Need a motor but don't know the type

[DumboJet]

Hello!

I recently had an unusual idea flushing through my brain and I definitely need some help figuring out the kind of parts I need (in case I find spare time to implement it - if it is possible that is...).
I need an electric motor that will have the ability to :
- Run/Spin at a fairly constant speed.
- While spinning I must be able to accelerate or decelerate it at high velocity (I guess that means I need a low response time). What I have in mind needs to accelerate or decelerate the spin  for time spans of duration let's say equal to 1/4 of the rotation period and then stop.

I don't know:
- If it's doable.
- If I should use a DC or AC motor and what type.
- I guess there are more things that I'm missing here...

If someone has free time to bother with my post, it will be great! 
Not that I will build anything useful or interesting, but you never know...

[robrenz]

I am no expert but I think you need a very high perfomance servomotor with a matching amplifier. Not going to be cheap. You will need to know the rotational inertia of the load you are driving and the max radians/sec/sec accel and decel that you need to accomplish your task. Then you will be able to size the servo motor you need.

[DumboJet]

Thanks. I can calculate the rotational inertia of the load and the max radians/sec/sec after I decide the exact scale of the construction.
I have a lot to learn on servo motors...
Here are some stuff that I've found:

Matching servomotors and amplifiers:
http://fab.cba.mit.edu/classes/MIT/961.04/topics/amplifier_servo.pdf

I guess that I will need a controller too, right?

[Mechatrommer]

i recently bought a knock off dremel that uses ac motor variable speed circuit, should be not that hard to look up in internet or some book maybe...

[NiHaoMike]

If a very small motor is all you need, a stepper or servo motor out of a really old hard drive might work.

Another trick that might work is to take a really oversized, slow motor (maybe out of a treadmill or something) and gear it up. By keeping the motor slow, its inertia would matter less.

[robrenz]

These sigma5 from Yaskawa are very good.

[Psi]

A RC brushless DC motor (inrunner or outrunner) can be controlled for the speed you want.
And since its pulses you can control the rotational speed exactly.

[robrenz]

Not trying to be rude, but the OP's requirements for controlled rapid accels decells, and stop all within 90 degrees of rotation requires high bandwidth bidirectional servo type drive.  Going from max speed to zero in 90 degrees is not going to happen with a stepper or a regular motor speed control. 

[Psi]

That's what gearing is for

[chickenHeadKnob]

Note to original poster, in physics delta V accel and decell are just sign changes but in practice there is a big asymmetry. Rapid acceleration is going to be more costly than de-acceleration  as you can use a clutch or brake for the latter and just dissipate the heat. Check your requirements.

[firewalker]

Someone could use a K.E.R.S. for rapid acceleration.

Alexander.

[Zero999]

i recently bought a knock off dremel that uses ac motor variable speed circuit, should be not that hard to look up in internet or some book maybe...

That's not a true AC motor but a universal motor. It appears to use phase controller to alter the speed.

That combination may work well but it depends on the application.

How much torque and at what speed to you require?

[DumboJet]

Not trying to be rude, but the OP's requirements for controlled rapid accels decells, and stop all within 90 degrees of rotation requires high bandwidth bidirectional servo type drive.  Going from max speed to zero in 90 degrees is not going to happen with a stepper or a regular motor speed control.

It is not necessary to completely stop the rotation within 90 degrees, but I surely need the motor to be able to make rapid (and quite strong) accelerations that may change in direction (whether it will accelerate or decelerate) within 90 degrees. I guess, however, that a motor able to do this will be able to stop the rotation within 90 degrees too.
"High bandwidth bidirectional servo type drive" seems interesting. I will definitely check it out! 

Note to original poster, in physics delta V accel and decell are just sign changes but in practice there is a big asymmetry. Rapid acceleration is going to be more costly than deceleration  as you can use a clutch or brake for the latter and just dissipate the heat. Check your requirements.

Yes, I have thought of that too! However, brakes are usually mechanical in nature (and may wear off in time) so perhaps their addition lowers the overall system reliability (assuming that the motor itself is a simple and reliable mechanical system). But I don't know. I definitely need to put some thought in that. 

Someone could use a K.E.R.S. for rapid acceleration.

Alexander.

I think that could help too, in order to save energy!  I haven't thought of that before!
I just don't know how much it will affect the complexity of the system. It certainly seems to need a lot of research and effort for integrating a KERS in a system.

So, I guess that the best bet for now is a high bandwidth bidirectional servo type drive.


Thank you all !
You were all very helpful ! 

[robrenz]

It is not necessary to completely stop the rotation within 90 degrees, but I surely need the motor to be able to make rapid (and quite strong) accelerations that may change in direction (whether it will accelerate or decelerate) within 90 degrees. I guess, however, that a motor able to do this will be able to stop the rotation within 90 degrees too.
"High bandwidth bidirectional servo type drive" seems interesting. I will definitely check it out! 

Reversing direction requires stopping

Yes, I have thought of that too! However, brakes are usually mechanical in nature (and may wear off in time) so perhaps their addition lowers the overall system reliability (assuming that the motor itself is a simple and reliable mechanical system). But I don't know. I definitely need to put some thought in that. 

