sign in sign up
<<
Home , Frequency modulation

High Design FREQUENCY

Implementing Functions in Microwave Frequency Synthesizers

By Alexander Chenakin Phase Matrix, Inc.

he main function This article reviews the of a frequency syn- typical methods for imple- Tthesizer is to deliv- menting modulation within er a stable and clean sig- a frequency nal. However, many instrument or system applications require not only a fixed frequency signal but also various modulation functions ranging from simple pulse, amplitude, fre- quency and to complex dig- Figure 1 · Pulse modulation is implemented ital modulation formats. Although modula- by inserting a switch into the synthesizer out- tors are usually realized as external devices, put path. they can also be incorporated into a frequen- cy synthesizer core. Inside any synthesizer there are many circuits that can carry multi- quency (also called rate) can be between DC ple functions and be reused to increase the and several megahertz. Pulse modulation is functionality without a significant increase in practically implemented by inserting a switch cost. This results in a more cost efficient and (or a chain of switches for a higher on/off ratio) versatile design. The most commonly used into the synthesizer output path as depicted in modulation schemes are briefly reviewed in Figure 1. The switch can be built using PIN- this article. Further details on modulation theory and implementation techniques can be found in [1-7].

Pulse Modulation Pulse modulation is probably the simplest modulation form. It is achieved by switching the output signal on and off in accordance with the applied modulating pulses. The result is a sequence of RF pulses that replicate (or tend to replicate) the input modulating signal. The minimum RF pulse width, rise time, fall time and overshoot are important characteristics that define how well the modulating signal is replicated. Typical rise time and fall time num- bers required are in the order of 10 nanosec- onds. Pulse modulation on/off ratio is another critical parameter. A typical specification is Figure 2 · A pulse modulator is integrated 80 dB or higher. The modulating signal fre- into a switched filter bank.

20 Electronics High Frequency Design FREQUENCY SYNTHESIZERS diodes or FET devices that support modulation index or depth) is nanosecond switching. A high-pass achieved by setting the output filter follows the switch to sup- power level in the middle of its press the leakage of the modulat- control range. Another important ing signal (called video feed- requirement is linearity because through) to the synthesizer output. the modulator must translate the Alternatively, the pulse modu- modulating signal with minimal lator can be conveniently com- . This may further limit a Figure 3 · is real- bined with a switched filter bank realizable modulation depth. ized using a voltage-controlled attenuator. used for harmonic (or subharmon- Various linearization techniques ic) rejection. The idea is to reduce can be applied to minimize AM sig- the design complexity and cost by nal distortion. In some cases, it is utilizing the same devices for desirable to implement not a lin- both functions. Furthermore, no ear but a logarithmic modulating additional loss is introduced that scale, meaning that the output eases requirements for the output power changes in dB per volt. This power . In this case, the mode is utilized for large power pulse modulator incorporates a variations (e.g., for simulation of digital decoder (as shown in Fig. rotating patterns) and is 2) to control the switches in such called deep AM. a manner that they will provide Alternatively, amplitude mod- the highest possible isolation ulation can be implemented by (on/off ratio) for any given fre- summing the modulating signal quency subband. into the ALC (automatic level con- trol) loop as shown in Figure 5. In Amplitude Modulation general, the ALC-based ampli- Figure 4 · Amplitude modulation is com- Amplitude modulation (AM) tude modulation offers better lin- bined with an open-loop amplitude con- historically has been one of the earity and repeatability charac- trol. most popular methods for carrying teristics. However, the modulation via RF . It depth may be limited by the avail- is realized by varying the output able ALC dynamic range, which, signal amplitude in accordance in turn, depends on the utilized with an applied modulating signal. . The maximum modulat- The simplest way to implement ing signal rate is also lower com- AM is to control the insertion loss pared to the open-loop alternative of an attenuator inserted into the because of the settling time of the synthesizer output circuit as closed-loop ALC system. depicted in Figure 3. This can be naturally combined with an open- Frequency and Phase loop amplitude control as depicted Modulation in Figure 4. The synthesizer is first Frequency modulation (FM) is commanded to set a desired output another popular form of analog power level by programming DAC modulation that offers better sig- (digital-to-analog converter) volt- nal immunity compared to AM. age. Then a modulating voltage is The process of producing a fre- applied over the DAC voltage to quency-modulated signal involves vary the output signal around its the variation of the synthesizer nominal value. Naturally, the out- output frequency in accordance put power cannot be set at its with the modulating signal. The highest (or lowest) level because frequency where the certain headroom is needed to synthesized signal fluctuates is allow further power changes. The proportional to the peak ampli- Figure 5 · Amplitude modulation can be maximum power variation (which tude of the modulating signal and realized by summing the modulating signal can also be expressed in terms of is called frequency deviation. FM into the ALC loop.

