Frequency measurements are critical to many tactical systems, such as direction-finding (DF) subsystems in electronic-warfare (EW) systems. Measurement results must be delivered instantly, requiring the use of specialized assemblies known as digital frequency discriminators (DFDs) and instantaneous-frequency-measurement (IFM) receivers to determine the frequencies of detected signals, which are typically high-speed pulsed waveforms. Specifying a DFD or IFM requires a sorting through a myriad of available hardware options and understanding the significance of different performance specifications.

DFDs are building blocks for IFM receivers, but are also embedded within a wide range of military systemsincluding radar warning receivers (RWRs), electronic-countermeasures (ECM) systems, and electronic support measures (ESM) platforms where they provide instantaneous frequency-measurement capability. A DFD can be thought of as a broadband measurement module, while an IFm surrounds one or more DFDs with additional signal processing hardware to form a complete receiver measurement system. DFDs typically provide output voltages that are proportional to the frequency of an input pulse or continuous-wave (CW) signal, and those voltages are often used to drive a display to show details on a received signal. An IFM receiver, on the other hand, can process those voltages by means of a high-resolution analog-to-digital converter; perform calculations by means of dedicated software; and effectively provide real-time information on the frequency, amplitude, pulse width, and time of arrival (TOA) for detected pulsed and CW signals across bandwidths as wide as 0.5 to 18.0 GHz.

In the case of IFMs developed at Wide Band Systems, IFM receivers can be supplied with extensive additional functionality, including information on pulse angle of arrival (AOA), frequency modulation on a pulse (FMOP), phase modulation on a pulse (PMOP), and various customer-defined parameters. To achieve such extended capabilities, IFM performs advanced digital signal processing (DDP) at the back end of the system, but also relies heavily on its DFDs for initial frequency measurements.

The term "digital frequency discriminator" is somewhat misleading since it suggests the use of one or more high-speed analog-to-digital converters (ADCs) near the input port to digitize unknown detected signals and provide digital code for identifying those signals. This is one possible implementation of a digital DFD, but is not the architecture currently used for most DFDs, which should technically be termed "analog DFDs." While it is true that a well-designed analog DFD can decipher signals to 18 GHz across an unambiguous phase range of 0 to 360 deg., and do so with as much as 14-b resolution, the digital code is generated at the back end of the assembly, rather than at the front end. At the front end, an analog DFD employs a series of discriminators or correlators, typically based on microwave mixers, to convert an incoming RF/microwave signal to a voltage that is proportional to the phase and frequency of that signal. It is that voltage that is processed by means of an ADC or application-specific integrated circuit (ASIC) to provide digital code for frequency/phase analysis.

From the block diagram (Fig. 1), it is apparent a DFD is a critical part of an IFM. DFDs typically rely on delay lines to measure the frequency of an incoming signal, using the relationship between the nondelayed signal and different delay times to determine information about the phase and frequency of the incoming signal. In the design approach used at Wide Band Systems, input signals are divided by means of a broadband, matched-phase power divider into multiple signal paths, which are then processed with and without delays. Delays can be external to or built into a component known as a microwave correlator. Delays are chosen for specific center frequencies across the full frequency band of interest, with each successive delay time doubling in length compared to the delay time for the previous correlator. The shortest correlator delay determines the unambiguous bandwidth that can be covered in a DFD, while the longest correlator delay sets the frequency measurement accuracy and frequency resolution.

This 2:1 delay line sequence makes it possible to use delays associated with a series of center frequencies that are sufficiently separated to allow broadband frequency coverage. The delayed and nondelayed signals are then processed by means of a correlator to produce output voltages that are mathematically related to the relative phase between the delayed and nondelayed signals. Since the delay time of the delay line is precisely known for the selected center frequency, the relationship between the delayed and nondelayed signals at that frequency are theoretically also known. That information can be used as a reference to determine details about delayed and non-delayed signals at other frequencies within that correlator.

Each signal path to a correlator uses a delay line with delay of t (or 2t, 4t, etc., as in the Wide Band Systems correlator design). That value, t, represents a fraction of the period for a center frequency of interest, f0. By comparing the relative phase of an input signal between paths, a determination can be made on input frequency. By digitizing the voltages from the DFDs, additional information can be learned about an input signal, such as amplitude and TOA. The relative phase between the two signal paths, θ, is a function of the frequency and the time delay:

θ = 2πf0t

Each correlator generates a pair of output voltages that is proportional to the sine and cosine of the signal phase:

V1 = 2sin(2πf0t) = 2sinθ
V2 = 2cos(2pf0t) = 2cosθ

These voltages can then be digitized by high-speed, high-resolution ADCs for further analysis in an IFM. The length of the delays in each correlator will determine the frequency range that can be measured, so covering a wide frequency range involves the use of parallel signal paths using multiple correlators with appropriately chosen delay times. Because of this architecture, IFMs and DFDs are designed to process single CW input signals or pulses and cannot accurately measure multiple CW signals or pulses at the input to the receiver. But such interference can usually be avoided by using some form of bandpass filter at the front end of an IFM receiver or prior to the input of the receiver. In the case of known interference signals, such as from a broadcast station, a band-reject or notch filter might also be used to suppress or eliminate the unwanted signals from the spectrum under study.

Because DFDs are so critical to accurate frequency measurements in such a wide range of EW systems, specifying a DFD (Fig. 2; picture of a DFD) requires a careful analysis of key performance specifications. These include instantaneous or operating frequency range, unambiguous bandwidth, frequency resolution, frequency accuracy, number of bits resolution, dynamic range, minimum pulse width, simultaneous signal level, signal-to-noise ratio (SNR), probability of detection, probability of false alarm, power consumption, and throughput for a given update rate. Mechanical and environmental specifications include size, weight, type of packaging, operating temperature range, and performance under shock and vibration, among other operating environment factors.

