RFI Measurement and Monitoring Facilities R. Fisher, March 1997 This is a rough outline of the Green Bank Observatory RFI measurement and monitoring facilities to be used as a general guide for the Interference Protection Group in designing instrumentation for the site. Requirements will evolve with time and experience. The long range instrumentation goals for on-site Green Bank interference protection are: 1. spectrum and time domain measurements in the indoor test range with sensitivity sufficient to detect radiation at least 10 dB below the levels detectable by typical spectral line, continuum, and pulsar observations on the GBT, 2. spectrum and impulsive noise detection capability on a roll-around cart for finding interference sources in the lab and other work areas, 3. several monitor stations around the site with instantaneous direction-finding antennas and RF transmission lines to a common location where the signals can be analyzed, and 4. an RFI analysis facility, which, combined with the monitor stations, has sufficient spectrum and time domain resolution and sensitivity to detect interfering signals seen in typical GBT observations. A long term design goal is to make both the indoor and outdoor systems look like simple spectrum analyzers that cover 30 to 3000 MHz. Manual intervention to switch between bands will limit the usefulness of the systems. Signal analysis and display will rely heavily on existing Observatory software, particularly aips++. User interfaces adapted to the RFI measurements need to be developed so that a trained technician can carry out most of the measurements and evaluations. Automated measurements and special signal analysis techniques are long range goals and will require considerable software development effort. Manually operated measurement facilities need to be implemented quickly. Indoor Test Range The indoor test range is a 11.3m x 4.6m x 4.6m shielded room with absorber on the interior walls. The absorber has reflectivity better than 30 dB above 1 GHz. For RFI tests, where only average radiation levels are important, it is probably effective well below 1 GHz. Where standing waves are prominent at lower frequencies, measurements with several antenna placements may be necessary. A non-conducting table is required to support equipment under test for RFI. The table should have a manually rotatable platform on top about 1.5 meters in diameter capable of supporting a large computer workstation and a few accessories like disk and tape drives. The table can have a few metal parts as long as they don't shield the RFI from the measurement antenna or create a significant ground plane near the equipment under test. A few heavy or bulky items under test will need to sit on the floor, so the table must be easy to remove. The primary RFI test frequency range is 30 to 3000 MHz. When a piece of equipment under test has the potential for radiation above 3000 MHz, the frequency range can be extended for particular measurements. Two broadband antennas can cover the 100 to 3000 MHz frequency range, and a number of simpler narrow band antennas are needed below 100 MHz. These antennas need receive only one polarization, either linear or circular. All of these antennas need to have specified or measured gain and reflection losses to an accuracy of about 1 dB over their usable frequency range. Test antenna preamps should have a noise figure of 3 dB or better with enough gain to overcome the noise of following mixer and analyzer stages. Various filters, mixers, LO's, and IF amplifiers will be needed to cover the full frequency range. These will be specified in a more detailed design. Several calibration signals (CW and broadband) of known power will be required to calibrate the intensities measured with the test antennas. If the test receiving antenna's gain and loss characteristics are known, the calibration can be injected between the antenna and preamp. Otherwise, the signal can be broadcast by another antenna of known gain. The broadband noise calibrator should have a modulating switch driven by the continuum or spectrometer backend in use. A calibration modulator makes data analysis easier in the signal processing software. Since the mobile cart can usually reside in the indoor test range control room, several pieces of expensive equipment can be shared - the spectrum analyzer, oscilloscope, and communications receiver, in particular. Permanently installed in the indoor test range control room will be a PC or workstation tied to the LAN and capable of displaying X-windows and other software tools associated with control of the spectral processor and display of data with aips++ or IDL, for example. This PC or workstation will also acquire data using its own peripherals such as an A/D convertor or an IEEE device controller as required. Three primary analysis signal paths will be used. All must provide for permanent data storage. One will be to a reasonably automated