Optical fiber provides many undisputed benefits for data-transmission systems. It also offers significant performance advantages in RF transmission systems, in spite of the fact that RF engineers have mistakenly considered fiber-optic cables to have lower dynamic range and higher costs than metal coaxial cables. Because of these beliefs, optical fiber has been limited to RF applications where long-distance transmissions have been beyond the capabilities of coaxial cables. With the emergence of new coaxial-to-fiber converters and all-optical switches, RF engineers may want to take a second look at the benefits of optical fiber transmissions lines for their systems. These new components are achieving enhanced dynamic-range performance with lower noise figure levels, all at lower systems costs.
One large federal systems integrator recentlydiscovered the benefits of using this fiberoptic technology combination in the design of a system that can dynamically connect RF receivers to an antenna array with each antenna array having multiple octave-band outputs. The system, intended for use in a classified application, was facing budget constraints. The system had to receive signals from multiple octave-band antenna deeds from multiple antennas in an array for a broad spectrum of frequencies to 18 GHz. The system had to connect up to four receivers to any single octave band in the array at different times. At a cost of more than $100,000 per receiver, equipment costs alone could have made the project impractical if implemented in the traditional manner in which microwave frequency receivers are co-located close to the antenna.
But by using fiber and sharing assets via an optical switch, it was possible to use fewer total receivers in a pooled asset configuration while meeting the sensitivity and dynamicrange requirements of the project. The fiber and optical switch would replace the conventional solution in which the receivers and an RF switch are placed close to the antenna and RF signals are routed over the long coaxial cable runs through an intermediate-frequency (IF) switch to the processing subsystem. The longest distance between an antenna array and the operations center was 1000 feet. Other runs were shorter but still at distances considered too long for conventional coaxial cables. Fiberoptic cables, unlike coaxial cables, can easily transmit RF signals cleanly over this distance without sustaining unacceptable insertion loss. With this approach, it was possible to bring the RF signals back to the receivers enclosed in the relatively benign environment of the operations center allowing them to be utilized in a pooled configuration, versus positioning the receivers near the antenna arrays. Therefore a smaller number of receivers were required to implement the system. The design, however, required some leaps of faith that flew in the face of traditional RF design strategies.
The two-tone third-order dynamic range of the system had to be at least 50 dB as measured in a 1 MHz resolution bandwidth. Based on conventional RF thinking, the dynamic-range performance of an optical fiber system was only going to achieve 40 to 45 dB. Also transmissions from each of the antennas would have to be connected and sometimes multicasted between the antennas and a shared number of receivers by using a switch. This was another area for potential signal losses. But RF switches capable of this type of dynamic-range performance that also support nonblocking, full fan-out connectivity are very expensive. The need to multicast RF signals and connect them to multiple receivers implied a need for amplification, which would add cost and noise to the system. For an optical solution, a high-performance optical switch was required, along with optical amplifiers capable of good lownoise performance. After all, an optical solution capable of covering the 1000-foot distance, passing through RF-to-fiber converters, through the optical amplifiers, moving through the optical switch and terminating at each receiver would have to exhibit noisefigure performance of less than 5 dB to be effective.
Part of the solution involved assembling a risk-reduction test consisting of single-mode optical fiber, RF-to fiber converters from Optium Corp. (www.optium.com), a 1550-nm optical amplifier, and an Intelligent Optical Switch from Glimmerglass www.glimmerglass.com> (see figure). The test system was evaluated by several RF test engineers and found to provide excellent performance. It consistently delivered dynamic ranges of 63 to 70 dB for a 20-Mb/s receiver bandwidth as measured with a 1-MHz spectrum analyzer resolution bandwidth with no variation over the various different spans of optical fiber, from 650 to 1000 feet. The system also delivered a dynamic range of 62 dB at 18 GHz, which was more than 10 dB better than had ever been tested in this particular installation because of the minimum distance at which the receivers could be mounted. The measured frequency response from 12 to 18 dB was flat within 1 dB over these considerable spans, compared to the significant expected gain slopes experienced by coaxial cables. This is a significant benefit for systems that have wide receiver bandwidths 500 MHz or greater.
In fact, the RF signal path to the receivers was so clean that receiver sensitivities ranged from -110 to -112 dBm, results that could not be matched by directly connecting a test signal generator to the inputs of the receivers. This was because the residual gain arriving at the receiver-input (10 to 13 dB) had the effect of lowering the cascaded noise figure of the entire path to the receiver output from 2 to 3.2 dB. The receiver as a stand alone exhibits a noise figure of 11 to 15 dB.
WORST-CASE NOISE FIGURE
The optical multiple antenna-toreceiver hookup was so effective that in the 12-to-18-GHz octave from the system's low-noise-amplifier (LNA) input to the RF connector at the receiver input, the worst-case noise figure was only 1.7 dB over a 1000-foot run of optical fiber from the antenna. The system's tangential sensitivities were about -110 dBm with total gain variations of about ±1 dB-essentially the performance of the LNA.
Using this blend of optical components, a system was designed using one-half the originally estimated number or receivers. The entire implementation was deployed for essentially the cost of a single nonblocking, full-fan-out RF switch. The combination of low cost and outstanding RF performance eliminated any consideration of fielding a system based on coaxial cables. This opens new options for systems designers who have systems problems in which the antennas would best be placed at significant distances from the operational areas. It is also useful in applications where the reduced weight of the fiber yields an advantage.
The combination of optical fiber cables, RF-to-fiber converters, and the high-performance optical switch made it possible to reduce the number of receivers while costing less than a system designed to share receivers among multiple antennas using a conventional RF switch and copper-based cables. RF system engineers must take a second look at the improved capabilities of fiber-optic cables and components to appreciate the lower cost, lower noise figure, improved sensitivity, and enhanced dynamic range possible especially in applications with long distances from antenna to receiver.