A brake is not what you want for deceleration performance. However depending on your application you may want a failsafe brake that locks the motor shaft on loss of power

I think that could help too, in order to save energy!  I haven't thought of that before!
I just don't know how much it will affect the complexity of the system. It certainly seems to need a lot of research and effort for integrating a KERS in a system.

Some servodrives can do regenerative deceleration and as the DC bus goes over voltage it gets pumped back into the ac input. a KERS is not necessary in that case

[Kremmen]

Back in the day i was involved in heavy industrial motion control systems. Based on that experience i would start by putting some numbers on paper first, because can be this will not be a low cost solution.
- The OP says spin at fairly constant speed. But what speed exactly 10 rpm? 10 000 rpm?
- what is the mass of the spinning objet. Or more to the point - what is its moment of inertia? Once you know that, and the angular velocity, you can then calculate the necessary torque to accelerate / decelerate it at whatever rate you need. You don't need to know exactly, even ballpark numbers will give an idea of the order of motor size.
- proper servodrives in the double digit Newtonmeter range won't be cheap. Check eBay - proper Yaskawa gear that is commonly available, say Sigma II series, will set you back about a kilobuck for the motor and servopak. And after that you need a system to drive the servo.

I have to say that it does not look promising/easy.

[DumboJet]

Reversing direction requires stopping

No, I mean reversing the torque direction for the acceleration (so that acceleration becomes deceleration and vice-versa).
I don't mean reversing the direction of the rotation.
Basically, I need the motor to turn in one direction carrying some weight with inertia and accelerate/decelerate the rotation rapidly.

A brake is not what you want for deceleration performance. However depending on your application you may want a failsafe brake that locks the motor shaft on loss of power.

I don't think that I will need a fail safe break.

Some servodrives can do regenerative deceleration and as the DC bus goes over voltage it gets pumped back into the ac input. a KERS is not necessary in that case.

But is there a motor that is both high bandwidth bidirectional servo type drive and able of regenerative deceleration at the same time?
Anyway, saving energy is not so important right now I guess.

Back in the day i was involved in heavy industrial motion control systems. Based on that experience i would start by putting some numbers on paper first, because can be this will not be a low cost solution.
- The OP says spin at fairly constant speed. But what speed exactly 10 rpm? 10 000 rpm?
- what is the mass of the spinning objet. Or more to the point - what is its moment of inertia? Once you know that, and the angular velocity, you can then calculate the necessary torque to accelerate / decelerate it at whatever rate you need. You don't need to know exactly, even ballpark numbers will give an idea of the order of motor size.
- proper servodrives in the double digit Newtonmeter range won't be cheap. Check eBay - proper Yaskawa gear that is commonly available, say Sigma II series, will set you back about a kilobuck for the motor and servopak. And after that you need a system to drive the servo.

I have to say that it does not look promising/easy.

Actually, what I have in mind can be scaled in size so I hope it can become cheap at the right size (but I'm not sure it will).
Assuming that the volume of the motor will be 2x2x4=16 cm^3 approximately (which may be the smallest size possible), based on my calculations, I will need a torque of 0.26 N*m (hopefully).
What I don't know, is if there is a motor with these approximate dimensions, that can respond so fast and that can create such torque.
I have to look around... There are so many electronics companies and so many different motors, that make me feel dizzy... 

Here there are some DC motors with small dimensions that MAY do the job (or not):
http://www.velmex.com/motor_torque.html
Perhaps I could try working with some cheap DC motors first and see what happens.

[Kremmen]

Please note that what you linked are _not_ DC motors, but steppers. Vexta is a known quality mfg so you may trust their spec sheets. You will notice that those motors are just capable of producing the torque you require, but only at the very low end of their speed range. It is critically dependent on your rpm requirements whether they will do the job or fail miserably, since at any speed their torque will drop to a fraction of the max value. This is a typical feature of steppers and all will exhibit this behavior. Another issue is their tendency to resonate at certain step frequencies dependent on the combined inertia of load and rotor. The resonance is worst in full step mode, which also produces the most torque so watch out for that.

Can you be more specific as to what is the constant speed and deceleration/acceleration profiles of your application? Also, if you can tell what kind of mass you rotate, i can help in calculating the mission parameters for the motor. If you want help, that is.

[DumboJet]

...but only at the very low end of their speed range.

That can be a problem.

The mass will be 26g attached at a distance of 2cm from the axis of the rotor.
As I said before, I need a torque of 0.26 N*m while the motor runs (so I guess stepper motors are not so good choice).
The average speed of the motor should be around 250 rpm.

I think a motor that runs with DC voltage would be more handy (I guess that it will require less gear, e.g. no need for AC power supply, etc) but I don't know if I will be able to accelerate/decelerate it.

[robrenz]

The mass will be 26g attached at a distance of 2cm from the axis of the rotor.

If I understand, you are saying that it will have 52g.cm of imbalance.

As I said before, I need a torque of 0.26 N*m while the motor runs

If that is steady state at 250 rpm what is that torque from?

