Hilbert Transform
The impulse response for a 10-term approximation to a Hilbert transformer is shown in Figure 3-46. The odd-symmetry and zero-valued coefficients are both exploited to generate an efficient FPGA realization. The coefficient data file for the Hilbert transform must contain the zero-valued entries. For example, the .coe file corresponding to Figure 3-46 is shown in Figure 3-47.
图3-46显示了对希尔伯特变压器进行10项近似的脉冲响应。 奇数对称和零值系数都被用于生成有效的FPGA实现。 希尔伯特变换的系数数据文件必须包含零值条目。 例如，图3-46对应的.coe文件如图3-47所示。
In practice, some optimization methods used for designing a Hilbert transform can lead to the presence of small even-numbered coefficients. If the Hilbert Transform filter class is used in the FIR Compiler, you must force these terms to zero.
实际上，用于设计希尔伯特变换的一些优化方法可能导致存在小的偶数系数。 如果在FIR编译器中使用希尔伯特变换滤波器类，则必须将这些项强制为零。
Just like the half-band filter, the zero-valued entries for a Hilbert transformer can occur at the filter impulse response extremities. However, these values do not contribute to the result.
就像半带滤波器一样，希尔伯特变换器的零值条目可以出现在滤波器脉冲响应末端。 但是，这些值对结果没有贡献。
This condition is detected when the filter is specified. If the number of taps is such that the zero-valued coefficients form the first and last entry of the impulse response, the filter length is reported as an invalid value. The number of taps N for a Hilbert transformer must obey N=3 + 4n, where n=0,1,2,3,…. For example, a Hilbert transform filter can have 11, 15, 19, and 23 taps, but not 9, 13, 17, or 21 taps.
指定过滤器时会检测到此条件。 如果抽头的数量使得零值系数形成脉冲响应的第一个和最后一个条目，则将滤波器长度报告为无效值。 希尔伯特变压器的抽头数N必须服从N = 3 + 4n，其中n = 0,1,2,3，.... 例如，希尔伯特变换滤波器可以具有11,15,19和23个抽头，但不具有9,13,17或21个抽头。