A remote diagnostics system for the MERLIN array D. Kettle, N.Roddis Jodrell Bank Observatory University of Manchester Macclesfield SK11 9DL 1. Introduction Jodrell Bank Observatory operates an array of multi-frequency radio telescopes, known as MERLIN (Multi Element Radio Linked Interferometer Network) on behalf of the Particle Physics and Astronomy Research Council (PPARC), see figure 1. The most distant of these, a 32 metre dish, is located near Cambridge; it is equipped with a suite of 4 low noise receivers covering various frequency bands from 1.3 GHz up to 24 GHz. All of the radio telescopes are operated by remote computer control and only the site at Jodrell Bank is manned; signals from the remote sites are sent back to a central processing centre at Jodrell Bank via analogue microwave links. Receiver maintenance and fault finding are carried out by engineers and technicians based at Jodrell Bank; for health and safety reasons it is normal practice for two people at a time to visit remote sites, and in the case of the Cambridge telescope there is usually an overnight stay involved. Until recently there was very little information relayed to Jodrell Bank to enable remote diagnosis of faults in the receiver systems. Hence, in the event of a fault being reported by astronomers using MERLIN, the normal response was to despatch two technical staff to the site to investigate the fault and determine the nature and location of the problem. This would usually be followed by a second visit by a specialist technician or engineer (plus assistant), armed with the appropriate spares. We developed the remote diagnostic system described here partly in an attempt to reduce the number of costly visits to remote sites, thereby saving money and releasing technical staff effort for other tasks. However, another attraction of the system is that it will enable technicians to test receivers at remote sites from Jodrell Bank on a routine basis, allowing any receiver degradation to be spotted early. A pilot system has been successfully installed at the Cambridge telescope, and similar systems will be installed at all remote sites when funding permits. 2. Remote measurements system 2.1 Measurement functions A simplified block diagram of the Cambridge receiver system, together with the points used for remote diagnostics, is shown in figure 2. A purpose built switch matrix, under control of the diagnostics PC, is used to route the signal from any point to any one of up to four measuring instruments; currently available are a spectrum analyser, a power meter, and a data acquisition card. Together with the diagnostics software, these instruments are used to make the following measurements, which are described below. Spectrum (band shape) display at IF, baseband and composite signal output System noise temperature System sensitivity (including antenna gain) Receiver total power stability 1
Each instrument is controlled and/or read by the diagnostics PC, which can in turn be controlled by a remote PC via a telephone line. A signal generator, controlled via GPIB, is used as a local oscillator reference during stepped frequency measurement, for example system noise temperature. In addition to its signal routing function the switch matrix unit is used to select whether the local oscillator reference source is the signal generator or the on-site synthesiser (which is used for radio astronomy). It also controls the positioning of microwave orber material over one of the feed horns during system noise temperature tests. 2.2 Software Two commercial software packages have been used in the development of the remote diagnostics system: National Instruments LabVIEW and Symantec PCAnywhere. Programs written in LabVIEW are used to control and read the instruments, to control the switch matrix, to perform calculations on the measured data, and to display results. PCAnywhere is used to provide communications with the remote PC and printer, via a telephone line and modems. 3. System Noise Temperature Measurements. System noise temperature is determined by making two power measurements at each frequency of interest. The total noise power output from the receiver is measured with the antenna pointing at cold sky, to give the value P sky. The feed horn is then completely obscured by microwave orber material, of known temperature, and the measurement is repeated to give P. The Y factor is then calculated, and used to determine the system noise temperature. Y = P P sky T + Trec Tsys =, Y where T is the physical temperature of the microwave orber in kelvins, and T rec is the noise temperature of the receiver, measured in the laboratory. 3.1 Remote controlled orber A purpose built orber positioning mechanism is fitted to the feed horn of the receiver. Figure 3 shows the orber positioning mechanism on the C band (4.5 to 5.2GHz) feed on the MERLIN 32- metre diameter radio telescope at Cambridge. The orber mechanism is driven by a DC motor, with two limit switches setting the range of travel. A local control box, mounted adjacent to the mechanism, provides the necessary drive power to the motor and allows local control of orber. The local control box is connected to the diagnostics system switch matrix via a multi-core cable down the radio telescope, thus allowing the diagnostics system PC to control the orber. 2
