Swept Sine Test
Whereas a Random test generates many frequencies over the band of interest at once, a swept sine test generates only one frequency, and sweeps this frequency through a pre-set range. Feedback from the control signal is then used to adjust the output amplitude such that the response amplitude of the UUT matches a test profile. The test profile is a graph of amplitude (usually defined as peak acceleration) versus frequency.
RSD, or Resonant Search and Dwell, is an extension of the Swept Sine test.
The Sine Control Process
The swept sine control process consists of generating a sine wave output to excite the device under test, detecting the control signal input amplitude, comparing the detected level with the reference amplitude, and updating the drive signal amplitude appropriately.
To measure the level in the incoming control signal, the detector can use a tracking filter, or can measure the RMS, peak, or mean value of the signal. When using a tracking filter, amplitude and phase data are produced while the other measurement methods only produce amplitude data.
If more than one control channel is used, then the output of each detector is combined in the Channel Averaging Block.
Tracking filters greatly reduce the noise and harmonic signals above and below the sine drive frequency. Their center frequency is always tuned to the current drive frequency, allowing all other signals to be rejected from measurement and control. The filter bandwidth can be either fixed or proportional to the current frequency.
The Spider system continually updates the tracking filter coefficients based on the current center frequency and bandwidth. It has a stop band rejection of about –60 dB. The output of the filter is averaged to produce a control amplitude value, which is then used by the comparator to correct the output drive amplitude.
In general, the narrower the bandwidth of the tracking filter, the sharper the resonance that the control system can calculate. As shown in the picture above, the red line that uses 7% bandwidth can show sharper resonance than the green line which uses the 25% bandwidth. However, the bandwidth of the filter also affects the speed of response time of a control system. The system response time is inversely proportional to the filter bandwidth. Therefore, choosing the right bandwidth of the tracking filters is usually a trial process.
The Peak, Mean, and RMS measurement methods analyze the data in blocks, with a length determined by the bandwidth settings:
Block length (second) = 1/bandwidth (Hz)
Both fixed and proportional bandwidths can be used. The block duration is constant for fixed bandwidth, and changes with the drive frequency when using proportional bandwidth.
If N is the length of this block, in samples, then the RMS is defined as
And the Mean is the statistical Mean Absolute Deviation:
With RMS, Mean, and Peak measurement strategy, signals at harmonics of the drive frequency can contribute to the final overall measurement result. Therefore, the drive level may be lower than when a tracking filter is used. In other words, when tracking filters are turned on, the UUT might get over-tested; when RMS, Mean and especially Peak are used as measurement strategy, the UUT might get under-tested.