22 High Frequency Electronics High Frequency Design FREQUENCY SYNTHESIZERS

the overage output frequency remains correct. For proper opera- tion, the modulating signal rate has to be well above the loop filter bandwidth. Thus, the PLL filter bandwidth is adjusted (narrowed Figure 6 · Frequency and phase down) to allow lower modulating modulation. Figure 9 · A phase shifter provides a rates. As a result, the phase phase modulation function. usually increases when FM is enabled. Typical achievable mod- ulating rates range from a few kilohertz to tens of megahertz. What if we need to apply a lower-frequency modulating sig- nal? Obviously, we have to fur- ther decrease the loop filter bandwidth, which may not Figure 7 · Frequency modulation is always be possible because of realized by modulating the VCO tun- prohibitory high VCO free-run- ing voltage. ning phase noise at low frequen- cy offsets. An alternative solution is to modulate not the VCO but the reference oscillator as shown Figure 10 · Frequency modulation is in Figure 8. If the modulating realized by controlling the DDS output signal rate is sufficiently low, the frequency. PLL will track the reference fre- quency change and, hence, trans- late the modulation to the VCO Figure 8 · A modulating signal is of the phase, it is possible to convert output. This mode is often called nar- applied to the reference oscillator. FM to PM (and vise versa) by adding rowband FM because the modulating an integrator (or differentiator) cir- frequency must be within PLL filter cuit into the modulating signal path. bandwidth. The loop filter bandwidth can also be described by modulation How can we modulate the synthe- should be set as wide as possible to index, which is the ratio of the maxi- sizer output frequency? From first allow higher modulating rates. mum frequency deviation to the fre- glance, it is quite straightforward—we Typical rates start from nearly DC to quency of the modulating signal. can simply modulate (i.e., change) the a few tens of kilohertz. Thus, the nar- Note that we can vary not only the VCO (voltage-controlled oscillator) rowband mode complements its wide- frequency but also the phase of the tuning voltage around the value where band counterpart to extend the over- synthesized signal, thus producing it is settled. The problem, however, is all modulating frequency range. phase modulation (PM). Both process- that the synthesizer’s PLL (phase- A disadvantage of this technique es are quite similar because in both lock-loop) core will tend to correct any is low achievable deviation caused by cases we vary the argument (the voltage change we introduce. Most very low tuning sensitivity of the ref- angle) of the same sine function as likely, we will lose this battle unless we erence oscillator. Although the refer- illustrated in Figure 6. Hence, the change the tuning voltage so fast that ence frequency deviation is multiplied angular modulation is a more general the PLL will not be able to react to the up by the PLL at the 20logN rate, the case that represents both FM and PM. change. This is exactly the idea that synthesizer output deviation may be The difference is not in the output sig- stands behind a so-called wideband insufficient. A higher deviation can be nal waveform but rather in the modu- FM modulation mode. The FM modu- achieved by changing not the refer- lator circuit configuration; that is, lator is built by adding a circuit (e.g., ence frequency but rather its phase as what parameter (frequency or phase) an operational amplifier) that sums an depicted in Figure 9. This represents is directly proportional to the ampli- external modulating signal with the a classical phase modulation; howev- tude of the modulating signal. Because control voltage delivered by PLL as er, both forms are interchangeable. the instantaneous angular frequency depicted in Figure 7. The PLL remains Practical implementation requires a is mathematically the time derivative locked all the time, thus ensuring that phase shifter that can be purchased