In comparing DFDs from different suppliers, it is importance to "normalize" the performance specifications, since manufacturers may provide performance data in different ways. For example, some manufacturers may provide accuracy performance at ambient temperature (room temperature or +25C), over the full operating-temperature range, or both. Given the delay-line nature of the correlators uses in DFDs, with delay lines being typically sensitive to changes in temperature, it is critical to review the performance of a DFD across the full expected range of operating temperatures. The highest accuracy tends to occur at +25C, with somewhat degraded accuracy at the lower and upper temperature extremes. Depending upon the manufacturer, DFDs are available for numerous operating temperature ranges, including -20 to +50C, -40 to +85C, and 0 to +60C; it is important to gauge accuracy as a function of the full operating-temperature range. Various techniques can be used within the correlators or a DFD to stabilize the delay lines with temperature (Fig. 3), including the use of internal heater circuitry, although this approach tends to increase the power consumption of a DFD.

Accuracy can also be impacted by the integrity of the impedance match between a DFD and its application. This can be noted by a DFD's voltage-standing-wave-ratio (VSWR) specification, typically for a 50-Ω impedance match. This value should indicate a low VSWR, typically of 2.0:1 or better, to ensure minimal phase distortion of input signals to the DFD.

Continue on Page 2

Page Title

Some manufacturers will provide only the instantaneous bandwidth for their DFDs, while others will also note the unambiguous bandwidth, or total frequency range over which the DFD can provide measurements. Of course, accuracy and resolution specifications will only apply to the specified instantaneous bandwidth. As part of the frequency range specification, some manufacturers will list the minimum detectable pulse width, typically in the range of about 50 ns, depending on the total frequency range of the DFD.

DFD dynamic range is usually listed as a span of amplitudes, from minimum to maximumsuch as -50 to +10 dBm or -60 to 0 dBmin both cases a 60-dB dynamic range. It may also be given in terms of signal sensitivity, such as -60 dBm, and maximum input level. It is important to note that another amplitude-related specification, variously given as signal over noise, pulse-on-pulse level, or detection level, refers to the acceptable power difference between two pulses required for detecting one of the simultaneous signals with the specified DFD accuracy. A signal/noise or pulse/ pulse level of 3 dB, for example, denotes that one of two simultaneous signals must be at least twice the amplitude level of the other signal in order for an accurate measurement; the second signal is not detected in this case. For a DFD with this specification, simultaneous signals separated by less than 3 dB in amplitude cannot be differentiated for an accurate measurement. When needed, a DFD can be designed to provide an error message or flag of some form to indicate a possible error when faced with a measurement on two simultaneous signals.

It is also important to understand time-related specifications for a DFD, since most manufacturers will list several different pulsed-based parameters, including minimum pulse width, throughput time, and processing time. The minimum pulse width is simply the shortest pulse that can be resolved over the frequency range of the DFD with the specified resolution and accuracy. In most cases, the root mean square (RMS) accuracy for a DFD may be twice as much as the frequency resolution, such as an RMS accuracy of 4 MHz for a unit with 2-MHz frequency resolution, although this different varies from manufacturer to manufacturer. In all cases where a DFD has been conservatively specified, the frequency resolution will be a smaller number than the frequency measurement accuracy.

Not all DFD suppliers list throughput (or recovery) time and processing time, which are specifications that exceed the duration of the minimum pulse width. The throughput time is the minimum amount of time needed for a DFD from one pulse to the next to make an accurate measurement. The processing time includes the time needed for the internal or external trigger signal to settle to where it is ready for the next measurement. For a DFD with 100-ns pulse measurement capability, typical throughput and processing times might be 200 and 300 ns.

DFDs can be specified for operation with internal or external triggers; some suppliers will offer units that can operate in both internal and external trigger modes. In the internal mode, a DFD will typically make a measurement on all pulsed or CW signals that are within the dynamic range and, for simultaneous signals, above the minimum threshold level or necessary amplitude separation distance for the multiple signals. In the external trigger mode, the DFD will make a measurement every time it receives an external trigger signal, such as a TTL pulse.

PACKAGING AND POWER
Selecting a DFD for an application is often as much about fitting mechanical requirements as meeting electrical performance needs, and DFDs are available in a number of different footprints. These range from standard 19-in. rack-mount units to more compact modules, and even include VME-based plug-in units. These housings can be equipped with a variety of control interfaces, including TTL, Ethernet, and RS-422 connections, depending on the application. Power consumption, which is specified in watts (W), will vary from unit to unit depending upon the amount of functionality provided in a DFD. For example, a DFD that provides amplitude measurement capability in addition to its traditional frequency measurement capability will consume more power than a unit without that capability. DFDs may also incorporate amplification, such as limiting amplifier to stabilize incoming signal levels, and video amplifiers prior to digitization of output video signals. Additional amplification will also account for increased power consumption.

In short, choosing a DFD calls for a careful comparison of a number of key specifications. The requirements of a particular application will also dictate which of those specifications will carry more weight. For example, some applications may not require the highest frequency resolution. In such cases, DFD manufacturers such as Wide Band Systems have supplied DFDs with as little as 6-b resolution. At the other extreme, the firm has also supplied units with as much as 16-b resolution. The key to successfully specifying a DFD is to match its performance to the needs of the application.