QEX

rejection to 40 dR, so we could detune by 20%. With the 90° phase shifters previously described, one could only detune by 1.3% for the same degradation. Detuning from the design frequency with the quadrature hybrid does have one drawback, in that the circuit then becomes somewhat sensitive to LO level.

A Truly Broadband Circuit

By combining the quadrature coupler just described with a pair of limiter amplifiers, one can make a circuit that is no-tune over several octaves of frequency. The limiter amplifiers are placed between the output ports of the quadrature coupler and the LO ports of the DBMs. The limiter amplifiers convert the LO-drive signals to constant-amplitude square waves. Now only the phase errors with frequency change are important. These small errors result from losses in the inductor and capacitor(s). We need experiments with real hardware to see how well we can do in practice.

LO drive can be developed digitally, with an oscillator at four times the desired output frequency followed by a divide-by-four quadrature circuit, Such a digital drive circuit has the potential for superb performance over a very wide range of frequency.

Even with a truly broadband LO-drive circuit, a receiver or exciter will still need a low-pass filter in the RF path to suppress the unavoidable third-harmonic response of the mixer. For this reason, I think that a no-tune design over one octave would be sufficient in practice and readily realizable.

Phase Shifting in the RF Path

So far, our circuits have followed conventional wisdom. We've placed the RF phase shifter or quadrature hybrid in the path from the LO to the DBMs' LO ports and used a 0° or 180° hybrid on the RF ports.

We can also connect an RF phase shifter or quadrature hybrid in the path to (or from) the RF ports of the DBM, with the LO ports fed by a 0° or 180° hybrid. This configuration is advantageous because the RF port of a DBM is much better behaved than its LO port. The RF port is a time-vary-ing-linear circuit, while the LO port is a nonlinear circuit. The performance of the previously described phase shifters connected to an RF port is much closer to their performance driving a 50 Q load. I recommend the transmission-line phase shifter, the pi-network

0 0

Post a comment