Page 1 of 1

speed control versus torque control

Posted: Wed, 23 Nov 2011, 11:03
by John Caldwell
Which is better, speed control or torque control?

speed control versus torque control

Posted: Wed, 23 Nov 2011, 12:07
by tesla500
True speed control would be very unpleasant. A small change in throttle position would result in a sudden jerk as you accelerated at maximum torque, then another jerk once the set speed is reached and the torque drops down.

Torque control is very smooth, but relies on the driver to regulate the vehicle speed by varying the throttle position.

A DC drive using open loop voltage control is in between the two. It's basically a softer, less jerky version speed control, but isn't as smooth as torque control. Some people prefer soft speed control, others prefer torque control.

For AC drives, torque is what's intrinsically controlled, so you'd have to build a feedback system around that to emulate speed control if that's what's desired.

speed control versus torque control

Posted: Wed, 23 Nov 2011, 12:39
by Electrocycle
with AC and speed control you get regen fairly simply - whereas with torque control zero throttle is zero torque, ie just freewheeling.

speed control versus torque control

Posted: Wed, 23 Nov 2011, 13:08
by John Caldwell
It is the last comment by Electrocycle that worries me. We absolutely must have regeneration. I don't see any good way to get regeneration with torque control. We are using an AC motor and drive.

speed control versus torque control

Posted: Wed, 23 Nov 2011, 14:29
by tesla500
You certainly can have regen with torque control. You could program an offset into the torque command so that zero pedal position = negative torque (regen). Also, you could have a second pedal sensor on the brake pedal that applies additional negative torque when you step on the brake.

All these negative torque commands need to have the qualification that vehicle speed is greater than zero, otherwise you'd start going backwards after stopping!

speed control versus torque control

Posted: Wed, 23 Nov 2011, 14:31
by John Caldwell
Thanks for your thought guys.
JC

speed control versus torque control

Posted: Wed, 23 Nov 2011, 15:04
by Johny
tesla500 wrote:All these negative torque commands need to have the qualification that vehicle speed is greater than zero, otherwise you'd start going backwards after stopping!
That's not quite true. In an AC system, regenerative braking occurs when the motor is running in negative slip. That is, the controller has requested a lower RPM than the RPM that the motor is actually turning.
Generally, in an EV environment, the controller is set to maximum speed and the accelerator provides torque control. A a secondary system - possibly linked to the brake pedal, changes the controller to 0 speed and requests torque proportionally with movement of the brake pedal.

An alternative is that the 0 speed and braking torque is linked to the accelerator pedal perhaps by position of the pedal or some smart system that takes into account vehicle speed (and trend - accelerating or de-accelerating) and accelerator position.

Given the above, a demand of 0 speed and any amount of torque will not result in the vehicle going backwards - that requires negative torque (or in some contollers, negative speed).

speed control versus torque control

Posted: Wed, 23 Nov 2011, 16:06
by tesla500
Johny wrote:That's not quite true. In an AC system, regenerative braking occurs when the motor is running in negative slip. That is, the controller has requested a lower RPM than the RPM that the motor is actually turning.
Generally, in an EV environment, the controller is set to maximum speed and the accelerator provides torque control. A a secondary system - possibly linked to the brake pedal, changes the controller to 0 speed and requests torque proportionally with movement of the brake pedal.

An alternative is that the 0 speed and braking torque is linked to the accelerator pedal perhaps by position of the pedal or some smart system that takes into account vehicle speed (and trend - accelerating or de-accelerating) and accelerator position.

Given the above, a demand of 0 speed and any amount of torque will not result in the vehicle going backwards - that requires negative torque (or in some contollers, negative speed).


That's true for an inverter that has a speed control loop. I was thinking of an inverter that gives you just a pure torque command input.

The internal vector control algorithm used by the inverter will do everything it can to make the torque be what you request. Any speed control is a separate control loop that drives the torque command.

speed control versus torque control

Posted: Wed, 23 Nov 2011, 21:51
by weber
Guys! This is a solved problem.

See the "AC drive programming and pedal mapping" thread, starting here
viewtopic.php?p=30613&t=1859#p30613
and subsequent posts on the following page.

