Frequency domain reflectometry vs. time domain reflectometry Here are the fundamentals of two measurement technologies and a nuts-and-bolts look at the practical implications for RF antenna systems.

RF cable and antenna testing is increasingly important to sustain performance quality and to reduce maintenance costs. Frequency domain reflectometry (FDR) is rapidly gaining popularity in applications above 200MHz. How does FDR technology stack up against the traditional cable testing technique, time domain reflectometry (TDR)?

The simplified drawing in Figure 1 below identifies several common devices on a typical antenna tower. The drawing illustrates how both TDR and FDR techniques rely on the scientific principle that no transmission line component is a perfect impedance match. At any impedance mismatch, some of the energy of an incident signal is reflected backward toward the source. If these reflections are excessive, the antenna system does not operate properly.

Several important factors should be considered when testing RF antenna systems. First, TDRs have been in common use for many years and, as a necessary and practical consideration, experienced technicians have adopted clever tricks and techniques to interpret the displays. Historically, TDRs have been smaller and much less expensive than FDR-based analyzers. FDR systems were only purchased by technicians who needed the higher FDR frequency coverage for microwave applications. Although there no longer are size and price discrepancies, the fundamental technical differences remain. (See Table 1 on page 9.)

Stimulus signal Both TDR and FDR have similar acronyms, and both test for excess signal reflections in RF cables, but there the similarities end. The FDR technique sweeps the antenna system at RF frequencies. A TDR injects dc pulses. Figure 2 on page 9 shows how the different stimuli respond in a typical RF antenna.

The RF sweep used by FDR is sensitive to RF problems and accurately identifies the return loss of the antenna. A TDR's pulsed dc stimulus reflects little energy at RF faults. Further, almost 100% of the TDR's source energy is reflected by the antenna or any other in-line, frequency-selective device such as frequency combiners, filters or quarterwave lightning arrestors. This behavior is predicted according to the spectral energy density of the stimulus sources shown in Figure 3 below right. The TDR's spectral content is splattered across a wide frequency range. The spectral magnitude of the output pulses tends to roll off rapidly at high frequencies. Typically, less than 2% of a TDR's pulse energy is distributed in the RF frequency ranges.

FDR techniques generate an RF sweep that includes only the frequency range selected by the operator, allowing frequency-selective characteristics to be displayed clearly. Measurements include return loss (or standing wave ratio, SWR) vs. frequency and return loss (or SWR) vs. distance. Instruments with frequency ranges as high as 1,200MHz and 3,300MHz are available for $4,000 to $7,000. Coverage to 8.6GHz and 20GHz typically costs $22,000 to $37,000.

The need for higher bandwidth drives the requirements for more accurate TDR time bases, faster pulse generators and faster sampling rates. These requirements drive costs. Several 100MHz-bandwidth TDRs are available today for $3,500 to $4,500. Models with higher bandwidths typically exceed $8,000 to $12,000. Bandwidth is determined by the TDR's pulse rise time. A TDR's pulse generator injects pulsed dc signals or fast-rising dc steps into the antenna cable. The equation in Figure 4 on page 10 is a good estimation for both.

How can a TDR test the antenna system when the lightning arrestor is one of the new quarterwave types instead of the standard gas-discharge capsule? How can the TDR test through duplexers or interference rejection filters? Simply stated, the TDR cannot test bandpass RF signal devices. Thus, TDRs tend to identify dc parametric values that identify performance characteristics below 100MHz to 200MHz. In contrast, the FDR technique pinpoints RF performance information throughout the RF and microwave ranges.

Distance accuracy Again referring to the example in Figure 2, as reflected signals return at 60% to 88% of the speed of light, the TDR's fast-sampling analog-to-digital (A/D) converter digitizes the reflected pulse amplitudes. Synchronization for the generator and digitizer is controlled by a time base, the precision of which decides the distance accuracy in the resulting display. It is no wonder that the most accurate TDR-based instruments come from leading oscilloscope manufacturers.

