Torque Measurement

Depends on your definition of "good enough".

The biggest unknown(s) in calculating motor torque will be the losses/inefficiencies through any gearbox/sheave/pulleys. You may be alright within 5%.

The second biggest possible source of error would be if you are calculating the hp of the motor from the voltage and current where you need to account for the motor efficiency. You may want to look into something like this:

texasiof.ces.utexas.edu/texasshowcase/pdfs/presentations/d5/rschiferl.pdf

A torque transducer offers uncertainty down to 0.05% usually around 0.25%.

The best torque measurement via motor current is higher 5% and with the efficiency (friction) of gears and bearings it goes over 10%.

Depends what you measure for.

A torque transducer need not be very expensive. Also, if you are handy, there are various ways of building torque transducers using strain gages and simple instrumentation.

But what can be expensive is the calibration of it, the temperature compensation of off-set and sensitivity and also the correct design in order to avoit cross influences as bending or axial transverse forces.

All depends on required uncertainty of results.

In a perfect world, temperature remains constant during during the test, so thermal coefficients can be ignored. In the real world temperature changes must be accounted for. Engineers have tools to provide temperature compensation, if required. In a perfect world, we can calibrate the torque transducer with a torque wrench. In the real world, that may not be sufficient. So, design and calibration can be expensive, but one of my functions, as an engineer, is to size up problems, and recommend workable solutions. Until I know more about the conditions under which the test is to be performed, I would not rule out the possibility of constructing test instruments.

Thank you welderman, in fact you support what I wrote, I highly appreciate.

You will allow me to make some comments to what you wrote:

"Engineers have tools to provide temperature compensation, if required"; of course but to compensate it one should first know how is its influence, linear, not linear. To know it a temperature run has to be done since theoretically it is, considering all factors influencing the behaviour of a strain gage bridge, impossible to do it. This is the reason why when needed sensors are first exposed to variable temperatures to gain the information. As you know there are, depending on the kind of strain gage used, several parameters variable as function of temperature as off-set, sensitivity and also gage coefficient and impedance. Without a temperature run to compensate the effects is not possible In some cases even with the information it is difficult to do it.

You interpreted what I wrote the wrong way I never thought to avoid DIY sensors but as I wrote it depends on the required uncertainty. What I wanted to stress is that one should not suggest to make a DIY sensor without knowing more about the application and its needs.

Since torque wrenches present the highest uncertainty due too many factors even if one does not need a too high precision it will never the less be recommendable to calibrate by comparison with an other torque sensor the last being calibrated with a dead weight rig.