Sound and Vibration Measurement: First Demo of the test jig

Hello Jan,

You are hitting a problem I recall well from 48 years ago when I first tried to measure microphone frequency responses using a Bruel & Kjoer setup. There was a little speaker in a small enclosure with a reference microphone (known as an 'Artificial Mouth'). The sweep oscillator was mechanically controlled by the pen plotter (which used rolls of paper and real ink pens). The sweep oscillator had a compressor loop which took input from the reference microphone via a built in mic amp and a broadband rms responding rectifier ciruit. . The compressor loop would increase/decrease the drive to the speaker to keep the sound level constant. At low frequencies it would increase the drive level to the speaker until either the speaker or amplifier overloaded enough to make sufficient high freqency content to reach the correct soundlevel. if you set the system up to demand output that the speaker could not manage all frequency responses plotted looked suspiciously good !

Eventually I built a fully analogue system with tracking filters to plot microphone and accelerometer responses.

In your video I can hear lots of sound at an indicated 22Hz test frequency but it isn't 22Hz but much higher harmonics. The speaker or amplifier are overloading.

It would be very instructive to look at the Fourier transform (or even scope like amplitude/time response)  of the captured data (I think Labview can do this.)

MK

Michael, yes. I'm going to run into all possible beginners problems with my test jig. 
It was never my intention to build a good jig. I wanted to have a handle to hang my DAQ automation story on.

In the setup in this video, a fibre cap is sitting loose on the speaker. The piezo sensor is sitting loose on top of that one, hanging from a desk lamp by the cable .
Nothing is bolted down, glued, attached. So many options for rattling, and this setup seems to deploy all those options to the max.

I will be happy if I can use the DAQ to show how imperfect this jig is. Get data points and correlate them to the reality of the device at hand.
my goals are:

• The custom instrument software on the Pi should be industry strength
• Same for the LabVIEW driver
• LabVIEW examples should prove that the DAQ is up to the job in a real automated process.
  They should show that the DAQ is fully controlable, and that the data can be processed (waveform viewed, stored to file, analytics like FFT, RMS, AVG)

for the jig, I will settle for:

• it can shake, with programmable frequency and amplitude.
• the sensor can be mounted to it.
• it can carry a small design (pcb, element14 puppet) that can be analysed

I'm as good as finished with the first part. The design I have allows for a streamlined process to set up the device and stream acquired data to the test computer.
I've been using 4 strategies at the moment to consume the data: waveform, spreadsheet, averaging, RMS. 
It's not difficult for the test writer to consume the data: the process exposes a loop where each iteration gets fed by a single sample. Designer can chose to eat those samples one-by-one (what I did for the waveform and spreadsheet) or collect them for analytics (what I do for RMS: that collects 30000 samples then calculates - and AVG: that collects 64 samples and averages over theat set)

That said, I will try to make a good test jig. I'm visiting balearicdynamics in the Drongen maker space this Saturday. We're going to try and make a stable platform.

michaelkellett, I now have a better built device.

Here's the FFT when sending a 100 Hz sinus to the jig:

30 Hz signal when there's no obvious vibration:

With signal, same settings, with something loose so that it can ramble:

That's showing the distortion at very roughly 3% (on the good curve).

Not bad for an open loop system  - do you know the mechanical resonant frequency of the set up ?

MK

going by using my hands to feel the resonance, I'd say the cabinet resonates around 50 Hz, 100 for the plate we mounted on top of it with the sensor attached to it.

going by using my hands to feel the resonance, I'd say the cabinet resonates around 50 Hz, 100 for the plate we mounted on top of it with the sensor attached to it.