The FDR technique requires that a frequency oscillator feed a swept-frequency signal into the transmission line. A radio frequency receiver measures the interference pattern generated when the swept RF source output signal adds and subtracts with reflected signals from faults and other reflective characteristics within the tested transmission line. The vector addition of the signals creates a ripple pattern vs. frequency. A fast Fourier transform (FFT) calculates the distance-to-fault. The identical block diagram is used for RF network analysis. Predictably, the FDR-based products come from leading manufacturers of vector network analyzers.

The number of ripples in the return loss (or power) vs. frequency display is directly proportional to the electrical distance to the reflective point on the transmission line, as shown in Figure 5 on page 16. This electrical distance indication is inherently accurate and suffers no accuracy degradation due to TDR-type time base controls.

The electrical distance is related to physical distance according to the propagation velocity, kp, similar to the TDR's velocity correction factor for the same property. Cables with different dielectric types maintain different electrical propagation speeds. In the following equation, the integer number of displayed ripples, n, relates to fault location, L, according to:

L = (n*c*kp)/(2*[DELTA]f

where

c = speed of light [DELTA] f = frequency sweep width

Thus, the FDR analyzer's distance accuracy is determined by knowledge of the cable's propagation velocity.

In general, TDRs have better distance resolution when testing short cables. When cables are less than 10 feet long, the FDR technique requires a wide frequency sweep to achieve adequate resolution because FDR's fast Fourier transform requires several frequency display ripples before adjacent problems are isolated. This is the primary reason why TDRs are preferred for short cables that cannot exceed 100MHz frequency range. A good example is a twisted pair cable for computer local area networks (LANs). The FDR approach is better for coaxial cables, as commonly are used with Ethernet-based LANs.

Improving FDR distance resolution is the reason design engineers will use 67GHz and 110GHz coaxial vector network analyzers (VNAs). The frequency range is needed to isolate segments of the circuit designs to sub-millimeter distances using the VNA's "time domain" mode.

If we consider all the factors involved in a typical antenna system measurement, including not only the accuracy of TDR and FDR techniques, but also the ability to actually measure the physical length of an installed cable, distance accuracy requirements are not particularly strenuous. Field service technicians need accuracy and resolution to a few feet for transmission lines from 10 to 1,000 feet long. This is easily accomplished by both TDR and FDR techniques.

Return loss (SWR) accuracy FDR techniques are National Institute of Standards and Technology (NIST) traceable and include several RF measurements such as insertion loss, gain and RF power in addition to return loss and SWR. TDRs do not display RF return loss nor any other RF parameter. Theoretically, the reflected responses from an ideal pulsed dc signal stimulus can be converted into frequency domain information by integrating the pulse reflections over the interval of reflection from the device. (See Photo 1 on page 10.)

Today, fast Fourier transform of TDR data is the subject of advanced research. Some limited functionality is available in some GHz-rate oscilloscopes. Practical limitations for low-cost field instruments include size, sample rates, time base control and traceable reference calibration standards. Despite commercial claims of high equivalent bandwidth, pulse TDRs do not provide sufficient effective directivity for accurate RF frequency tests such as return loss and insertion loss.

By contrast, the accuracy of the FDR technique is traceable and well-defined in terms of directivity and measurement linearity. Source match and load match of modern FDR analyzers are excellent, reducing these error sources to negligible levels.

Further, the FDR methods do not suffer from the "dead zone" problem common to TDR implementations. The dead zone results from the time required for the TDR's A/D sampler to begin sampling after the dc pulse generator injects a pulse.

If it's important to reduce costs and improve quality through preventive maintenance, return loss accuracy to less than +/-1.0dB is an absolute requirement. Accuracy determines test repeatability, which determines the ability to spot the initial symptoms of the problem. Under the failure prevention philosophy, return loss accuracy is much more important than distance accuracy.

Sensitivity Sensitivity is excellent on both FDR and TDR analyzers; however, they're sensitive to different characteristics. Whereas the TDR is sensitive to lumped-dc parametric characteristics, it is likely to miss RF characteristics such as corrosion, slight pin gaps and damaged RF components. The TDR has severe difficulty "seeing" past any RF component with a passband characteristic, such as filters, duplexers, quarterwave lightning arrestors and antennas. Figure 6 above shows the basic limitation of TDR techniques to frequency-selective performance problems. This problem is caused by the large proportion of dc spectral content of the TDR's stimulus signal.

