I have been experimenting with an ADXL377 accelerometer and have a question as to the validity of the data I am seeing. All three outputs have a 10 nF capacitor making the bandwidth about 500 Hz. The ADXL377 unit is mounted to a test sample attached to a metal block which is struck by a swing arm impact hammer. I am measuring each output at a rate of 400 kHz. The impact event appears to last for about 10 msec. During that time the calculated acceleration oscillates positive and negative which I am interpreting as the sample effectively "vibrating" back and forth. When I mount the ADXL377 on the test sample the frequency of the vibration is about 400-700 Hz. When I mount the ADXL377 directly to the frame of the tester I measure a higher frequency of about 2500 to 3500 Hz which is imposed on the same 400-700 Hz signal. This higher value exceeds the bandwidth of the unfiltered output in the datasheet (Table 1 on page 3). I am not sure how to interpret these results with a "500 Hz" bandwidth filter on the outputs. Seeing the frequency change depending upon what the device is mounted says the output is "real" but measuring a frequency response beyond the filtered output bandwidth of 500 Hz seems suspect. I am not an EE so I am not sure of how to interpret what I see in these data and if it is real or even realistic. So my basic question is this: could these results be real, or am I trying to use the ADXL377 well beyond its intrinsic capabilities?
Here si the output when the ADXL377 is mounted on the impact tester block:
And with the ADXL377 mounted on the test sample (which is mounted to the tester block):
The goal of the test is to quantify the acceleration experienced by the test sample, namely the peak acceleration: X+Y+Z.