifters this year instead. It wasn't worth the fear of possibly popping the main breaker in any given match.

[philso]

Originally Posted by SoftwareBug2.0

We have used six CIMs in our drivetrain. However, our drivetrain never used more than four of them at full power at once which probably helped avoid any current/voltage problems. No special software was needed.

What do you mean by this?

Are you delaying the start up of one of one, or more, of the motors on each side to spread out the current peaks?

[philso]

Originally Posted by Richard Wallace

capture the effects of losses that depend on loading, particularly those that become more significant at higher loads.

This would require a calibrated "rolling road" dynamometer and instruments to measure the instantaneous battery voltage and current and the load that the dynamometer was applying. The effect of wheel spin would could be approximated if the dynamometer rollers were wrapped with the proper type of carpet.

Maybe your local race shop might be able to help out

[Ben Martin]

We ran 6 CIM motors through this VEXpro gearbox at 6:1 reduction, the standard KOP belt system, and 4" wheels. Worked great for us, no issues whatsoever all year. When we increased the weight up to 120 + 20 for bumpers + battery and installed traction wheels, the motors gave us that little bit of oomph that we needed to power through to the feeder station. If we were in more sustained pushing matches, we would have found ourselves more at risk of breaker tripping.

[SoftwareBug2.0]

Originally Posted by philso

What do you mean by this?

Are you delaying the start up of one of one, or more, of the motors on each side to spread out the current peaks?

We didn't have any delays to spread out current peaks. We used six CIMs and each of our wheels was powered by two CIMs, so we had three wheels. Our drivetrain would have just been very hard to control if we had turned all the motors on full at once because it would have wanted to spin very quickly.

http://www.youtube.com/watch?v=oMBTmfBnjsc

I was wondering whether people would read closely enough to see that what I had to say didn't make sense for a typical 6-wheel drive.

[Oblarg]

Originally Posted by Ether

That's the 64 Thousand Dollar Question.

Find a surface on which you know the CoF of the wheels to reasonable accuracy, get a high-speed camera (or accelerometer), and measure the acceleration/current draw of the robot running at full power pushing various objects whose dynamic friction on that surface has been measured?

Not perfect, obviously, and rather time-consuming, but it should give you something to compare against theory.

[drwisley]

Originally Posted by Ether

That is probably a good test to determine relative efficiency (e.g when trying to tune the alignment, chain/belt tension, etc of a drivetrain), but how would you tease the absolute efficiency from just that one piece of data?

I use the no load losses from test stands, which we can derive from the current, and add 1% per gear mesh under load. I have some powertrain industry experience to correlate this practice from. This is a good starting point for shift points. Although the 6 CIM rimpull calc's I have put together don't require shifting.

[Ether]

Originally Posted by Oblarg

Find a surface on which you know the CoF of the wheels to reasonable accuracy, get a high-speed camera (or accelerometer), and measure the acceleration/current draw of the robot running at full power pushing various objects whose dynamic friction on that surface has been measured?

Not perfect, obviously, and rather time-consuming, but it should give you something to compare against theory.

Here's the theory to compare it to.

Adjust Kro, Krv, and Kf until the model matches the test data.