The FDR technique is highly sensitive to any dimensional tolerance change within the RF conductor path. Even a tiny variation in return loss, such as a connector changing from 25dB to 22dB return loss, is clearly detected. Although a 22dB figure would still pass specification, it is important to identify the change in impedance. If the connector has changed, then the weather seal is also likely to have changed. Over time, water will seep under the seal and destroy the cables. During this time, the cell site coverage area will contract. Holes within the coverage area of the transmitter will expand, reducing service quality. The connector will corrode and form a dissimilar metal-to-metal contact. This "diode" will cause passive intermodulation distortion and adjacent channel power problems for digitally modulated signals.

My company takes the position that it is better to find these small problems before complete failure causes outage and expensive repairs. Because of the lack of adequate sensitivity and RF accuracy, TDR techniques amount to a "fix-after-failure" service philosophy. Only FDR techniques provide the ability to implement preventive maintenance plans.

Immunity to interference For RF system design, zoning and economic reasons, RF transmitter antennas are frequently co-located. The explosion in wireless communications implies that the sources of RF interference will increase in the future, not decrease. TDRs, being baseband sampling devices, will produce extraneous readings in the presence of nearby transmitters. Averaging helps, but if the interference is above a few milliwatts, TDRs frequently fail to measure anything.

Modern FDR analyzers are built to reject interference by using ac detection or auto-zeroing and advanced designs such as phase-tracking vector receivers and spread-spectrum techniques. The FDR measurements must be immune to external interference so tests can be performed without shutting the site's transmitters down. Site locations at hill top "antenna farms" can be tested even with very high ambient RF energy into the antenna. (See Photo 2 on page 12.)

Compensates for cable insertion loss FDR tests the cable and antennas at RF, their frequency of operation. By measuring RF characteristics directly, FDR provides a clear performance indication. By virtue of using the same RF frequencies as these devices, FDR techniques easily measure the antenna system and accurately reflect the return loss performance of each device versus distance.

FDR also compensates for the insertion loss of the RF cables; thus, the antenna return loss display is not perturbed by the cable's inherent insertion loss. (See Photo 3 on page 12.) This allows easy identification of problems, such as moisture collection or antennas damaged by lightning, at the top of the tower.

Commercial TDRs do not compensate for RF insertion loss values because the cable's RF insertion loss characteristics are frequency-dependent.

Test device types FDR techniques have a clear advantage testing antennas, filters, duplexers and waveguide. Because these devices are found in antenna systems, FDR techniques are also preferred for coaxial antenna feeds. When checking coaxial cable for length, TDR or FDR is adequate. If return loss is also tested, FDR is clearly a better choice.

Testing twisted pair cable is typically best performed by a TDR device because twisted-pair cables have a limited frequency range, reducing the resolution of any FDR display. Further, twisted-pair cables are typically used for baseband digital signals, not RF. Coaxial cable, such as those for Ethernet LANs, can be adequately tested with either TDR or FDR instruments.

In short, TDRs are appropriate for baseband dc tests and FDR techniques for RF device tests.

Commercial FDR products The 25MHz-to-3,300MHz Site Master (Photo 3) solves several nagging problems. First, the synthesized source and novel phase tracking receiver technology reject ambient interference. Second, vector network analyzer measurement techniques ensure test accuracy. Third, FDR technology is used to identify very small changes in RF impedance versus distance. Finally, the battery-powered package is lightweight and rugged.

In addition to return loss, SWR and distance-to-fault, the 54100A network analyzers also measure insertion loss, gain, precision return loss, directivity, power and group delay.

Summary FDR-based distance-to-fault techniques can detect minute problems. If problems such as vapor intrusion, loose connectors, antenna bandwidth drift, corroded conductors or damaged lightning arrestors can be identified before severe damage/failure occurs, corrective action can be taken. The timely replacement of the problem component mitigates the need for replacement of the complete antenna system later.

FDR finds potential problems quickly and reliably, allowing cellular service professionals to implement preventative maintenance plans and reduce costs per cell expenses. As a trouble shooting tool, FDR techniques pinpoint damage and impending failure conditions.