Soft dB Zen-Ten User guide User Manual
Page 24
User manual for the ZEN 3-channel X-LMS controller
Soft dB inc.
The main function of the band-pass filter is to force the control to work only on a selected
frequency band. The band-pass filter also acts like an anti-aliasing filter: high-frequency
sampling (78.125 kHz) along with the decimation technique, allows a very short group
delay but the aliasing can be a problem without the addition of a numerical anti-aliasing
filter. The red cursor in the upper graph of the design interface indicates the anti-aliasing
filter frequency. The band-pass filter must be designed to ensure at least 30 dB of
attenuation at the red cursor.
The low and high frequencies of the band-pass filter can be set along with the filter type
and filter order. For certain types of filters (Chebyshev and Elliptic), the Ripple (dB)
parameter allows the user to specify the level of ripple in the bandwidth of the filter.
On the DSP, this band-pass filter is implemented using an FIR technique. The user has to
specify the number of coefficients for the FIR filter running on the DSP (control Nbr
coeff). The graph at the bottom of the design interface (see Figure 14) helps in
determining whether the FIR length is adequate. The FIR filter is long enough if its
amplitude at the end of the impulsion response is close to zero. Also, the frequency
response of the filter on the upper graph is computed using the impulse response for both
the FIR and IIR versions of the filter. In this way, the truncation effect can be easily
analyzed and the number of coefficients for the FIR filter can be adjusted as-required.
Note: The design of the band-pass filter must be done in such a way as to minimize
the length of the filter. The length of the impulse response has an important impact
on the distance between the reference sensor and the control plan. To obtain a
compact control system in which the reference sensor is close to the control plan (or
the control sources), the impulse response must be short.
Primary control path parameters (filters CxRx)
The parameters for the primary path filters are: the number of coefficients of the filters
CxRx, the adaptative step size of the identification LMS and the amplitude of the white
noise output used in the identification process.
The length of the CxRx filters is not critical. These filters are used during the control
phase to subtract the contribution of the controller outputs from the reference signals. If
the subtraction is not very precise, control performance will not be greatly impacted.
Also, if the primary control path amplitude is not significant, the user can remove the
filter and disable the subtraction. The best way to disable the CxRx filters is to use only
two coefficients, a white noise amplitude of zero and Mu=0.
ZEN User Manual
p. 24