DC Motor running as a Generator

[IrisLab]

Can someone explain to me the effects/consequences of running a DC motor as a DC generator?

In particular, suppose we have a DC motor where we attach a variable resistor R across the leads of the motor. This R is our load for our generator. Now, suppose we turn the shaft of the motor (generator?) with a torque T and angular velocity w. This would induce a voltage v across R and a current i through R.

I'm assuming (neglecting losses) that Tw = vi = (v^2)/R = R i^2.

Suppose that I have can turn the shaft with a constant T and that w is free to seek whatever angular velocity. Now, if I vary the R, I would change the current draw and thus vi would vary. With T constant, w would have to vary to compensate, right?

Any help on this matter would be greatly appreciated! Thanks.

[Mike Betts]

"neglecting friction"... Yes.

To expand: Energy must be conserved: wT == vi

[Al Skierkiewicz]

Originally Posted by IrisLab

Can someone explain to me the effects/consequences of running a DC motor as a DC generator?


I'm assuming (neglecting losses) that Tw = vi = (v^2)/R = R i^2.

Dave,
I am guessing you are asking if mechanical power in would equal electrical power out. That assumption is correct except for the introduction of an efficiency adjustment. (Of course this only holds true for true DC brush type motors, brushless motors won't generate a voltage.) Motors are not designed for generating power in most cases so the efficiency is not optimized for generation. I am guessing 80-90% efficiency at best. Keep in mind that a lower R is translated into a greater demand on torque input. Electric trains use just such a system for braking but the motors are designed for this duty. The control system just switches in more resistance in parallel (lower R) to reduce speed. When the robot speed controllers are in the brake mode, they just short the output leads and take all the mechanical energy converted to current into the short. The back EMF generated in the motor by the high current effectively brakes the motor.

[IrisLab]

Al, Thanks!

Originally Posted by Al Skierkiewicz

Keep in mind that a lower R is translated into a greater demand on torque input. Electric trains use just such a system for braking but the motors are designed for this duty. The control system just switches in more resistance in parallel (lower R) to reduce speed.

My intent was to use one of the FIRST motors in the Kit of Parts as a "controlled brake" where we'd hang a variable resistance across the leads of the motor. By varying the resistance, we could adjust the torque/speed on the DC motor (generator) shaft.

It is difficult for me to explain the configuration that I have an interest in but another example might be illustrative. Consider a pully type configuration where a heavy weight is on one end of the pully and a DC motor (generator) is on the other end.

If the heavy weight is allowed to fall due to gravity, could I control the resistance across the leads of the motor to gradually bring the falling weight to a controlled stop?

(A) I realize other methods may be suitable for this task, but again, I'm only using the task to illustrate the concept. It's not the actual task that I'm interested in. If this makes sense, what motor in the FIRST kit of parts would make sense to use?

(B) For heavy loads, I'm thinking that the variable resistance controlling the torque/speed of the DC motor (generator) will see a very high current. That resistance control will need to absorb a lot of power. Any ideas on how to set up a "variable resistance" that could handle the power? One thought I had is to use a power transistor and run it in an ohmic region. (I really don't think I could design this control because I'm not that smart.)

(C) As an extension of (B), I would like to avoid a potentiometer (if one even exists to handle the power). I would like to control the resistance through an output of the FIRST robot controller either an analog or a couple of digital outputs. Any thoughts on a design here?

I know this seems like a crazy thing to be thinking about, but the primary objective here is to use this "controlled break" to avoid any current drain from the battery. In others words, if you just reverse everything above and run the DC motor as a motor and use it slow the fall of the weight, then you have to pull a certain amount of power from the battery that has a limit relative to the fuses (30 amps?).

But, as generator, your concern (which is probably a big concern) is if you'll smoke the motors. I'm not so worried at this point about smoking the motors, I'm more worried about drawing more than 30 amps.

Does this all make sense?

[Joe Johnson]

The speed-torque curve still is in play. Actually the voltage-speed-torque curveS ARE still in play. Not only that but the voltage-current-torque curves and the voltage-torque-effeciency curves apply as well.

I think about this situation as the following sequence of events:
#1 Input a torque, T.
#2 The shaft will start spinning, that generates a voltage, and current will flow if you have a load across the motor leads.
#3 The depending on the electrical resistance the shaft will reach a steady state speed, w.
#4 Given a T and a w, a steady state voltage, V, is defined (essential there are a family of parallel T-w lines for the motor that vary by voltage, once you pick a particular point T&w pair, you pick a particular T-w line, that T-w line corresponds to a particular voltage).
#5 Given a T and V, a particular torque-effeciency curve is defined. You already know T, so you simply look up effeciency on the torque-efficiency curve the corresponds to that voltage.
#6 Given effeciency, you can use the following formula to figure out how much power you are generating: Power Out = V*I = Power In * Effeciency = T * w * eff.

