Measurements on CW signals

The amplitude of a CW signal may be measured in many ways, one traditional instrument being an RF millivoltmeter. These used a diode detector and could measure signals in the range (typically) 10 kHz to 1 GHz. They typically had a high input impedance and so could be tapped across an RF line to make a ‘through’ or ‘bridging’ measurement with minimal disturbance to the circuit under test, or used in conjunction with a 50Ω termination for terminated measurements. The measured value with such an instrument could be affected by the presence of odd order harmonics and, in many cases, even order harmonics also, so their popularity has waned. For higher frequencies, terminating (50Ω or 75Ω) true rms power meters are normally used. The sensors may be thermocouples, or diodes operated at a very low level - where their response is rms rather than linear. A good example is the Boonton model 4300, which is illustrated in Figure 16.1.

The determination of the exact frequency of an RF signal was in former days a complicated business but is now simply a matter of connecting it to a digital frequency meter. Nowadays, a frequency counter function is built into many general purpose DMMs (digital multimeters), such as the Fluke 187 (up to 1 MHz), whilst bench-top timer/counter/frequency meters offer a wider range. A typical example is the Fluke PM6681 which reads up to 300 MHz on its high impedance input and 300 MHz-8 GHz on an optional 50Ω input.

The phase noise of a CW signal can be measured in various ways, the simplest being to use a high grade spectrum analyser. The harmonics of an RF signal can also be measured with a spectrum analyser. This is such a versatile instrument that it is covered in detail later in the chapter.

Modulation measurements

For the measurement of AM, FM or PM the most convenient instrument is a modulation meter. In addition to measuring the modulation depth or deviation, most modulation meters will also make a high-quality demodulated output available for monitoring purposes, and additionally make measurements such as carrier frequency and level, frequency response, signal to noise ratio, stereo separation, etc. It is possible to measure the AM of a signal which also carries FM (or PM) and vice versa. Usually, in addition to manual tuning, an auto-tune function is available to instantly tune the instrument to the only (or largest) carrier present. However, general purpose modulation meters are being replaced by the modulation facilities built into specific radio equipment test sets. Figure 16.2 shows one such instrument, with the versatility to test to many standards, including GSM, PCS, PCN, DECT and CDMA.

Spectrum and network analysers

These instruments are so fundamental to the RF engineer that they deserve a section to themselves. The spectrum analyser is a development of the earlier panoramic receiver, which was a swept receiver displaying the amplitude of any signals it encountered within the frequency range over which it was swept. Apart from greater stability and selectivity, the main difference is that the modern spectrum analyser can display the signals on a logarithmic scale covering (typically) 80 dB at 10 dB per vertical division. Additionally, for finer amplitude discrimination, a vertical scale of 2 dB/division and also a linear scale are usually available. Manufacturers of spectrum analysers include Agilent (formerly Hewlett-Packard), Tektronix, Aeroflex, Anritsu, Rohde & Schwarz and a number of others.

A spectrum analyser may be used for a wide range of measurements, including determining the relative amplitude of any harmonics of an RF signal. It may also be used to measure the phase noise (sideband noise) of an unmodulated carrier, provided of course that the phase noise of the spectrum analyser itself is lower than that of the CW source under test. Another important test conveniently carried out using a spectrum analyser is intermodulation testing. A typical application is testing the linearity of an HF SSB transmitter, by the two-tone test method. Here, two equal amplitude audio-frequency tones, say 1000 Hz and 1700 Hz, are combined and applied to the transmitter’s modulation input, taking care to isolate each tone from the other so that intermodulation does not occur between them, e.g. in the tone generators’ output circuits. A sample of the transmitter’s output is then applied to the spectrum analyser, and if no intermodulation has occurred, the only signals found will be (assuming for example USB modulation) two equal amplitude components at 1000 Hz and 1700 Hz above the suppressed carrier. In practice, the carrier suppression will not be complete, though the usual specification calls for it to be at least 40 dB down on PEP (peak envelope power).

In the two-tone test, assuming that intermodulation is not severe, PEP will be 6 dB above the level of either of the two RF tones. If third order intermodulation occurs in the transmitter, as is bound to be the case to some extent, additional components will be seen in the output, offset by the separation between the tones, e.g. at 700 Hz above the higher frequency tone and at 700 Hz below the lower. The permitted level of these tones depends upon the applicable specification, as published by the FCC (Federal Communications Commission, applicable in the USA), ITU-R (International Telecommunications Union, Radiocommunication Bureau, formerly known as CCIR - International Radio Consultative Committee), or whatever.

