Sure, beam deflection, and other "slop" in the setup does induce
error. As you suggest, I'm intending to minimize that error by having
it deflect as little as possible - both mimimizing length (I only need
6-10 inches, really, to get to the torque ranges I'm after) and making
sure it's as stiff as possible.

That said, even if the setup were to deflect, say, 15deg, cosine error
induced would be about 3.5%. Overall effect on the system is
dependent on the relative length of the extension compared to the
wrench. If I were to really worry about it, it's pretty a pretty
simple trig problem to figure out exactly how much cosine error I have
(by measuring actual beam deflection) and factor that back into my
torque wrench click setting. Also consider that my *calibrated*
wrench is only certified to 4% accuracy anyway. All said, it's well
within the range of accuracy I need where the torque specs required
are on the order of +/- 15% - it's just not that critical of an
application.

As to verifying torque, one can get a good check by clamping the setup
in a solid vise and applying known weights to the extension/wrench
combo. Measure the distance and you know if you're in the ballpark or
not.


"Wild Bill" wrote in message
I think the example in the spooky-torque design is probably considerably
inaccurate for a couple of reasons.

Beam deflection, like the principle of a beam-type torque wrench. Depending
upon the choice of material, the amount of deflection could change with use.
As the deflection changes (from a straight line), so does the distance to
the center of the far end.
The breaker bar swivel pin and yoke deflection.

Maybe the best chance at achieving a predictable multiplier would be a
single piece of very strong material (no fasteners or joints), and the
centerline kept perfectly straight, while keeping the axis of the torque
wrench drive at precisely 180 degrees to the drive socket.

I don't think an accurate torque reading/indication can be achieved with a
breaker bar and/or the extension shown in the example.
There wouldn't be an easy way to test the accuracy of a homemade extension
without having it checked against a calibrated standard.