Left: FR-4 PCB output waveform with 100 MHz input, via Tektronix 7S11 sampler. Photo courtesy of Dennis TillmanWhat's awkward about building comb generators and other broadband multipliers is that they're traditionally based on exotic components such as step-recovery diodes (SRDs) or nonlinear transmission-line devices (NLTLs). SRDs have previously seen application in the Amateur microwave community.note 4 They're difficult to source as new components, though, since their exotic nature and limited commercial applications have kept them out of the catalogs of the major retail parts distributors. Advanced Semiconductor, the only SRD vendor I've been able to work with directly, has a US $250 minimum order for their ASRD 800-series parts. So when I needed a comb generator for a project recently, it made sense to look for alternatives. Circuit NotesAny active or passive device with nonlinear gain characteristics will generate harmonic distortion in some form. My design uses a series of four cascaded active devices, each of which can provide gain from DC through several GHz. Harmonic generation occurs as a consequence of deliberately overdriving the active devices.The first two stages consist of MC100EL16D differential line receiver ICs from ON Semiconductor. These chips are designed to condition low-level analog signals for introduction into a fast ECL logic environment. As such, they exhibit significant gain, not unlike a comparator without hysteresis. (In fact, self-oscillation in the absence of an input signal is normal.) The input impedance at U1 is 50 ohms, as is the nominal output impedance at U2 with the chosen values of R7 and R8. (See notes 7 and 8 below for links discussing impedance-matching issues with ECL.)With a rated toggle frequency of 1.75 GHz and output edge speeds in the 100-300 ps range, the ECL line receivers are fast enough to serve as useful harmonic sources by themselves. Their differential input stages are great at isolating the output waveform's characteristics from those of the input signal, a very desirable trait in a comb generator. However, the output harmonics fall off more rapidly than desired, and the even-order harmonics are somewhat underrepresented. Right: Output signal from two cascaded MC100EL16 ECL line receivers, driven at 33 MHzThe circuit presented here achieves improved performance in both respects by following the ECL line receivers with two inexpensive MMIC amplifiers. The monolithic amplifiers are undercoupled with 1-pF capacitors in an attempt to equalize lower- and higher-frequency harmonic generation. My device of choice was the Mini-Circuits GALI-5+, which exhibits about 16 dB of gain at frequencies up to 4 GHz and is still usable at twice that frequency. By combining these two different device types, we can take advantage of the input-level independence offered by the ECL line receivers and the superior harmonic-generation characteristics of the MMICs. Results and MeasurementsAmong the many applications for a comb generator is as a driver for a harmonic mixer or sampler. These devices, in turn, find wide application in equipment from synthesizers to network analyzers. My own intended use for the comb generator required specs similar to those of the SRD-based generator in the popular HP 8753A/B vector network analyzer, so it proved helpful to compare the performance of my own circuit to that of its counterpart in my HP 8753A during development. Below: 30-MHz harmonics from HP 8753A internal SRD source. Difference between strongest and weakest harmonics is 26 dBOne useful performance metric for a comb generator is the differential amplitude of the strongest comb harmonic and the weakest. Within a given output range, this parameter tends to improve with increasing drive frequency, so measuring it at 30 MHz does a good job at representing worst-case results. The HP 8753A's generator is excellent in this regard; only 26 dB separates the weakest comb tooth (at approximately 1400 MHz) from the strongest, near 750 MHz. Shown here is the output spectrum from the HP 8753A's comb generator at 30 MHz; results at 60 MHz are similar. Under identical test conditions, my ECL/MMIC generator yields good results as well, but not as good as the far-more-complex and -expensive Hewlett-Packard design.Below: 30 MHz harmonics from KE5FX comb generator. Difference between strongest and weakest harmonicsis 37 dB
Microwave Experimenter's Manual: Antennas, Components And Design Download Pdf
2ff7e9595c
コメント