Tektronix, Inc.

Ahmed M Said

Michael C Seckora

A finite impulse response (FIR) digital filter (100, 170, 175, 180) incorporates both symmetric and transposed structures in a configuration that suites ASIC implementation. The filter coefficients can be arranged in ascending order to implement coefficient block floating point, or in descending order. To further facilitate coefficient block floating point, filter coefficients are chosen so that each cell has an exponent value equal to or greater than that of a preceding cell. A right shift is provided between adjacent cells to equate the exponent value of a sum output of a preceding cell with that of the succeeding cell. Variable decimation is provided by downloading appropriate filter coefficient and adjusting a decimation forward clock. Multiple channels can also be accommodated.

Symmetric transposed FIR digital filter

A method and apparatus are disclosed, that is suitable for digital or analog spectrum analyzers, for accurately and rapidly ascertaining the frequency of a spectral line by determining its location from the response of two Gaussian shaped filters whose center frequencies bracket the frequency of the spectral line. The difference is taken between the amplitudes in decibels of the responses of the two Gaussian filters to the spectral line input signal. The frequency of the spectral line is then found from the linear relationship fx=delta-log-ampl.*c1+c2, where c1 is proportional to the square of the standard deviation of the Gaussian filters and inversely proportional to the difference between the center frequencies, f1 and f2, of the Gaussian filters times the logarithm of e, and where c2 is the midpoint between the center frequencies, f1 and f2, of the Gaussian filters, G1 and G2. Alternatively, a sweeping local oscillator output can be mixed with the signal containing the spectral line of unknown frequency and the resulting signal applied to one Gaussian filter at two different times to produce equivalent results. In this case, c1 is proportional to the square of the standard deviation of the Gaussian filter and inversely proportional to the difference between the local oscillator frequencies, f,LO+1 and f,LO+2, at times t1 and t2, times the logarithm of e, and c2 is the average of the local oscillator frequencies, f,LO+1 and f,LO+2, plus f, the center frequency of the Gaussian filter.

Method and apparatus for measuring the frequency of a spectral line

Methods and circuits for implementing them determine the phase angle of a complex electrical signal from the in-phase and quadrature components of the signal by truncating the magnitudes of the real and imaginary components of the signal and combining them to form an address which is then used to access a lookup table containing angular values. However, if both of the values are small, before they are truncated they are both expressed in terms of a lowest common exponent, i.e., shifted left as much as possible without affecting their ratio. The phase angle from -180 to +180 degrees may be found directly from the lookup table, or, in an alternative variation on the method, a single-quadrant angle is looked up in a shorter table and then supplemented according to quadrant information contained in the signs of the values to produce the total phase angle. Circuits are disclosed for implementing both methods for inputs in 2's complement or sign and magnitude form.

Growing community of inventors

Beaverton, OR, United States of America

Ahmed M Said