Do Hub-Motors Slip? - Exploring Traction Control & Limited-Slip Diff

[Raag90]

Hey guys,

I'm doing my final year project on an Electronic Drive System using 2 Brushless motors as hub-wheels for a rear-wheel drive. Just using steering angle and the accelerator to determine the differences in speeds for the drive wheels around a corner, etc.

Now I also wanted to implement some sort of Traction Control. I was thinking of maybe regulating the slip that each wheel could experience:

slip ratio = (Velocity(wheel) - Velocity(vehicle)) / Velocity(wheel)

So if I continously drive the motor at a constant speed, I was wondering whether there is a possibility that the wheel may at all slip? Logic tells me that if I give the motor some input, then it should more or less follow that reference speed. This is assuming that if the wheel rotates at, ie 2000rpm, then even if I lift the rear wheels off the ground (thus creating a slip scenario) then my wheels would continue to spin at 2000rpm?

So my question is, do I need to even take this slip into account?

Even more importantly, how do you design a limited-slip different for this?

Help much appreciated

 

[DickL]

If it is an induction motor there will be slip, and the slip will be some function of the applied torque. If it is a synchronous motor, then no slip. If a stepper motor, I think there is no slip up until a critical value of torque is reached, then it may lose steps or stall.

 

[Raag90]

@DickL:
Thanks for the reply. Yeh we are using BLDC motors (Permanent-Magnet Synchronous Motors) and therefore I assume they will experience no slip. Now that's the issue I'm facing. If both rear wheels start on the road, then they will spin at a speed, say 2000rpm. Now what if I lifted one wheel off the ground. Would it experience slip in the sense that the speed would increase from 2000rpm? As of now, the only way I can think of implementing a LSD is by comparing whether both drive wheels are different to each other, and therefore adjust the applied rpm to each motor accordingly.

 

[Unrest]

@DickL:
Thanks for the reply. Yeh we are using BLDC motors (Permanent-Magnet Synchronous Motors) and therefore I assume they will experience no slip. Now that's the issue I'm facing. If both rear wheels start on the road, then they will spin at a speed, say 2000rpm. Now what if I lifted one wheel off the ground. Would it experience slip in the sense that the speed would increase from 2000rpm?

For a practical vehicle, no matter what type of motor, the wheel would almost certainly slip without traction control.

If the controller tries to deliver maximum power when the driver floors it, then its speed will depend on torque. Even though the motor might not technically be "slipping", the wheel certainly can.

If the controller tries to achieve a set speed, then there will be times (driver floors it) when it's trying to accelerate as fast as it can to get there. Losing traction would only help it speed up. Again the wheel would slip.

 

[DickL]

If you lift a wheel off the ground, its speed will not change - still 2,000 rpm. That is what synchronous motors do. Now imagine 1 wheel staying on the nice dry pavement and the other wheel on wet ice. The wheel on the ice will still turn at 2,000 rpm, but it will be slipping relative to the ground speed (or at least not transmitting much torque).

For a vehicle to run around curves, the drive wheels need to turn at different speeds, or else they will slip relative to the ground. Vehicle drive systems similar to the 1 your describing generally have some means of determining they are curving. I've know of this being accomplished using steering inputs to the motor controllers, or responding to torque differences. Both approaches have difficulties and the choice of approach may be influenced by the conditions under which the vehicle will be used.

Using the average speed of the 2 driven wheels for the traction control can work for many situations, but there are real conditions that can result in 'funny' behavior. If you can also have a speed input from a non-driven wheel or wheels that are using an independent controller that will add an improved level of confidence in the traction control behavior over a wider range of conditions.

 

[Unrest]

If you lift a wheel off the ground, its speed will not change - still 2,000 rpm. That is what synchronous motors do. Now imagine 1 wheel staying on the nice dry pavement and the other wheel on wet ice. The wheel on the ice will still turn at 2,000 rpm, but it will be slipping relative to the ground speed (or at least not transmitting much torque).

Yes but that's not a practical vehicle. Imagine you're stopped and want to accelerate as fast as possible. You say "I want 2000rpm". The motors can't achieve that instantly, so they'll start off slower, either by the motor slipping, or by the controller keeping the speed down to prevent slip.

While it's accelerating you jump off so it's much lighter. Now you'd expect the vehicle to accelerate faster. If it doesn't (because the motors are synchronized with the controller's predetermined acceleration) then it won't be much fun to drive because it'll have the same performance with 10-yr old kid as an overweight aunt.

To say it another way, a "practical vehicle" should be able to take advantage of reduced load to achieve higher speed and acceleration. Spinning wheels is a form of reduced load so they should speed up unless it has some traction control like what the OP is after.

 

[Raag90]

So now that we've deduced that the 'slip wheel' will stay at around 2000rpm whether on ice or on the pavement, we can deduce that comparing relative rpms of the two motors would not be an accurate measure of slip. Instead, if there was a way we could measure the current through each motor, and as torque is dependant on the current through a motor (and its motor coefficient), we can work out each torque each wheel produces.
As T=I*Kt, ------>Kt is known (for each motor).
Now if I divide the torque by the wheel/motor radius, then I should get the force that each one is exerting onto the ground, right? Could this force not be a good indication of how much a wheel is slipping by (lower force means less traction?).

I was also thinking that while taking a turn, if we can calculate each of these forces respectively, and then taking into account the radius of the turn, as a measure of the steering angle, then we could calculate the lateral component of these forces, and combine them as the total lateral force exerted by the vehicle to keep traction (and hence work these components out as accelerations, as we divide by the mass of the vehicle over the two drive wheels). F=ma.

