All of the observations in this paper were made with an InSb detector system built by the author. The preamplifier used follows the design of Hall et al. (1975): the detector (supplied by Santa Barbara Research Center) is externally biased to 0 volts, and radiation is detected by measuring the variation in generated current. The only noise source is then random thermal electron motion (Johnson noise), for which the noise current is proportional to the square root of the ratio of detector temperature to impedance.
The detector is operated at pumped nitrogen, temperatures (T ~ 60 K). Prior to pump-down, and following a technique discovered by K. Matthews at Caltech, the detector is flashed by exposing it to extremely intense 1.2 µm radiation for several minutes, The apparent effect of this bizarre procedure is to remove a surface charge layer that builds up on the protective oxide coating of the InSb crystal, thereby eliminating a spurious conduction current from the oxide to the p/n junction (K. Matthews 1976, private communication). Typical results of flashing are an increase in detector impedance from a value of 3 × 109 ohms to 1011 ohms. The system noise is thus determined not by the detector, but by the feedback resistor, whose resistance of 1010 ohms is the highest stable value that is available commercially.
At 2.2 µm, the system NEP is 10-15 watts / (Hz)1/2, and using a focal plane chopper, the system is background noise limited (in the K-band) on the Mt. Hopkins 60-inch telescope at f/10 with a 2 mm aperture. Using the 60-inch, an object having a K magnitude of 9 can be measured (at K) to a statistical level of ~ 1% in one minute of integration time. This represents about a tenfold improvement in signal-to-noise over the best previously available PbS system.