3.2 Measurement hardware and software and example results The remote diagnostics system uses a commercial microwave/rf power meter to determine the two power levels at the baseband points (15MHz bandwidth) in the radio astronomy receiver system. Full band analysis of the noise temperature is performed by selecting the diagnostics system signal generator to be the system reference local oscillator, and then stepping across the full receiver band recording the sky and orber power levels in turn. Should greater frequency resolution be required then additional narrow band filters may easily be added in front of the power meter head; this may be either in the baseband or at IF. The user defines the measurement by choosing various options from a set of drop down menus: Noise Temp.vi LO Control Observing Frequency Output MERLIN signal gen. L-Band L-Band 5GHz 22GHz DAQ card power meter synthesiser Low High The user is presented with a virtual front panel which is completed with name and temperatures etc. The program is executed and the signal generator steps across a desired band, whilst the power meter takes a reading. The orber arm is driven into position over the end of the feed, thus introducing a known quantity of noise, and the program is executed a second time. Figure 4 shows the corresponding virtual front panel. The calculations described above are performed and the system noise temperature, in kelvins, is plotted against frequency. The plot is displayed using HiQ. This is a high-performance problem solving environment where a LabVIEW user can analyse, visualise and print out data and graphs. 4. System Sensitivity Measurement. We can measure the overall sensitivity of the radio telescope, Tsys in Jy, by comparing the received power level from an astronomical calibrator of known flux with the power received when the telescope beam is pointed at cold sky. We define r to be the ratio of these two powers for a 26 calibrator of flux S janskies (1 Jy = 10 Wm -2 Hz -1 ). S Then T Jy = ( r 1) Essentially, the hardware and software configuration for sensitivity measurements are similar to that used for measuring the system noise temperature. However, instead of using the orber, the total power is determined when pointing the radio telescope at a radio star calibrator of known strength and then at cold sky. This measure of sensitivity includes the effect of antenna gain, and is the best indicator of the overall radio telescope performance. 3
5. Receiver Stability Measurements. Radio astronomy receivers must be extremely sensitive in order to be of use, since the signals that they receive are very weak. However, for most applications sensitivity alone is not sufficient; it is also vital that the total power output from the receiver should be stable in time. We use the microwave/rf power meter measuring system described above to measure the total output power at the baseband point. This is recorded at 25 ms intervals for a period of 1 minute to give an indication of the power variation with time. The data are then processed (fourier transformed) to generate a frequency spectrum which gives a 1/f noise signature for that receiver channel, and shows up periodic variations such as microphonics. Stability measurements are the most straight forward as they only require the program to control the power meter and take a reading every 25ms as described above. Figure 5 shows the virtual front panel of the stability program. The program automatically displays the time series and corresponding frequency spectrum, obtained by fourier transform when the data collection is completed. 6. Spectrum Analyser The most valuable single feature of the remote diagnostics system is the ability to view, remotely, the frequency spectrum of signals at various points in the radio astronomy receiver system. Almost all fault diagnosis begins with the examination of such spectra, and it is of immense value to be able to do this without the need to visit the site. In common with the other measurements, a series of drop down menus is used to define the spectrum analyser measurement. Once the user has made a decision on the required measurement, the program remains in a suspended state until spectra are collected. The appropriate band-shape is displayed on the graph and is up-dated until the stop collecting button is pressed. There is an option to print out the band-shape that has been collected. Figure 6 shows the front panel of the program, together with a C Band receiver IF band pass response captured remotely. 7. Conclusions and Future Work. There are substantial benefits of this scheme: causes of faults can be determined remotely, allowing repairs to be carried out more efficiently, and reducing the number of costly site visits. Routine testing of receivers is made very easy, allowing faults to be picked up early and facilitating the detection of degraded receiver performance. The measurement system is fully functional, but further development of the software is planned to simplify the user interface, making it more accessible to non-specialist staff, and increase the level of automation. Similar systems will be installed at all MERLIN remote sites. 4
microwave orber modem GPIB card diagnostics PC DAQ card receiver front ends UHF, L-band 5 GHz, 6 GHz, 22 GHz IF1 IF2 LO ref synth sig gen inputs switch matrix spectrum analyser power meter LHCP RHCP IF points IF converters baseband points IDS modulator IDS modulator point microwave link 5