I'm trying to perform FFT analysis on data from an engine "rpm sweep" test to find resonance peaks on the DUT. There are three tri-axis accelerometers along with time, engine speed, engine torque, channels. One accelerometer is placed on the engine "base" and two on the DUT. Data is sampled at 5 kHz and sync'd across all channels.

I'm wondering how to go about setting up the FFT windowing to do this? I'm not very familiar with vibration. I have tried setting up the FFT with a one window across all the data, but the result was a lot of low magnitude data with no discernible resonance peaks. Is this approach correct, or am I setting up the FFT window incorrectly? I'm using DIAdem for the signal processing.


If you are measuring vibration there is no need for fft to find resonance in the first place. If the accelerometer has a resonance peak the output will increase. This should be visible in the accelerometer raw data.

If you find this, you can map it to the sweep (you know at which time the peak occurs, and you know how your sweep is setup) and therefore calculate at which frequency the peak occurs.

Most accelerometers have output filters, so it can be that you wont see any resonance.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.