The relevant ITU-R specification is Recommendation 326, and this has been embodied in the national regulations of many European companies. This specification calls for the third-order intermodulation products in an HF SSB transmitter operating in J3E mode (formerly known as A3J mode) in normal speech service to be 26 dB down on either of the two tones. The earlier versions of Recommendation 326 were unfortunately worded in such a way that the requirement could be interpreted as being 26 dB down on PEP. My suggested re-wording was submitted to the ITU by CCIR UK Study Group 1, ratified by a Plenary Assembly, and is incorporated in the current version. The requirement for transmitters where a privacy device is fitted is tighter, at 35 dB down on either tone. The higher figure is because a device such as a scrambler will disperse the speech energy throughout the sideband, resulting in a greater likelihood of significant intermodulation products falling into adjacent channels. Both carrier suppression and IMP (intermodulation products) are quickly and simply tested with a spectrum analyser.

Another instrument important to the RF engineer is the network analyser. This measures the analogue characteristics of electronic products including components, circuits and transmission lines. Consequently it is widely used in many fields from R&D to mass production, for analysing the transmission, reflection and impedance characteristics of these products. Manufacturers of network analysers are much fewer in number than those of spectrum analysers. Further, some manufacturers of network analysers produce only scalar instruments, rather than the more generally useful vector instrument. Basically, a network analyser comprises a swept signal source of constant amplitude, and a receiver of constant sensitivity which is always tuned in sympathy with the instantaneous frequency of the source.

In a vector network analyser, the receiver is phase-sensitive and its output can be displayed on the instrument’s display device (formerly usually a cathode ray tube but nowadays usually a colour LCD display) as amplitude and/or phase against frequency (a Bode plot), or on a polar plot, or on a Smith chart. The reference for phase measurements may be the swept source’s output or may be obtained from one of the accessories which are available for use with the network analyser.

A scalar analyser is similar, except that the receiver produces only amplitude information. If the unit under test produces an output frequency different from the source frequency (e.g. a mixer or frequency changer unit), there is no meaningful relation between its output phase and that of the source, so a scalar measurement is the only possible one.

Other instruments

RF signal generators have long been fundamental items in the RF engineer’s armoury and their design has advanced enormously since the days of the Marconi TF144G, known to a generation of engineers, from its wide squat shallow case, as ‘the coffin’. Early types such as the TF144H were simply LC oscillators tuned by a variable capacitor in conjunction with a turret of coils for different ranges. They were designed in such a way as to minimize both the variation of output level with tuning and the amount of incidental FM which was caused when amplitude modulation was applied - and in later models fitted with a facility for frequency modulation, the amount of incidental AM caused when frequency modulation was applied. All high-class signal generators nowadays employ synthesis, so that their medium- and long-term frequency accuracy is equal to that of their ovened crystal oscillator reference. One scheme offering very low noise is direct synthesis: this technique is not to be confused with direct digital synthesis which is discussed in Chapter 9. Early synthesized signal generators using direct synthesis, such as those from

General Radio, used decade synthesis whereas later generation models from Eaton/Ailtech used binary synthesis, considerably easing the design problems and resulting in a generator whose output phase noise really is nearly as good as a prime crystal oscillator. However, for reasons of economy (a direct synthesizer is complicated, and therefore expensive) most modern high-class signal generators use a VCO/PLL approach. An example of such an instrument, of advanced design, is shown in Figure 16.5. This instrument offers a 150 KHz to 2000 MHz frequency range with 10 Hz setability, at a +/− 1 ppm frequency stability from the internal standard. It is also capable of locking to an external frequency standard. The output level range covers − 127 dBm to +7 dBm amplitude, in 0.1 dB steps. Modulation facilities include AM, FM and Phase modulation, internal or external. Manual control is via an 80 character back-lit LCD display, keyboard and rotary encoder control. Full remote control is available through RS232 and GPIB interfaces.

Using the traditional approach, for tasks involving many measurements such as testing a complete radio communications system, a considerable number of different test instruments would be required. There would further be many different interconnection set-ups required during the course of testing, all of which makes this approach unattractive, especially when the test equipment has to be taken to the radios rather than vice versa. For this reason, special purpose radio communications test sets are available from a number of manufacturers. An example is the 4403 Mobil Phone Tester from Willtek, see Figure 16.2.

The humble oscilloscope, although not normally considered as a piece of RF test gear, should not be forgotten. A conventional analogue oscilloscope, given adequate bandwidth, can be used for many RF tests. Obviously, it can be used to measure directly the peak-to-peak amplitude of a CW signal, the rms value being obtained by dividing by 2.828. This assumes that the harmonic content of the signal is low, a point which can be judged adequately if the bandwidth of the oscilloscope exceeds three times the frequency of the signal. Circuit misbehaviour, such as squegging of an oscillator, is instantly revealed by the oscilloscope where otherwise the problem might not be at all obvious.