spectrum analyzer for moderate sensitivity scans of a broad frequency range. The second will be an IF connection to the spectral processor for detailed, high sensitivity measurements of bandwidths up to 80 MHz per measurement. The third will be a square-law detected output sampled at rates of at least 10 kHz for periods of at least one minute for detection of broadband transient or periodic emissions. Several filters are needed to limit the pre-square-law-detector bandwidth to selections from about 1 to 100 MHz (center frequency to be determined by available filters). Mobile Cart The mobile cart is intended to track down sources of interference in the Jansky lab or other indoor work areas. Since it will work in a fairly high interference environment, it needs the ability to discriminate between many signals, but it doesn't need as much sensitivity as the indoor range or the outdoor monitoring stations. Below is a list of the main pieces of equipment on the cart. The spectrum analyzer will allow frequency discrimination and semi-quantitative measurement of signals. Below 200 MHz or so, impulsive noise is strongest. This is most conveniently monitored with a tunable receiver set to the frequency where the RFI is strongest but where there are no narrow band signals. Selectable bandwidths from about 3 to 100 kHz are generally best. A speaker, headphones and an audio-bandwidth oscilloscope are best for picking out impulse signatures. The capabilities of the mobile cart are close to what is needed for tracking down local power line and narrow band interference off-site. I don't think that we can share all of this equipment between the indoor range, the mobile cart, and Wes's truck, but we should have the same capability in all three places. To the extent possible the mobile cart should be rugged and stable enough to allow it to be pushed across the outdoor walkways and loaded on a truck and taken to another building. The cart will contain Spectrum Analyzer (30 to 3000 MHz) Preamp(s), 3 dB noise figure and enough gain to overcome spectrum analyzer input noise Antennas Broadband above 300 MHz Electrically short or narrowband below 300 MHz Communication receiver up to ~200 MHz with audio output Low frequency oscilloscope for detected output of communications receiver or unswept spectrum analyzer Outdoor Monitor Stations The outdoor monitor stations contain the antenna and front-end portion of a site RFI monitoring system. Eventually, we want direction-finding capability from 30 to 3000 MHz at a number of places around the Observatory. The purposes are both locating interference seen at the telescope and continuous monitoring of the RFI environment. The signals from all outdoor stations will be piped to one location (probably where the spectral processor is) and share a common set of signal detection hardware. Except for sharing the spectral processor, this will be independent of the indoor test range. The most appropriate antenna configuration for sensitive direction finding over the desired frequency range remains to be determined by experiment. We can start with one station instrumented for, say, the 600 to 1700 MHz frequency range at a location (probably the Jansky lab or the 140-ft) convenient for sending its signals to the spectral processor. This frequency range generally has the most cases of reported RFI, so it will be of most immediate benefit. One idea for the outdoor station is four log-periodic antennas pointed at 90 degree intervals around the horizon. The signals from two antennas are sent to the signal analysis hardware at the central monitoring location. By comparing the relative intensity of the interference on successive antenna pairs a reasonably accurate direction can be computed. In addition to four antennas, this requires four preamps, a 2-pole-4-throw RF switch, two mixers/filters/IF amplifiers, a fixed frequency LO, a frequency multiplexer to get the two signals on one optical fiber, and an RF fiber modem. We can start with modest noise figure, inexpensive preamps until we settle on a usable antenna strategy. A way to remotely control the RF switch will be required. The rationale behind using antenna pairs for direction finding is that interference can vary quite rapidly with time, which makes direction finding with a rotating antenna a tricky and time consuming business. Once the two antennas with the strongest response to the interference are identified by quickly switching between them, the ratio of intensities measured simultaneously by two antennas will give a direction independent of source strength. Initial calibration of the azimuthal antenna patterns will require that the test antennas be mounted temporarily on a mechanical rotator. If this rotator is inexpensive enough, we can consider permanently mounting all antennas on rotators for convenient recalibration. We may find that multi-antenna direction finding is more appropriate for lower frequencies where impulsive