I was thinking of using an IMU (Inertial Measurement Unit) with the accelerometer and gyroscope located at the centre of gravity of the vehicle to determine the amount of outward acceleration experienced onto the vehicle, and then comparing this to the previously calculated vehicle tractive acceleration, in order to determine any tractive limits that may be reached or not.

Would this be one way of doing things?

 

[DickL]

I'm making several assumptions: 1) the tires are more or less conventional pneumatic tires, 2) this is a robot or remotely controlled vehicle, and 3) it is not intended to be used in a wide variety of terrain situations nor at high g forces (e.g. no racing). If 1 is not correct, then I need to know more about the vehicle and the surfaces it will run on. If 2 is not correct, then some aspects become easier, drivers can do a lot. If 3 is not correct, then it becomes more difficult because adhesion on the different wheels can vary greatly and controlled wheel slip may be needed for full and effective control.

Motor current can, and is frequently used, for torque feedback and control. All of the variable speed motor controllers for vehicle propulsion that I've worked with included some level of motor current measurement and feedback. Of course getting a hold of that data can be a problem with commercial motor controllers. Still you can always add an external current sensor and build your own feedback loop. Using the torque/motor current feedback is a reasonable way to make a differential (limited slip or not).

If the speeds are not great and the vehicle is staying on pavement or solid ground you may find it easier to use the steering input as way to effect the differential action.

Once you move out of the well behaved adhesion, reasonable speed and turn radius arena, control becomes considerably more complex because of the many different combinations that must be considered and designed to be handled.

An example of a real situation that was encountered by an automated transit vehicle was smooth ice on the guideway, and a small up grade shortly after its starting position. The vehicle started up, accelerated into the grade and lost traction on all wheels (all wheel drive). The vehicle's wheels sped up per the program, went into cruse mode and were measuring the distance traveled, all the while it stayed in one place on the guideway with its wheels spinning at the commanded speed. Torque feedback would have helped, but only if it were of the necessary sensitivity, overly sensitive and too gross can both cause problems.

Then of course there are the problems encountered by the Mars rovers. If I recall correctly, at 1 point it got caught in soft dust and all wheels spun, much like the above.

When attempting tight radius turns on soft ground, if the differential action doesn't assist the steering, the vehicle will just push the front/steering wheels straight ahead, especially on soft ground.

Frequently in marginal adhesion conditions (heavy, thick snow, muddy ground) it is necessary to have a relatively high amount of wheel slip to generate maximum traction, but if the wheel slip becomes too high traction falls off the cliff. On nice dry pavement, maximum adhesion occurs with a small amount of wheel slip, somewhere around 1% (I'm ignoring drag racing here, which utilizes a different adhesion process), but in snow or thick mud max adhesion may occur upwards of 10% wheel slip.

As I understand the DRPA Baja robot vehicles, they all used conventional (more or less) automotive drive systems (e.g. AWD, differentials, slip-slide control, etc.) coupled with several different types of input data, such as radar for determining actual vehicle velocity, and probably inertial platforms as well.

The conditions in which the vehicle will be used, particularly the boundary conditions will be very important in figuring out how to control the motors and incorporate traction control.

 

[Raag90]

If you lift a wheel off the ground, its speed will not change - still 2,000 rpm. That is what synchronous motors do. Now imagine 1 wheel staying on the nice dry pavement and the other wheel on wet ice. The wheel on the ice will still turn at 2,000 rpm, but it will be slipping relative to the ground speed (or at least not transmitting much torque).

But since Torque and RPM have a negative gradient relationship, if the wheel does remain at constant rpm as you suggest, how would there be any change in torque? I thought that any change in either torque or rpm would cause an inverse change in the other? So even if the wheel is lifted, and continues to spin at 2000rpm, my torque should not change in value compared to when the wheel was spinning at that speed on ground? (even though logically, there would be less torque exerted on the wheel in the air).

 

[DickL]

Synchronous motors work a bit different. Their speed is strictly controlled by the frequency of the current. They stay synchronized with that frequency, unless the break out torque is exceed, then they usually stall. Although the motor rotates at the current's frequency, there is a small change with load. That small change is a phase angle change, not a change in rpm. The motor's torque is primarily a function of the motor's current. With a small phase angle the motor's impedance will be high and the current will be low, and with large phase angles low impedance and high torque (actually it is more complex than that).

Going back and looking at synchronous motor characteristics again has raised a question in my mind as to whether you are using them. I've worked with some brushless synchronous motors, but the common ones have brushes and slip rings. Could your motors be induction motors, or brushless DC motors? A sudden reapplication of traction can cause a synchronous motor to break out of synchronization and stall. Most electrical motors do have a drooping rpm characteristic with torque, but not synchronous motors. Your comment about the torque / rpm relationship also suggests the motors maybe some other type.

 

[Raag90]

We are using Brushless DC motors for the project.

Your point makes me think about something else. I've always wondered how the following scenario holds true:
BLDCs have characteristics similar to DC motors in the sense that:
->Ea=Wr*Kt
->T=Ia*Kt

Now we know that RPM is proportional to the Voltage (controlled by PWM). So the more voltage you apply, the faster your motor spins. But we also know that Torque is proportional to Current. Now if my motor characteristic impedance stays constant, would'nt my Voltage be proportional to the Current? Then why the inverse relationship between torque and RPM?

 

[HowlerMonkey]

Until you provide more torque than the drive wheel to pavement union can withstand, there will be no need for traction control.

You could always do what the audi guys did with their rear wheels when the AWD was declared illegal in their class.

They left the worm gear differential connected but left it undriven.

In your case, why not solve this with a mechanical link between them?

I work on traction control for lexus and it is immensely complicated and still unable to accurately control the rpm of all wheels.