[DumboJet]

If I understand, you are saying that it will have 52g.cm of imbalance.

Yes.

If that is steady state at 250 rpm what is that torque from?

I've expressed it wrong. I meant:
"As I said before, I need a torque of 0.26 N*m applied on frequent intervals while the motor runs."
The 250 rpm will be the mean value of the speed and every now and then I will need some torque to accelerate/decelerate it.

[robrenz]

What is your delta rpm both plus and minus from the 250?
If I have followed so far you want that delta rpm to occur in .06 Second (1/4 rev at 250 rpm).
What steady state rpm variation is allowed at each of the 3 speeds?
What % overshoot and undershoot is allowed during the speed change?

[Kremmen]

I did some quick calculations. Have to check them later when i have the time, but initially it looks like this:

- assume you start from 250 rpm and decelerate to (say) 50 rpm over a quarter turn
- the time is 0,15 s
- the (constant) acceleration will be 139,6 rad/s^2

- from your description of the mass (26g, 2cm offset) the moment of inertia is 10.4 * 10^-6 kgm^2

The torque required to accelerate that inertia in 0.15 s over a speed difference of 20.9 rad/s is just 1.5 * 10^-3 Nm. This looks real tiny but could be due to the small mass. I will check later but now i have to run.

Edit: torque exponent is negative.

[robrenz]

I think 1/4 rev at 250 rpm is 0.06 seconds.

[DumboJet]

Let's see:

250 rounds -> 1 min = 60 sec
1/4 rounds => 60/(4*250) sec = 0.06 sec

u1=250 rpm  =>  w1 = 25*pi/3 rad/sec
u2=50 rpm  =>  w2 = 5*pi/3 rad/sec

The difference in the kinetic energies equals the torque work:
dKE = W  =>
1/2*I*w1^2 - 1/2*I*w2^2 = F*r*angle  =>
1/2*I * (w1^2 - w2^2) = F*r*angle = T*angle  =>
T = 1/2*I * (w1^2 - w2^2) / angle  =  1/2* (m*r^2 + I_rotor) * (w1^2 - w2^2) / angle  =>
T = 1/2* (0.026*0.02^2 + 0) * [ (25^2*pi^2-5^2*pi^2) / 3^2 ] / (pi/2)  =>
T = 0.0021781709065 N*m 

I am a little bit rusty at physics and calculations but perhaps I got it right this time.
Anyway, a motor with an average volume of 16 cm^3 that can give a torque greater that this, will be useful too.
In my calculations I assume that the inertia of the rotor is zero and that the motor responds and accelerates instantaneously.
Generally, I don't think any of these are true in a real application so it good to demand a greater torque than this.
I don't think that a motor at this scale will be such a great cost, so perhaps with a more powerful motor I could even increase the average rotation speed (I don't really have an upper limit on this value). The ability to increase the weight if needed would be nice too.

Perhaps we should get it the other way around : from DC motors with an average motor-volume of 16 cm^3, what's the maximum torque (without sacrificing the rotation speed) that I can get and which motor provides it?

EDIT : This site does not seem to support unicode so I had to change some unicode letters that I had used.
EDIT2 : Added some more info.

[robrenz]

These people have what you want.  There are surely much cheaper alternatives thatn them but the performance is there.  You may look at RC electric car motors.

[DumboJet]

What is your delta rpm both plus and minus from the 250?
If I have followed so far you want that delta rpm to occur in .06 Second (1/4 rev at 250 rpm).
What steady state rpm variation is allowed at each of the 3 speeds?
What % overshoot and undershoot is allowed during the speed change?

A very likely scenario is to be moving between [50-250-450] rpm.
I don't really care much for the precision of the speed (at any of the 3 states of velocity). An error up to 5 rpm might be acceptable (for the values mentioned previously).
I don't care much for the overshoot either (although a large overshoot than reverses the direction of the rotation might not be such a good situation to have).

These people have what you want.  There are surely much cheaper alternatives thatn them but the performance is there.  You may look at RC electric car motors.

I'll take a look. RC electric car motors might be a good idea too. I check them out too. 

EDIT : Added the " An error up to 5 rpm might be acceptable.".

[DumboJet]

T = 0.0021781709065 N*m 

It seems that for the acceleration from 250 rpm to 450 rpm it needs a larger torque (although its the same difference in rpms):
T = 1/2(0.026(0.02)^2+0)(((3*15pi)^2-(25pi)^2)/3^2)/(pi/2) =>
T = 0.0050823987818 N*m

Perhaps we should get it the other way around : from DC motors with an average motor-volume of 16 cm^3, what's the maximum torque (without sacrificing the rotation speed) that I can get and which motor provides it?

It seems I've started becoming greedy! 

[DumboJet]

Thank you all.
Perhaps a torque of 0.02 N*m or 0.01 N*m is enough and more than enough for what I want (assuming that I got the math right...).
Now it's time for me to decipher manufacturer sheets and figure out where they keep their price tags hidden... 
Someone must write a book called "Electric Motors And Manufacturer Sheets For Dummies".