- What is an SRS and how is it calculated?
- What do the different status panel colors indicate in Shock Synthesis?
- What is the Recalc Comp feature and how does it work?
The Shock Response Spectrum (SRS) is a composite response function which is composed of the maximum amplitude responses from a "bank" (i.e. a cascaded set) of specifically defined passband filters. These filters are usually defined based on 1/n octave center frequencies. The time domain signal is independently passed through each filter and the maximum time domain response noted. The resulting set of response amplitudes are plotted against the corresponding filter center frequency which provides the basis of the SRS plot. The definition of these passband filters essentially tries to "model" a single degree of freedom system. As such, the damping and filter center frequency must be provided to complete the filter definition. For data acquisition purposes, since the filters have a bandwidth associated with their response, data acquisition operations must provide sufficient frequency (passband) data to allow proper response of the filter. This requirement generally means that the acquisition bandwidth must exceed the highest filter center frequency by a sufficient margin to allow a proper response of the filter to be obtained. For example, if the highest filter is 1000 hertz, the acquisition should allow data with frequency content to at least 2000 hertz (actual upper limit depends on the 1/n octave resolution) to be sure the 1000 hertz SRS filter bandwidth is satisfied. In addition, since the SRS is based on a set of maximum time domain (SRS filtered) responses, the data is generally "over-sampled" to provide better time domain resolution of the filter responses. The effect of over-sampling is to lower the available upper frequency content given the same sample rate. For example, data sampled at a rate of 5120 hertz is low pass filtered by the acquisition system at 2000 hertz (standard 2.56 : 1 over-sample ratio). SRS calculations generally employ at least a 5.12 : 1 ratio which means the maximum allowable upper frequency will be reduced to 1000 hertz. For this example, one could try to calculate the 2000 hertz SRS data value, but the result would most likely be under estimated and should not be relied on for a good measurement result.
A simplified voltage "tracking" system is provided by observing the channel status view panel. The colors of the various channel values will change based on the current channel voltage level being processed.
The following table allows the user to identify the channel voltage observed relative to the full scale (FS) channel range:
|Grey||Channel is inactive or not available.|
|Yellow||Channel voltage is less than 5% of the FS value.|
|Green||Channel voltage is less than 99% of the FS value.|
|Red||Channel voltage is greater than or equal to 99% of the FS value.|
The Recalc Comp feature allows the user to dynamically update the system transfer function while performing output pulses. The system transfer function is used to compensate the drive signal to accommodate shaker system characteristics.
There are several parameters which are associated with the compensation function. The following descriptions define their use as employed during the system transfer function determination (System ID) and updates (Recalc Comp). The parameters are found in the Schedules... dialog under the main menu.
The system transfer function is initially determined during the System ID phase of the program. This is generally considered part of the low level equalization of the program. The "Equalization Level" is defined as that output level where the averaging will occur to calculate the initial system transfer function (H(f)). The level is estimated by using the peak value of the output waveforms and setting the output drive amplitude accordingly. The reference level is determine by the maximum amplitude of the reference waveform.
The type of excitation to be used for the initial H(f) determination is selected under "Initial Excitation". The selections are random, pseudo (random) or pulse. The random selection generates a unique randomized waveform (approximately 3 sigma distribution) for each output sequence. The pseudo random selection utilizes the same random sequence as the first frame; just repeated for each additional output frames. The pulse option utilizes the reference waveform as the excitation source. The number of averages to be used during the system ID phase is determined by the "Average Weighting" value. The average count can be determined by an inverse of this field (e.g. 0.125 uses 8 averages). In general, the best H(f) calculation is found by using the most broad band excitation source. This source is usually random. Pulse excitation should usually be avoided since pulses rarely contain sufficient broadband energy to adequately excite all system H(f) frequencies at a proper level.
Updating of the system transfer function is performed during the test by use of the "Recalc Comp" feature. As long as this button is enabled, the program will attempt to update the system transfer function based on the following parameters and procedure. As mentioned earlier, all updates are performed on a spectral line basis.
After a pulse has been output, the system calculates an averaged error function which is based on the Control response and drive signal of the pulse. A correction to the system H(f) function is performed on a spectral line basis. The averaged error function utilizes an exponentially weighted averaging scheme for each pulse processed at the same level. Changing the output level will reset the averaging process. Averaging is required in order to determine the coherent power between the drive and control channel. The exponential weighting factor is the same value as used in the "Average Weighting" parameter described above.
The "Feedback Gain" is used to determine what weighting feedback is to be used on the error function when applying the correction to the system transfer function. For example, a weight factor of 0.75 indicates that the updated H(f) will contain 25% previous H(f) and 75% corrected H(f) weighting from the averaged error function.
The actual update of the H(f) function is further controlled by the 2 threshold limits of "Coherence Blanking" and "Compensation THold". The coherence blanking indicates the level of the coherence (between drive and control) which must be satisfied before any change to the H(f) is performed. This value has a range of 0 (No coherence) to 1.0 (total coherence).
Similarly, the Compensation Threshold is the level below which no H(f) updates will be performed. This level applies to BOTH the reference and system H(f) functions. For example, if the threshold is set to -100 dB, then no update of the H(f) line will occur if the reference power spectrum line is below 100 dB from the peak reference line. A similar operative description of the system H(f) function also applies.