I was running a test wherein i had a DAQ board accepting inputs from a torque transducer and a RPM sensor. Now earlier the whole setup was working fine except for the torque measurements which were all over the place due to a misaligned shaft. 
Fixed that, and about a couple of weeks later, i had this problem of the rpm readings going haywire anytime the RPM reached above 2100 rpm. 
For weeks we tried to figure out what was going wrong, sent the DAQ board out to be rechecked and all and then one of my colleagues just thought of lifting the box and found out the readings were way better for both and the RPM readings had no noise. 
Apparently mechanical vibrations being transmitted from the motor to the table and then to the box which just screwed with everything. 
Now we have a pillow below the box 
Any inputs?