Hi

I am in process of building a simple linear actuator that is based on two timing pulleys (same size) with a toothed belt running between them. The belt itself is attached to a linear slide with the moving object attached to that.

My requirement however is that the direction must be able to be reversed very quickly which poses some questions?

1. How fast can the direction of a DC Gear motor be reversed, repeatedly?
2. If the direction cannot be changed virtually instantaneously can I attach a motor to each pulley running in its own direction, switching between the two?
3. If a motor is attached to each pulley should I use a sort of uni-directional coupling between pulley and motor shaft so that motor 2 does not turn against the gears on motor 1 and what would such a coupling be called?

The physical requirements is to move a 1-2kg weight 13 cm then reverse direction and move 13 cm back. Rinse and repeat. The mechanism should be able to complete 3 complete movements per second. The pulley circumference will be +- 360 cm requiring a speed of 60 rpm on the motor .. I think.

 

Just use an air piston to make your job.

 

Yeah, ironically this is supposed to replace an air piston.

Additionally I am hoping to get a bit of position control via encoders ..

 

There are air pistons which have position control on them. They give out an analog voltage according to their position.

 

True true.
The problem however is that I cannot get compressed air to the job ..

 

Ok. If you want to change direction fast you have to use stepper or servo motors.

 

Using belts will wear out quickly. Dc motors get trouble on sudden reversal. You can also try a piston with a crank to slide mechanism, similar to air piston.

 

Using belts will wear out quickly. Dc motors get trouble on sudden reversal. You can also try a piston with a crank to slide mechanism, similar to air piston.

Bingo. That is a very good idea. The motor will rotate in one direction only but the weight will move back and forth.

 

I have built a scotch/slotted yoke mechanism for this already and it works very well.
It does not however allow for adjustable stroke which would be a nice to have ..

If reversing the motor is not an option then two motors will have to be used, one for each direction. Anyone know of a unidirectional shaft coupling that 'free-wheels' in the other direction?

 

Have you considered a stepper motor, or are you committed to a regular DC motor?

John

Edit: For free wheeling in the opposite direction, how would you get any force in the intended direction? It would have to be an active, e.g., magnetic, clutch. In fact, some active clutches/brakes are nothing more than stepper motors.

 

I am willing to try a stepping motor, it does not have issues with quick reverse I assume?
My only concern is the speed, how would one go about selecting the motor based on the required torque at the required maximum speed of say 60 rpm? Of course a stepper also solves the problem of positioning to a large degree...

Insofar the coupling is concerned I was thinking of some kind of unidirectional gear mesh with a spring forcing the gear to couple in one direction and push out the way in the other. This however requires the gear to slide along the shaft .. I'll post a simple pic of the gear in a bit.

 

Rudimentary picture of proposed coupling. Note that the view is from the side and displays only one gear tooth, it may end up having two or three. I also have the shaft descriptions reversed...

 

Your coupling illustrates exactly the problem I mentioned. It is easy to decouple in the same direction that the motor turns, for example, coaster bikes. However, consider when the motor drives the pulley CCW. The resisting force is CW. Now, you want it to disengage CW, so I don't believe it will work.

Steppers are used in a large number of instruments to positioning and placing. Check out an automatic pipettor for filing multi-well trays. You have not mentioned how much power you need.

John

 

Your coupling illustrates exactly the problem I mentioned. It is easy to decouple in the same direction that the motor turns, for example, coaster bikes. However, consider when the motor drives the pulley CCW. The resisting force is CW. Now, you want it to disengage CW, so I don't believe it will work.

Steppers are used in a large number of instruments to positioning and placing. Check out an automatic pipettor for filing multi-well trays. You have not mentioned how much power you need.

John

Keep in mind there will be two motors and two coupling sets... one engaging in the CW direction the other in the CCW direction. Two motors because I have been advised that fast reverse of a single dc gear motors is not a good option.

I need enough power to move 1kg (maybe 2 kg) at .36 meters per second.

As far as the stepper ratings are concerned, I see they often graph the torque curve against pulses per second. For a 1.8 degree stepper is one revolution then 200 pulses? (1.8 * 200 = 360)

 

Keep in mind there will be two motors and two coupling sets... one engaging in the CW direction the other in the CCW direction.

I don't think I was clear, or maybe I am not clear in my own head.

Consider: If a motor running CCW can apply torque to a pully, how can the pulley free-wheel on the motor shaft in the opposite direction without an active clutch to release it? You apply force against a resistance. If there is no resistance (i.e., it free wheels in the opposite direction), then there can be no force.

John

 

Truthfully I only thought of this coupling today so I may very well be off the mark.

The idea is that the coupling on the left pulley will engage with its motor when the left motor turns cw. At this time the motor on the right is switched off but has its pulley and coupling turned by the motor on the left but because its gear direction is reversed (vs the left coupling) its gears just slip over one another. The reverse happens when the motor on the right turns on.

Main thing I am counting on is that the couplings will have their respective gear teeth 'pointing' in opposite directions. In effect the coupling on the pulley always turns with the shaft but only engages the motor shaft in one direction which is the direction that the motor is set to turn to.

 

There are shaft couplings that are unidirectional - they grab one way and are free the other http://www.wmberg.com/catalog/produ...pp/default.asp?code=default_view&OpenTo=20010

 

by feeding back positional information from a DC motor, you'll have a servo drive.. Motion profiling is common practice.

Sudden reversals are not a problem for your motor, however your load is a different story. this is where decel/accel profiles win the day.

I have a pair of substantial DC motors that I pulled from an old teletype machine that have high resolution encoders built in. Ideal for motion profiling. You may still find them in your electronics dump.

 

Forget stepper motors for high-speed, high-torque operation, they 'slip'.

A DC servo motor will be fine for the job, however you need to calculate the torque needed to accelerate/decelerate the masses involved to get to the required position in the required time.

The application is typical of robotics or machine tool servo systems, and best places to look for the required calculations would be a servo manufacturor such a Bosch-Rexroth, Siemens or NUM etc.

(If you give them the requirements and ask for a price on a suitable motor & drive, they will do the calcs and tell you the motor specs as part of the quote..)

Tooth belt drives are incredibly reliable, you should not have any problem with that as long as it is adequately rated for the required power in the first place.

(Many commercial servos are now brushless high-frequency rather than 'DC', so if you talk to commercial suppliers just ask for a servomotor rather than a 'DC Servo')

If you want to do the calcs yourself, the biggest part of the load is probably the inertia due to acceleration.

You are best off using quite high motor revs and an inline gearbox to get down to the required belt speed. Inertial load is changed by the square of the gear ratio, so eg. a 4:1 reduces the inertial load seen by the motor by a factor of 16, while increasing the static torque or force at the load by 4 times.

Hope this is some help!

 

Thanks for all the replies.

Just a quick update, this guy uses a dc gear motor for his linear actuator design (chain driven) ..
http://buggies.builtforfun.co.uk/Sim/linear-act.html

So it would seem to me that a dc gear motor would indeed do the job quite well, and as rjenkins suggested, gear ratio is the key to both torque and possible inertial load problems.

Thanks again, I'll update with a pic if I get it going..

So far, this is the motor candidate:
http://za.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=0420621

Driven by this h-bridge:
http://www.sparkfun.com/commerce/product_info.php?products_id=9107