This may help to answer your issue

https://www.erbessd-instrument...torsional-vibration/

thanks!

so, I would measure torsional vibrations by placing the accelerometer in an eccentric position...

My understanding is that two tangential accelerometers, 180-degrees apart, are needed to cancel (by subtraction) the lateral motion from pure torsional motion. I would be more convinced with the Erbessed-Instrument "experiment" if another direct measure of torsional vibration was used for comparison. Perhaps for that experiment the torsional motion was dominant compared to lateral motion. I use the optical encoder method (located for high torsional displacement/velocity) and the Binsfeld strain-telemetry (located for high torsional strain).

A triaxial accelerometer would presumably be oriented for axial, radial and tangential (torsional) vibrations. A rotating coordinate system is expected to be different from the stationary X-Y-Z coordinates on the bearing housing. I know it is possible to balance a large slow speed fan using torsional vibrations! If you want to add any sensor mass that is eccentric to shaft axis, then you should balance it with an equal and opposite mass to maintain the integrity of the vibration measurement.

Walt

interesting Walt, for sure another mass 180 degrees apart is necessary to cancel the unbalance of the sensor

Then you answered one of my concerns. Setting up the sensor in the rotary part perimeter, tangential, axial and radial for the triaxial sensor, it is possible to still read an amplitude and phase value and perform a balancing. I will take in mind and try to do if I have the opportunity

i thought, beeing the unbalance 1x synchronous component and with a constant rotating speed (no tangential acceleration)the sensor wouldn't read anything in 1x harmonic.

Regards