Gamma-Ray, Neutron, and Alpha-Particle Spectrometers (GRS, NS, and APS, respectively) were included in the payload complement of Lunar Prospector (LP). Specific objectives of the GRS were to map abundances of Fe, Ti, Th, K, Si, O, Mg, Al, and Ca to depths of 20 cm. Those of the NS were to search for water ice to depths of 100 cm near the lunar poles and to map regolith maturity. Objectives of the APS were to search for, map, and provide a measure of the time history of gaseous release events at the lunar surface. The purpose of this paper is to document the mechanical, analog electronic, digital electronic, and microprocessor designs of the suite of spectrometers, present a representative sample of the calibrated response functions of all sensors, and document the operation of all three LP spectrometers in sufficient detail as to enable the full knowledgeable use of all data products that were archived in the Planetary Data System for future use by the planetary-science community.
 The prime focus of the Lunar Prospector (LP) Mission was to address a wide range of lunar science goals focused on understanding the origin, evolution and current state of the Moon and, by extension, of the terrestrial planets. These goals require determination of the bulk composition of the crust and its implication for the bulk composition of the Moon, the distribution and compositional variations of mare and highland petrological units, the location and extent of possible polar ice deposits and other lunar resources, and the magnitude of gas release from surface vents into the lunar atmosphere including a determination of the location of these vents. Lunar Prospector was launched in January 1998 and was fully operational in a spin-stabilized configuration, in a near-polar, near-circular orbit having an average altitude of 100 km, between 17 January 1998 and 25 October 1998. The spin axis of the LP spacecraft was flipped from pointing to the north-ecliptic to the south-ecliptic pole and its altitude was reduced to an average of 30 km in late October 1998. The mission was ended by crashing the spacecraft into the Moon near the south pole on 31 July 1999. For the most part, the Gamma-Ray (GRS), Neutron (NS), and Alpha-Particle (APS) Spectrometers performed as designed throughout the mission and all but one of the LP spectrometer scientific goals were achieved during the next three years of intensive analyses of the data. The one exception resulted from a failure of one of the sunlight blocking foils of the APS, which developed a light leak during launch. This failure, in turn, introduced noise into the APS front-end electronics chain. The data-reduction procedures for the neutron and gamma-ray data are detailed in accompanying papers [Maurice et al., 2004; Lawrence et al., 2004]. All data products described in the accompanying papers have been submitted to the Planetary Data System at Washington University, St. Louis.
 A brief description of the Gamma-Ray and Neutron Spectrometer experiments has been given by Feldman et al. . The purpose of this paper is to provide a more complete documentation of these two spectrometers as well as of the APS. In section 2 we briefly review the scientific goals of each of the three LP spectrometers. Section 3 provides detailed descriptions of the gamma-ray, neutron, and alpha-particle sensors. In section 3 we (1) translate the scientific goals into measurement requirements, (2) show the mechanical design of the sensors, (3) present an overview of the analog electronics, and (4) present the design of the Spectrometer Electronics Subsystem (SES), which contains the power supplies, data handling and formatting functions, and the spacecraft command interface. The results of some of the sensor calibrations conducted before launch, and representative performance checks conducted after launch, are given in section 4, and a summary of achievements of the Lunar Prospector spectrometers are given in section 5.
2. Scientific Goals of the Three Spectrometers
2.1. Crustal Composition, Volcanism, and Polar Ice
 The bulk composition of the lunar crust is important for estimating the composition of the Moon as a whole, a fundamental constraint on the origin of the Earth-Moon system. Specifically, a determination of the surface concentrations of key elements, when combined with models of lunar differentiation, imposes significant new constraints on the lunar bulk composition.
 Additionally, a determination of the variability of lunar surface composition significantly improves our understanding of the evolution of the highland crust as well as of the duration and extent of basaltic volcanism. For example, one important component of lunar rocks whose abundance was accurately mapped is KREEP (K-potassium, Rare Earth Elements, and Phosphorus). KREEP is enriched in incompatible elements and therefore probably represents the last crystallization product of a putative lunar magma ocean [Warren and Wasson, 1979]. A prime goal of the LP Gamma-Ray Spectrometer (GRS) was to provide evidence that KREEP may have contaminated most lunar magmas as they made their way to the surface from sources in the mantle or lower crust [Binder, 1982; Jolliff et al., 2000]. Major basin excavations and lateral variations of crustal composition may also have contributed substantially to the distribution of KREEP at the surface [Lawrence et al., 1999, 2000]. Thus KREEP is a tracer for understanding the evolution of the crust by volcanism and impact excavation/deposition.
