diodes in conjunction with these curves and conversion of the time-of-flight information to energy, it was thus possible to determine the neutron-energy spectrum. The calibration of the gamma-ray detectors consisted of determining, in one case, the charge liberated by a photodiode due to the scintillations of a small crystal of cesium todide (Cal) and, in the other case, the change in light transmission of a crystal of potassium bromide (KBr). The Cal detector was calibrated by its exposure to a calibrated gamma-radiation field produced by a Co™ source. The output currents of the photodiode were measured with a vibrating-reed electrometer. The CsI detectors were also exposed to '4-ysec bursts of X rays from the NRL beta- tron. The charge produced by the photodiode as a result of an X ray burst striking the crystal was measured. From the known radiation dose in each burst, it was possible to obtain a value of the photodiode current for a given radiation rate. It was found that the calibration figure in coulombs per roentgen for the Co™ source was about 32 percent higher than that for the betatron. This can be understood when it is realized that a larger percentage of incident gamma energy is deposited in a small crystal when the energy ia low. The X rays from the betatron were much more energetic than those from the Co®. Since the energy spectrum of the Co” more nearly approximated that expected from the device, the Co™ results were weighted more heavily. The calibration figure for the tube and crystal used here was 0.7 x 107" coulombs/r. The potassium bromide detector was also calibrated by exposure of the crystal to the Co™ source. The output current of the photodiode was measured as a function of time. From the known radiation rate and the photodiode current, the light transmission (defined ag the ratio of photodiode currents before and after exposure) of the crystals could be ex- pressed as a function of total gamma dose. tion of dosage is given in Migure 2.2. The curve of transmission (I/I,) as a func- 2.2 DATA ENCODER A block diagram of this section of the equipment is shown in Figure 2.3. Each detector output controlled a coding circuit consisting of a converter, or variable-frequency pulse generator, followed by a single-stage binary scaler. Each such converter had a repetition rate range of about 1,000 to 100,000 pulses/sec, and could be adjusted to free run at any repetition rate in this range. A positive signal increased the frequency. Both Lil detector signals had logarithmic load resistors. The so-called Log-R circuit is a nonlinear device for compressing a wide range of signal currents into a relatively narrow range of voltages, in a roughly logarithmic manner. It increases the probability that detector signals will not dirve the succeeding circuitry beyond its dynamic range. They are used in the Li and Lil detector channels. A simplified circuit of a Log-R is shown in Figure 2.4. As detector current I rises from zero, the only conducting path is through R,, the other branches being opened by the reverse-biased diodes. The slope of the V-I relation is thus R until V = V;, when VT1 becomes conducting. Further increase in I occurs with the function slope of R, and R, in parallel, V rising more slowly with I than before. Eventually, all diode paths are conducting and the function slope is that of R,;, R,, Ry, and R, in parallel. The design objectives were an approximate log characteristic for 0.001 s I s 100 ma, circuit resistances large compared with diode resistances, bias voltages large compared to 1 volt, a rise time of less than 10 ,sec, anda simple circuit. These objectives can 17