24 High Frequency Electronics High Frequency Design FREQUENCY SYNTHESIZERS

that this calibration has to be imple- mented at many frequencies across the entire operating range. Moreover, the calibration has to survive over time and temperature changes. Thus, achieving a good image and LO leak- age suppression for a broadband, high- frequency, direct IQ-modulator is a very challenging task. An alternative solution is to create Figure 12 · IQ-modulator block a desired IQ-modulated signal at a Figure 11 · A signal is presented as diagram. lower, fixed frequency and then a vector on a polar diagram. upconvert it to microwave frequencies as illustrated in Figure 13. Obviously, Hence, such a complex modulation is it is much easier to achieve better or can be built using discrete devices called vector or IQ-modulation. cancellation of undesired products at such as varactor diodes. Vector modulation can be applied a single (and lower frequency) point. An interesting solution is to con- directly at RF frequencies by utilizing However, the difficulty now moves to trol the division ratio of a frequency an IQ-modulator. It consists of two the upconversion side. We still need to divider inserted into either the PLL identical mixers driven with a 90- remove the undesired and reference or feedback path as illus- degree phase shift at their LO (local LO leakage posted by the second (reg- trated in Figure 10. The divider has oscillator) ports as shown in Figure 12. ular) mixer. However, because the to be a high-resolution device such as The base-band data signals are product separation is much larger a fractional-N divider or DDS (direct applied directly to the mixer IF (inter- (compared to the direct IQ-modula- digital synthesizer). The modulating mediate frequency) ports, upconverted, tion), a hardware filter can be used. signal is first digitized by an ADC and summed together with no phase For a broadband operation, a YIG- (analog-to-digital converter) and then shift between them. The resulting out- tuned filter is a simple and effective is summed with the DDS tuning put is an IQ-modulated signal at the solution. The disadvantage of the YIG word to vary the DDS’s output fre- same carrier frequency as the LO. The filter is slow tuning speed and rela- quency. Because the DDS offers quality of the synthesized signal can tively narrow pass-band that can be exceptionally small frequency incre- be tested by applying two base-band insufficient in certain applications. A ments and a fast update rate, a sim- signals of the same frequency and switched filter bank (Fig. 14) offers ple yet high-performance FM (or PM) amplitude with a 90-degree phase better characteristics. However, it modulator can be constructed. shift with respect to each other. For a requires a larger number of channels perfect modulator, only one sideband (compared to devices used for har- Complex Modulation should be present. However, in reality, monic and sub-harmonic filtering) More effective modulation formats the output signal contains another and, hence, is hardware extensive. are possible by simultaneously vary- sideband because of imperfect ampli- This results in a more complex sys- ing both amplitude and phase. The tude and phase balance. For example, tem design and posts other challenges simplest way to visualize such a com- equalizing the signal paths within 1 (e.g., achieving high isolation between plex signal is to draw it as a vector on dB (amplitude) and 10 degrees (phase) filter channels, etc.). a polar diagram. The amplitude and results in approximately a 20-dB side- In the past, complex microwave phase are represented as the length band rejection. Naturally, better rejec- assemblies were often built using and the angle of the vector as shown tion is required. Moreover, an LO leak- individual connectorized modules in Figure 11. In digital communication age also takes place. The undesired connected with coaxial cables. The systems, such a signal is expressed in sideband can be further suppressed by designer could easily isolate and I (in-phase) and Q (quadrature) terms, adjusting the amplitude and phase of refine individual blocks to make them which are projections of the signal vec- the applied IQ signals. The LO leakage perfect. These days, such assemblies tor on a corresponding orthogonal can be controlled by adjusting DC off- have to be made on a common PCB axis. Therefore, the amplitude and set voltages for the diodes used in the using tiny surface-mount parts. A phase modulation assumes the change balanced mixers. Therefore, it is gener- great effort is required to minimize of the signal vector, which can be con- ally possible to calibrate the modulator interactions between individual com- veniently accomplished by varying characteristics to a degree where it can ponents sitting on the same board. two independent IQ-components. be practically utilized. The difficulty is Furthermore, many parts are reused

26 High Frequency Electronics Figure 13 · Upconversion using a tunable filter.

Figure 14 · A broadband upconverter based on a switched filter bank. to accomplish different functions, which are distributed through the whole assembly.The net result is a significant increase in “design density,” meaning both component count and functionality per square inch. All these factors drastically complicate the design process. Nevertheless, this seems to be a “must” approach these days.

References 1. Manassewitsch, V., Frequency Synthesizers: Theory and Design, 3rd ed., NJ: Wiley, 2005. 2. Gardner, F. M., Phaselock Techniques, 3rd ed., NJ: Wiley, 2005. 3. Crawford, J. A., Advanced Phase-Lock Techniques, MA: Artech House, 2008. 4. Coombs, C. F., Jr., (ed.), Electronic Instrument Handbook, 3rd ed., New York: McGraw-Hill, 1999. 5. Terman, F. E., Electronic and Engineering, New York: McGraw-Hill, 1955. 6. Gentile, K., “Fundamentals of Digital Quadrature Modulation,” RF Design, February 2003, pp. 40-47. 7. Chenakin, A., Frequency Synthesizers: Concept to Product, Norwood, MA: Artech House, 2010.

Author Information Dr. Alexander Chenakin is the vice president of the Signal Sources Group at Phase Matrix, Inc., where he over- sees the development of advanced products for test-and-measurement applications. His pro- fessional achievements have been widely presented in trade publications and international conferences. Dr. Chenakin can be reached by phone at 408-954-6409 or by e-mail at achenakin@ phasematrix.com.