Image

Don't miss the C code in the final post.
viewtopic.php?p=33512&t=1859#p33512

This is working fine so far, with the Tritium WaveSculptor 200 in the MX-ϟ. Tritium just need to increase the rate of their speed feedback to about 20 samples per second to allow it to work more smoothly.

And it's not patented (as far as I know).

speed control versus torque control

Posted: Wed, 23 Nov 2011, 22:32
by bga
tesla500 wrote: True speed control would be very unpleasant. A small change in throttle position would result in a sudden jerk as you accelerated at maximum torque, then another jerk once the set speed is reached and the torque drops down.
Holden cruise control works like this and it's not good.
Each time a a hill is encountered, you get 4 or 5 kicks in the back as it steps up the throttle to manitain set speed (110kph), then backs off to idle at the crest and overspeeds on the other side of the steeper ones.

I don't use it for this reason and try to drive more a modified constant energy (kinetic+potential) profile, allowing the vehicle to slow up climbing the hills.

I think this hints at a desirable arrangement being a hybrid of torque and speed control. The linked thread mentions varying the zero torque point on the accelerator pedal with speed so that foot position provides some input to the speed setpoint.

...Must get to a point that I can experiment on this myself.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 11:39
by John Caldwell
I will step back in at this point. Jerky speed control does not mean that speed control is a bad thing. It means that the controller has the wrong settings. After listening to everyone's arguments, I am inclined to think that the best control system would be speed control with proportional gain only and that gain set low. With proportional speed control and low gain (e.g. a proportional band of 50%) you would get a result much like torque control. The more you press the pedal, the more torque the motor puts out. As speed increases, that torque qould automaticly drop off if you maintain the same pedal position. There is no reason at all why control will be jerky. Thanks everyone for your input. I will try speed control with low gain and see how it works.
Regards,
Johnny

speed control versus torque control

Posted: Thu, 24 Nov 2011, 15:55
by weber
tesla500 wrote: True speed control would be very unpleasant. A small change in throttle position would result in a sudden jerk as you accelerated at maximum torque, then another jerk once the set speed is reached and the torque drops down.
bga wrote: I think this hints at a desirable arrangement being a hybrid of torque and speed control.

I agree with both of you gentlemen. Pure torque control is tolerable, while pure speed control is not, but there is no doubt that the most desirable arrangement is a hybrid of torque and speed control.

The way in which you implement that may be constrained by what inputs your motor controller gives you and what control loops it lets you break into and how those loops are nested. You can start with speed control and make it more torque-like, or you can start with torque control and make it more speed-like.

John, if you haven't already, you may be interested to read the description by my mentor Ross Pink, who like you, began with speed control.
uploads/690/EvanControlMethod.pdf
Ross refers to a diagram, which you can see here.
viewtopic.php?p=23410&t=1859#p23410

speed control versus torque control

Posted: Thu, 24 Nov 2011, 16:54
by Johny
It's worth noting here that Tuarn (acmotor on this forum) used speed control with 100% allowable slip in his Danfoss configuration.
Pure torque control did not give him regen. braking (as a few have stated).
My controller only allows 20% slip so my "limp home" mode (shaft encoder failed so resort to v/f mode) uses speed control with 20% slip - should be just drivable. (Normally using torque control with speed entanglements similar to weber's driver control mapping).

Azure Dynamics did a manual on how theirs worked.
In retrospect it's a bit primitive and smacks of speed control with some kind of motor current tracking. It is where I got the idea of a pedal "dead band" though - just not as big as theirs appears to be.
http://www.azuredynamics.com/products/f ... Manual.pdf

speed control versus torque control

Posted: Thu, 24 Nov 2011, 17:17
by tesla500
The Nissan Leaf seems to implement something very similar to what weber suggests. One thing they're doing in addition is switching to power control rather than torque once the system is in the constant power region at higher speeds. The pedal feel is quite good in the Leaf.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 17:57
by Johny
tesla500 wrote: The Nissan Leaf seems to implement something very similar to what weber suggests. One thing they're doing in addition is switching to power control rather than torque once the system is in the constant power region at higher speeds. The pedal feel is quite good in the Leaf.
Hmmm. I haven't had mine up to field weakening speed yet - neither have weber/coulomb. I wonder if something strange happens to torque control in the motor's constant power region.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 18:27
by tesla500
Johny wrote:Hmmm. I haven't had mine up to field weakening speed yet - neither have weber/coulomb. I wonder if something strange happens to torque control in the motor's constant power region.