All this is very approximate because I neglect mechanical losses and motor curves are not linear in real life AND, finally, all the analysis is steady state -- dynamic effects may or may not be significant.


I don't know if this helps you or not, but it makes things a bit clearer for me.

Joe J.

P.S. I am an M.E. by training, so as like as not I have used the wrong wording or terminology here or there. Even so, I think that I have the jist of things more or less correct. As I often say at my day job, "I am not a E.E., but I play one on TV..." ;-)

[IrisLab]

Originally Posted by Joe Johnson

I don't know if this helps you or not, but it makes things a bit clearer for me.

Joe J.

P.S. I am an M.E. by training, so as like as not I have used the wrong wording or terminology here or there. Even so, I think that I have the jist of things more or less correct. As I often say at my day job, "I am not a E.E., but I play one on TV..." ;-)

Thanks! This does help. The main question that I was having and you hit the nail on the head is if the Toque/Speed curves are pretty much the same. The above thought flow clarifies it for me as well.

Dave

P.S. I am supposedly an E.E., but as I say, just because I'm an E.E. doesn't mean that I can fix your computer or wire your house. :-) My area is image processing and I have long since forgotten DC motors and discrete components. These past month or so with FIRST has been very refereshing...sorta like an old farm boy getting back to the ranch after living in NY city. Unfortunately, this farm boy has forgotten how to milk cows and feed chickens. It ain't exactly like riding a bike. (Hmmm...too many analogies?)

[Al Skierkiewicz]

Dave,
I am going to make the assumption that this application is NOT a FIRST robot since what you are asking to do does not fit exactly the rulebook. That being said, remember that a falling weight, slowing to a stop will have a varying input to the motor and thus a varying output current. Even a fully shorted motor will still be able to move so a mechanical brake will have to be employed as well. The choice of motor will depend on the weight and pulley design but it would seem that the drill motor with transmission ought to do a nice job for weights under say 10 lb. (You will have to experiment with motor depending on the weight and length of travel. i.e.the chippy seems to high an RPM to be effective for small weights over short distances.) You will be back driving the transmission so the motor load would be able to allow more finite control. As to producing the variable resistance, some form of feedback will be needed such as a tachometer on one of the pulleys. Integrate the output of the tach so you have some reference voltage, feed that to a simple multistage window comparator. Use the output of the comparator to switch in fixed resistors in parallel with the motor. Add an end of travel switch to click in the mechanical brake and you are done. For more sophistication use the output of the tach for speed and distance reference and servo the loop to the resistors and brake. A Basic stamp would work nicely I think.
True it's not like riding a bike but you know this stuff. You just have to remember where you put the info in the garage of your mind. A little search and you will find the right box. The downside is the other stuff you are going to find in the mean time. (No comments Joe!)

[IrisLab]

Originally Posted by Al Skierkiewicz

Dave,
I am going to make the assumption that this application is NOT a FIRST robot since what you are asking to do does not fit exactly the rulebook.

Actually, yes it is for a FIRST bot. To be honest, I don't know the rules completely except what I've picked up from this forum.

Would you not be aloud to run one of the FIRST motors as a generator? Is there a restriction?

It's difficult for me to explain, but what I've been looking at is the Toyota Prius drive system and in some sense the ThunderChicken CCT from the previous year. The CCT uses two DC Motors with a planetary gear system to achieve a CVT (continuously variable transmission). The Toyota Prius does something similar. Toyota however uses the car's internal combustion engine (ICE, would be nice to have on a FIRST bot :-), a generator and a planetary gear system to also achieve a CVT. (The Prius has another DC motor that provides power assist, but that's not really important at this point.)

After haggling over the Prius stuff a bit, I posted a few times on CD and some one sent me the White Paper on the CCT. After reading through the Chicken white paper, the current draw appears to be a major issue (or more of a concern than an issue). It seems to me that if you just do like Toyota and instead of use two motors to drive the planetary gears in the CCT, you could run one motor as a generator and avoid the current draw problem.

Would that fall in line with FIRST rules?

[Al Skierkiewicz]

Dave,
As a strict intrepretation of the rules from 2003 and those relating to custom circuit boards, custom circuits may not directly connect to any motors. They may connect to sensors on the robot and to the robot controller. As an alternative, you could use one of the provided servoes to actuate a brake band in a custom transmission and I think that would fit in the rules under both mechanical and electrical rules. If you were to design a transmission that needed to use a motor in freewheel at one time and brake at another time you could use a servo to actuate a microswitch that in turn connected to the brake jumper on the speed controller. This is an alternative that has been used by teams for many years. The servo, under software, moves to close or open the microswitch thus switching the speed controller from brake to coast.
I doubt that the staff will make much of a change in this area of the rulebook over last year. It is a safety issue more than anything else.
I hope Paul sees this post and comments but my understanding of the Thunderchicken tranny was that the second motor added more power to the system not simply to change speed. Look close and I think you will see that the ring gear can be either driven one direction or the other or stopped. I will defer to Paul who is way better at describing this.

