"external" torque sensor

[tomdarwinaustralia]

Hi,

here is my problem: I'd like to measure the torque produced by an electrical motor very accurately. (accurate<O.OO5 N.m)

The point is: I'd like to use an "external" sensor and not a classic axial torque sensor.
I've tried to use piezo electric sensors, in order to measure the reaction force between the housing of the motor and the ground.
But I have, with this solution, some vibration issues (1st natural frequency occurs at 260Hz..) because the pz sensors aren't stiff enough to support the motor.

Any idea/suggestion about external/indirect torque measurement ?
Any help would be appreciated!

Tom (french! sorry for my poor english, I did my best..)

 

[GregLocock]

What size and weight is the motor roughly?

I'm tempted to say mount it in a pair of bearings and then usa a small weight on an arm to counteract the torque

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[patprimmer]

Year. You could even have an accurately calibrated weight and move it out an accurately measured distance on the arm until it exactly counter balances the torque.

Or else you could lean the leaver on a load cell instead of the weight.

Regards

eng-tips, by professional engineers for professional engineers
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[BobM3]

How high a frequency response are you looking for? You might not be able to get there by measuring the reaction of the motor body. Can you strain gage the shaft? Optical methods may work if you have a long enough shaft that would produce a significant strain.

 

[tomdarwinaustralia]

the motor weights approximatly 25kgs, dimensions 15cms*15cms*25cms...
the solution with the arm/calibrated weight or load cell looks interesting but I am not sure it would be accurate enough regarding the accuracy I need !( <0.005 Nm)

I'm looking for 'external' sensor because I'm afraid the shaft can't support any strain gage sensor..

Basically I had this idea: mouting the motor body in a pair of bearing and use piezo sensors (the most accurate I could find) to counteract the torque. But I've still some natural frequency issues and I'm not sure the bearing are stiff enough to allow such precise measurememt !

Anyway,BobM3 would you like to tell me more about the "optical methods" you described ? Do you know any company which could sell such equipment ?

 

[BobM3]

I've only used strain gage devices but find the optical method interesting. This site describes the method:

http://vibrac.com/products/transducers.aspx

Sounds like you can't use an inline unit but maybe you can use the idea and make something that would work for your setup. I've also heard of using gear teeth and prox switches to measure the shaft deflection.

There's also magnetoelastic:

http://www.mdi-sensor.com/

 

[tomdarwinaustralia]

Thanks for these very interesting links but as you said I'd like to avoid the use of inline unit because of the lack of the accessibility of the shaft and these solutions look, as well, a little bit too expensive

" I've also heard of using gear teeth and prox switches to measure the shaft deflection." would you like to explain more precisely ?

Thanks for your contribution to this very helpful and professionnal forum !


Thomas

 

[Warpspeed]

The gear tooth method works reasonably well, but it requires a torsion bar between two slotted optical or magnetic sensors. For good sensitivity and robustness the torsion bar needs to be fairly long. I understand that may be a problem in your case?

This method definitely requires some software processing of the two recovered signals. As the torsion bar twists under load, the load end becomes time delayed with respect to the drive end.

It is a case of measuring the load end delay time, and dividing this into the tooth to tooth time interval. These time delays are both rpm dependent, but the ratio stays the same for a given transmitted torque. Problems with vibration and time jitter require some heavy software averaging of many successive readings to get a smooth high resolution torque reading.

I have a rolling road chassis dynamometer at home that uses this method of torque measurement, and I wrote the software for it, so am acutely aware of the problems and limitations.

I much prefer Greg Locock's idea of a couple of support bearings and a torque arm. Very simple, very easy to calibrate, and highly accurate. This can rest on a weighing scale, or use a load transducer cell. Physical weights mounted on a torque arm can be a problem, because they may tend to thrash around between the mechanical limit stops.

 

[BobM3]

Just seems to me that your accuracy requirement (and therefore, sensitivity requirement) would make it difficult to make the motor housing on bearings idea work. Also you had a problem with a 260 hertz natural frequency. Take a close look at your sensitivity needs versus your stiffness requirements to make sure you can make the rotating housing/torque arm work. You'll probably want to have a damper on the arm. Make sure your first natural frequency is far enough above your maximum frequency of interest.

 

[sreid]

You could build a mechanical support for the motor that has the stiffness/compliance you requir, say, 10 micrometer movement at full torque. Then use high accuracy position sensors to measure the deflection. There are capacitive, eddy current and optical sensors that can measure deflection in this range with a resolution in the 1 to 10 nanometer range.

 

[tomdarwinaustralia]

Thanks for all these valuable inputs ! We finally use a motor housing flexibly mounted and piezo/strain gage sensors to measure the deflection.

Thanks for your contribution to this very helpful and professionnal forum !

Thomas