noise is more prevalent and that a rotating horn will work at higher frequencies. We will learn with experience. We may also find that one or more of the monitoring stations must be too close to a building with many RFI sources to allow effective direction finding. Simple omnidirectional antennas may be more appropriate in some cases, relying on relative signal strengths at various stations to locate sources of interference. Ultimately the gain of the monitor antenna should be nearly equal to the ratio of the monitor receiver system temperature to the system temperature of the GBT receiver so that they have roughly the same sensitivity to interference. This assumes 0 dBi gain of the GBT feed in the direction of the interference, which may be a bit pessimistic. Central RFI Monitoring Station The primary themes of the monitoring network are to allow the telescope operator or other technical personnel to respond quickly to reports of interference from the observer and to head off potential interference from Observatory equipment by detecting and eliminating RFI before it appears in the astronomical data. To do this well the monitor stations and signal processing hardware must match the sensitivity of the telescope receivers as closely as possible. Also, the displays and signal analysis software must be convenient to use by the telescope operator or a trained technician. Much of the software developed for the indoor test range will be useful with the monitoring stations, but some additional capabilities will be needed for direction finding, for example. In addition to designing the outdoor monitoring stations for antenna gain to system temperature ratios roughly equal to that of the GBT receivers in the direction of interference, we need to provide the same integration capability and frequency and time resolution as is used for GBT observations. Hence, the spectral processor or its equivalent and some wideband time domain sampling hardware and analysis software (FFT's, noise statistics, synchronous demodulation, etc.) are needed in the central monitoring station. The list of post-mixer equipment needed for the indoor test range (filters, detector, A/D convertor, workstation, oscilloscope, etc.) applies to the central monitoring station with the additional requirement of two IF channels instead of one. A conventional spectrum analyzer is needed here, too, for broadband and rapid visual inspection of the passband coming from the monitor stations. Perhaps we can share equipment such as a spectrum analyzer and workstations at a location such as the 140-ft control room. Rough Equipment Summary Indoor test range several dipole antennas below 100 MHz log-periodic antenna 100 to 1000 MHz conical log-periodic antenna 1 to 3 GHz or higher various preamps to cover 30 to 3000 MHz range spectrum analyzer (shared with mobile cart) modest bandwidth oscilloscope (shared with mobile cart) mixer tunable LO (consider sharing tunable oscillators in spectrum analyzer) IF amplifier RF filters for image rejection IF filters (1 to 100 MHz bandwidths) fixed attenuator set square-law detector PC/workstation on the LAN with X-Windows A/D convertor for PC/workstation ( > 10 kHz sample rate ) interface for controlling LO and spectrum analyzer from PC/workstation fiber modem for IF signal to spectral processor audio amplifier with speaker/headphones on square-law detector lazy susan for equipment under test calibration noise source and CW signal of known intensity RF switch for noise calibrator directional couplers to cover frequency range for calibration injection RF switches for convenient and automated changes between measurement antennas, filters, and other routinely used devices Mobile cart non-resonant antennas below about 300 MHz log-periodic antenna 300 to 1000 MHz conical log-periodic antenna 1 to 3 GHz or higher (shared with indoor range) various preamps to cover 30 to 3000 MHz range spectrum analyzer (shared with indoor range) modest bandwidth oscilloscope (shared with indoor range) communication receiver up to ~200 MHz Outdoor monitor station 3 or 4 4 log-periodic antennas 600 to 1700 MHz 4 preamps 2-pole/4-way RF switch with remote control (switching time < 1 sec) 2 RF filters fixed frequency LO with power splitter 2 mixers and IF amplifiers 2 IF filters fixed frequency LO and mixer for multiplexing IF's onto fiber fiber modem Central monitor station spectrum analyzer modest bandwidth oscilloscope fixed frequency LO and mixer for demultiplexing fiber signal 2 IF amplifiers tunable LO and 2 mixers for selecting continuum center frequency within IF passband 2 sets of IF filters (1 to 100 MHz bandwidths) 2 square-law detectors PC/workstation on LAN with X-Windows 2 channel A/D convertor for PC/workstation ( > 10 kHz sample rate ) interface for controlling spectrum analyzer from PC/workstation fiber receiver modem audio amplifier with speaker/headphones on square-law detector cables and connectors everywhere