The minimization of dissipative losses in radio astronomy receivers is a never-ending engineering task. Any resistive loss also adds noise to the system. Just 1% (0.04 dB) of loss in an RF component at room temperature at the input of a receiver will add 3 Kelvin of noise to the system temperature. With a typical system temperature of 20 K, this translates to 15% loss of sensitivity and a 32% increase in the required observing time. Small resistive losses also degrade the quality of frequency-selective filters needed to reject strong interfering signals that can overload the low-noise amplifier stages.
Cooling a lossy component not only reduces the noise it adds to the system in proportion to its physical temperature, it also reduces the resistive loss, even in normal metals, such as copper. In the October 2008 issue of the International Journal of Infrared and Millimeter Waves, Ricardo Finger and Tony Kerr describe their careful measurements of low-temperature RF losses in a variety of metals typically used in front-end receiver components. The bulk resistivity of very pure copper is well known to drop by several orders of magnitude at temperatures below about 30 K. A factor of 50 is typical for a good grade of commercially available copper. However, RF current flows only near the surface of a conductor so a 50X reduction in bulk resistivity produces only a 7X improvement in RF loss, and this requires a smooth surface on the scale of the RF “skin depth” (less than a micrometer).
To test whether the RF loss reduction at low temperatures is realized in practice and whether some metals and surface finishes are better than others, Finger and Kerr performed two experiments. The first was a DC measurement of a thin copper strip immersed in liquid helium to verify that its bulk resistivity dropped by the expected factor of 50. The second experiment consisted of a simple resonant microwave structure, shown in Figure 1, where the sharpness of its resonances, its “Q”, was inversely proportional to the RF loss in the metal conductor in the structure.
Some of the results of the resonator tests on various forms of aluminum, copper, and gold are shown in Figure 2. The ratios between warm and cold losses ranged from about 2 to 4 instead of the 7 expected for copper, and the better materials at room temperature were not the better materials when cold, which suggest that surface properties play an important role. Interestingly, a particular gold plating process, called “Pur-A-Gold125”, gave the best cold results and will be given serious consideration in future component manufacture.
The ultimate low-loss materials are the superconductors. These are used in mm-wave SIS mixers, which, as the ‘S’ in the name implies, must be superconducting to work at all. However, these superconductors require refrigeration to ~ 4 K. Most cm-wave receivers are cooled to only 15 K or higher. High-temperature superconductors (HTS) work at temperatures as high as 60 K or above, but their manufacture is complex and they have been economical for RF applications only in high volume, such as the cellular telephone industry. However, this has changed recently with the advent of a small number of firms that specialize in HTS circuit layout and manufacture.
In a collaboration with Michael Lancaster and Srikanta Pal at the University of Birmingham (UK), the NRAO recently took delivery of two band-stop filters made with HTS. These have an extremely sharp 50 dB notch at the frequency of the very strong Sirius and XM satellite signals at 2.33 GHz with very low attenuation below 2.31 and above 2.35 GHz as shown in Figure 3. These filters have been installed after the cooled LNAs in the GBT S-band receiver with excellent results for wideband pulsar observations as shown in Figure 4. The top portion of this frequency range was unusable before the filters were added because the satellite signals overloaded the correlator digitizer and the front-end itself when the telescope was pointed in the general direction of the satellites. Roger Norrod specified the requirements of the HTS filter and measured its final characteristics as reported in Electronics Division Technical Note 210.
The extremely low loss required for the sharp filter characteristics shown in Figure 3 is not achievable with normal metal conductors. However, HTS material cannot be used for all RF components because it is currently available for microwave applications as thin films deposited on a relatively high dielectric constant substrate. Also, the RF loss of HTS rises faster with frequency than it does for normal metals (as the square instead of the square root of frequency) so both types of conductors will have their advantages in different applications. Another place where HTS shows great promise for radio astronomy is in power splitter/combiners and hybrids in the transition from waveguide to microstrip at the input to the first cooled LNA stage. Mike Stennes has designed and is now testing a number of these components in an OMT design for the EVLA and as an input circuit for balanced HFET amplifiers. Stay tuned for more HTS developments.