The oscilloscope can also be used to measure the modulation index of an FM signal. Here, the oscilloscope displays a few or many cycles of the RF as required, whilst triggered from the same RF. At the left-hand side of the screen, all traces will be in phase, but moving progressively to the right, the traces will diverge to the right or left of the average, according to whether the particular trace was written when the frequency deviation was negative or positive. The point where late cycles n cycles across the screen just meet early cycles n + 1 cycles after the trigger point is very clearly visible; the value n + where this occurs marks the point of ±180° peak phase deviation, from which, knowing the frequency of the modulating sinewave, the modulation index is simply derived. The oscilloscope can even be used for quite sophisticated measurements, such as eye diagrams for DPSK or similar digital modulation methods. Here, the oscilloscope displays the IF output of the transmitter modulator (or of the receiver IF) whilst it is triggered from the unmodulated IF carrier. This may be obtained from the carrier input to the modulator, or if the receiver uses synchronous demodulation, from the receiver’s carrier recovery circuit. (The receiver test may be carried out with the transmitter’s IF output patched into the receiver’s IF strip, or alternatively it may include the RF path. In the latter case, however, either the receiver first mixer should be driven from the transmitter’s final upconverter drive, or both TX and RX synthesizers should be run from the same reference.) Finally, a pulse whose frequency is that of the data clock and whose width is about 10% of the data period, is applied to the Z modulation input (bright-up input) of the oscilloscope. The pulse can be triggered by the transmitter’s data clock, or obtained from the receiver’s clock recovery circuit (see Figure 16.6). The bright-up pulse should have a variable delay with respect to the data clock edge: adjusting the delay to centre the pulse on the data-stable period will produce an ‘eye diagram’. Note that if the transmitter modulator includes an all-pass filter providing equalization for both the transmitter and the receiver IF filtering functions, the eye diagram at the receiver’s IF output should (in the absence of additive noise) be considerably cleaner and more ‘open’ than at the transmitter modulator’s output.

Finally a word about field strength measuring equipment - used for a variety of purposes, including EMC measurements. Measuring receivers are specialized instruments which are in some respects akin to a spectrum analyser, but very different in other ways - such as not possessing a visual display. Typical examples would cover 9 kHz to 30 MHz, or 30 MHz to 1 GHz, covering between them measurements to CISPR 16 (bands A to D). Detector response can be selected as average, peak or quasi-peak (CISPR), and in addition to spot frequency measurements, the band or any part of it can be automatically swept. The received level is output to a plotter, together a specification limit line, such as the relevant VDE limit.

Such receivers are used in conjunction with a special measuring antenna, or field probe. Simple E and H field probes have a response which, in terms of the signal strength delivered to a spectrum analyser or measuring receiver, is not constant with frequency. Nevertheless, since they are easily fabricated, they can be useful adjuncts in any RF laboratory. Figure 16.7 shows the response of simple probes in the VHF region, giving the incident field strength in terms of the measured level in dBm on, for example, a spectrum analyser, assuming the probe is in the far field of the source. More sophisticated measurement antennas cover a wide bandwidth, e.g. the HLA 6120 9 kHz-30 MHz HF Loop Antenna from Schaffner Limited. This is an active antenna, providing a constant antenna factor of unity over the whole frequency range, the measured output in dBμV being numerically equal to the field strength in dB μV/m. It is ideal for the 3 m magnetic field measurements to VDE 0871 and FCC 18. The model CBL 6112, from the same company, is in effect a compound antenna. It consists of a bi-conical (bow-tie) element and a log periodic section, permitting testing over the whole range from 30 MHz to 2 GHz with a single antenna. Primarily an emission test antenna, it will nevertheless accept powers up to 300 W for purposes of immunity testing, with field strengths up to 10 V/m or more.

The above measuring antennas are of course not isotropic, since, as was explained in Chapter 14, it is not possible to design an antenna to be isotropic. However, the EMC 20

Wideband Field Probe from Schaffner Limited covering 100 kHz to 3 GHz, is in fact isotropic. It does not infringe Maxwell’s equations, for the head contains three separate orthogonal sensors. The three sensors measure the electric field strength in the three axes individually, and the field strength is computed by the instrument’s processor by summing the squares of the three measured values. If placed in the near field of an emitter, it measures just the E field component of the field. If placed in the far field, at at least one wavelength away and preferably three wavelengths, it again measures the E field, in volts/m, from which the H field in A/m and the power flux density in W/m² can be directly derived, given that the wave impedance in the far field equals that of free space, namely 377 Ω - see Figure 9 of Appendix 11.