The Leaf uses a synchronous motor, so theoretically full torque is available up to maximum speed (no field weakening possible in a PM motor). They're artificially limiting it to 80kW, presumably due to battery limitations. If the battery could handle it, the Leaf drive system should be capable of 300kW at maximum speed.

I've had my inverter into field weakening (at limited input voltage), and the control naturally transitions into power control. A standard vector control algorithm already implements what the Leaf does.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 20:38
by weber
tesla500 wrote:The Leaf uses a synchronous motor, so theoretically full torque is available up to maximum speed (no field weakening possible in a PM motor). They're artificially limiting it to 80kW, presumably due to battery limitations. If the battery could handle it, the Leaf drive system should be capable of 300kW at maximum speed.

I can see how you might think this, but in fact PM motors can be field weakened by phase-shifting the inverter output, and with IPM motors like the Leaf's it can be done over a wider range. And it _is_ done in the Leaf. Dyno curves show that it only produces its full 280 Nm of Torque up to about 3000 rpm, which corresponds to about 45 km/h. This is a _voltage_ limitation, not merely a battery power limitation.

Field weakening PM and IPM
http://www.ornl.gov/~webworks/cpr/pres/104228.pdf
Leaf Dyno test
http://www.mynissanleaf.com/viewtopic.php?f=27&t=5323

The scary thing is what happens to a PM or IPM motor if the inverter fails at high speed. Suddenly the motor starts to generate its full back-emf and there could be some fierce regen braking or very high voltages generated. An induction motor is inherently safe in this situation.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 20:56
by Richo
I don't soley drive my ICE using the cruise control and hence I won't be driving my eV in the same manner.

speed control versus torque control

Posted: Thu, 24 Nov 2011, 21:39
by tesla500
weber wrote: I can see how you might think this, but in fact PM motors can be field weakened by phase-shifting the inverter output, and with IPM motors like the Leaf's it can be done over a wider range. And it _is_ done in the Leaf. Dyno curves show that it only produces its full 280 Nm of Torque up to about 3000 rpm, which corresponds to about 45 km/h. This is a _voltage_ limitation, not merely a battery power limitation.

Field weakening PM and IPM
http://www.ornl.gov/~webworks/cpr/pres/104228.pdf
Leaf Dyno test
http://www.mynissanleaf.com/viewtopic.php?f=27&t=5323

The scary thing is what happens to a PM or IPM motor if the inverter fails at high speed. Suddenly the motor starts to generate its full back-emf and there could be some fierce regen braking or very high voltages generated. An induction motor is inherently safe in this situation.


Thanks for those links. I wasn't too familiar with PM field weakening. It makes sense what's going on now with the power curve of the Leaf.

I definitely prefer induction motors for their intrinsic safety during an inverter failure. I wonder why so many manufactures prefer PM motors to IMs? Isn't Tesla the only manufacturer using IMs?

One thing I don't get from the paper on field weakening: Why wouldn't the motor field weaken itself when short circuited while spinning? Is it because the motor refuses to generate currents of the correct phase to field weaken?

My reasoning: If the inverter produces a reasonable amount of direct component current and can completely cancel out the field of the magnets, effectively no flux cuts the stator windings anymore, resulting in 0V across the windings. If you short circuit the motor while spinning, you've effectively (ignoring winding resistance) forced no flux to cut the stator windings, and there will be 0V across the windings. Why would one of these result in powerful braking, and the other not?

speed control versus torque control

Posted: Thu, 29 Dec 2011, 14:51
by Peter C in Canberra
I have a plain old series DC motor with a Kelly controller. It can do speed (voltage) control, torque (current) control or something it calls 'balanced' mode. Kelly were not very helpful when I asked what that actually means. Without bothering to try the other modes I went for balanced. It feels like it is torque control at lower speeds so I can get smooth, non jerky driving at carpark speeds but at higher speeds I find I don't need to move the pedal much to maintain a cruising speed up and down hills. All in all it is a nicer feel than my petrol car and don't see any reason to emulate the response of a petrol car.