[IrisLab]

Originally Posted by Al Skierkiewicz

Dave,
As a strict intrepretation of the rules from 2003 and those relating to custom circuit boards, custom circuits may not directly connect to any motors.

Ah! I didn't know that. That does eliminate that as an option, but I can definitely understand the safety concern there.

I've emailed Paul, and we are discussing his design at the moment to help me understand it better.

If anything comes from that discussion, I'll post it here or another thread.

Thanks!

[Rickertsen2]

Originally Posted by IrisLab

Ah! I didn't know that. That does eliminate that as an option, but I can definitely understand the safety concern there.

I've emailed Paul, and we are discussing his design at the moment to help me understand it better.

If anything comes from that discussion, I'll post it here or another thread.

Thanks!

If i understand you correctly, you are trying to build a variable reluctance brake by putting a varaible electrical load across the the terminals of a backdriven motor. There is a way to LEGALLY do this. The speed controllers have a jumper which when enabled, causes them to act as a short circuit and effectively a brake, whenever they are sent a signal inside of their deadband. Well anyway all you have to do, if you want variable braking is to send them a duty cycle modulated signal, that switched from being inside the deadband, to just barely ourside it, but not enough so to make the motors even move. In effect you sortof modulate a PWM signal onto a PWM signal.

As far as mechanically how well this works, I am very skeptical. With motors this size the brakign effect isn't all that great as anything but high speeds, and as mentioned above all the curves still apply, so as it brakes and slows things down, it will generate less back emf, and thus less "countertorque"(for lack of a better word), and thus not works as well. I implemented a system similar to this on an electric scooter as a secondary braking mechanissm and as sortof an experiment. It worked pretty well at high speeds but did little to nothing at low speeds.

[IrisLab]

Originally Posted by Rickertsen2

As far as mechanically how well this works, I am very skeptical. With motors this size the brakign effect isn't all that great as anything but high speeds, and as mentioned above all the curves still apply, so as it brakes and slows things down, it will generate less back emf, and thus less "countertorque"(for lack of a better word), and thus not works as well. I implemented a system similar to this on an electric scooter as a secondary braking mechanissm and as sortof an experiment. It worked pretty well at high speeds but did little to nothing at low speeds.

Pure braking is not so important as being able to control the DC motor (CCT) or generator (Prius) to augement/control/manipulate (not sure how to phrase this term) the planetary gears. The way I understand a CVT with a planetary gearing is that you have input DC motor that runs at optimal efficiency (To/2, wo/2). Then you run another DC motor or generator at some non-optimally efficient Tr, wr to achieve a constant T for any w at the output. To do this, you need to be able to really control the Tr, wr. I'm curious as to how the CCT does so.

[gc02]

Originally Posted by IrisLab

Then you run another DC motor or generator at some non-optimally efficient Tr, wr to achieve a constant T for any w at the output.

If you're trying to achieve constant torque it could probably be achieved more simply through the electronics. It would entail reading the motor's w and adjusting the speed controller's duty cycle accordingly.

[Al Skierkiewicz]

Originally Posted by Rickertsen2

If i understand you correctly, you are trying to build a variable reluctance brake... There is a way to LEGALLY do this. The speed controllers have a jumper which when enabled, causes them to act as a short circuit and effectively a brake, whenever they are sent a signal inside of their deadband. Well anyway all you have to do, if you want variable braking is to send them a duty cycle modulated signal, that switched from being inside the deadband, to just barely ourside it, but not enough so to make the motors even move.

Rick,
There is no way to legally accomplish this since the braking jumper on a speed controller cannot be controlled by the custom circuit board. It can only be controlled by a switch used as a sensor.
<R52> Custom Circuits may not:
• Directly affect any output devices on the robot, such as by providing power directly to a motor, supplying a PWM signal to a speed controller or supplying a control signal to a relay module.

[Joe Ross]

Originally Posted by Al Skierkiewicz

Rick,
There is no way to legally accomplish this since the braking jumper on a speed controller cannot be controlled by the custom circuit board. It can only be controlled by a switch used as a sensor.
<R52> Custom Circuits may not:
• Directly affect any output devices on the robot, such as by providing power directly to a motor, supplying a PWM signal to a speed controller or supplying a control signal to a relay module.

First of all, I would ask for clarification about this. I wouldn't be suprised if it was allowed since it was something that was never thought of before (such as the ruling last year that the programming port could be used as a general purpose serial port).

However, depending on how much you want to pulse that switch, you can use a servo to press a mechanical limit switch, or an electromagnet that turns on or off, activating reed switch, or a light that activates a optical switch.