 Other specific objectives of the GRS, when combined with the Neutron Spectrometer (NS), include (1) identifying and delineating basaltic units in the maria; (2) determining the composition of ancient or “cryptic” mare units found in the highlands, and searching for more of these units using mainly the Fe data [Lawrence et al., 2002]; (3) identifying and delineating highland petrological units; (4) searching for anomalous areas with unusual elemental compositions that might help identify specific events and processes that helped shape lunar evolution and/or have resource potential; and (5) determining the solar wind implanted H concentration to determine the degree of soil maturity and the distribution of these potential lunar resources.
 Finally, maps of H using the NS determine if significant quantities of water ice exist in permanently shadowed areas near the lunar poles as conjectured by, e.g., Watson et al.  and Arnold . Although returned samples show that the Moon is essentially devoid of intrinsic water, cometary and carbonaceous chondrite meteorite impacts have brought water to the Moon during its history [Feldman et al., 2001].
2.2. Outgassing Sources of the Lunar Atmosphere
 Using radioactive radon and polonium as tracers, the Apollo 15 and 16 orbital alpha-particle experiments obtained evidence for the release of gases at several sites, especially in the Aristarchus region [Gorenstein and Bjorkholm, 1972]. Aristarchus crater had previously been studied by ground-based observers as the site of transient optical events [Middlehurst, 1977]. The Apollo 17 surface mass spectrometer showed that 40Ar is released from the lunar interior every few months, apparently in concert with some of the shallow moonquakes that are believed to be of tectonic origin [Hodges and Hoffman, 1975]. These tectonic events could be associated with very young scarps identified in the lunar highlands [Schultz, 1972; Binder and Gunga, 1985] and are believed to indicate continued global contraction. Thus one goal of the APS observations was to determine whether at least some outgassing sites correlate with locations where relatively recent tectonic activity may be occurring. Because the youngest observed scarps may indicate continued global contraction, these observations will further constrain lunar thermal history as well as sources of the Moon's surface-boundary exosphere. Finally, the APS data should provide information on the locations of potential sources of N2, CO2, and CO for lunar utilization.
 Specific objectives of the APS are to (1) determine the rate of outgassing of the lunar interior as a possible major source of the lunar atmosphere; (2) determine the distribution of outgassing sites and their correlation with young impact craters and tectonic features (see S. L. Lawson et al., Recent outgassing from the lunar surface: The Lunar Prospector Alpha Particle Spectrometer, submitted to Journal of Geophysical Research, 2004) (hereinafter referred to as Lawson et al., submitted manuscript, 2004); (3) determine the global rate of tectonic events that may be responsible for shallow moonquakes and their relation to young scarps identified in the lunar highlands; (4) determine whether the lunar tidal cycle influences the timing of the outgassing events; and (5) provide an assessment of lunar volatiles for possible resource utilization.
3.1. Suite of Three Spectrometers Aboard Lunar Prospector
 Lunar Prospector (LP) [Binder, 1998] carried an integrated suite of three spectrometers. A Gamma-Ray Spectrometer (GRS) and a Neutron Spectrometer (NS) have provided global maps of major and trace elemental composition of the lunar surface, with special emphasis on a search for polar water-ice deposits implied by the H abundance. An Alpha-Particle Spectrometer (APS) was used to determine the frequency and locations of gas-release events.
 The LP spectrometer system consists of a common electronics package, the Spectrometer Electronics Subsystem (SES) located on the spacecraft bus, a gamma-ray sensor mounted on one science boom and the neutron and alpha-particle sensor assembly mounted on a second boom. A simplified overview of the sensor components of the LP spectrometers is given in Figures 1 and 3. Photographs of the GRS and NS/APS sensors are given in Figures 2 and 4, respectively. The spatial resolution of the three spectrometers is about 200 km at the nominal LP mapping altitude of 100 km and 60 km at the low altitude orbit of 30 km.
3.2. Gamma-Ray Spectrometer
3.2.1. General Overview
 The GRS sensor consisted of a 7.1 cm diameter by 7.6-cm long bismuth germanate (BGO) crystal, placed within a 12-cm diameter by 20-cm long, well-shaped borated-plastic (BC454) anticoincidence shield (ACS), as shown in Figures 1 and 2. The scintillators were viewed by separate photomultiplier tubes, and enclosed within a cylindrical, graphite-epoxy laminate housing. This housing material was chosen to maximize strength per unit weight, yet provide minimal background that could interfere with determination of lunar surface abundances. The gamma-ray energy range of the GRS extended between 0.3 and 9 MeV with a channel energy width of 17.6 keV. Its accumulation time was 32 s, which corresponds to a ground track distance of about 50 km.
 The sensor housing is wrapped with a thermostat-controlled heater, placed inside a thermal blanket that was designed to provide a stable operational environment that was no colder than −28°C for at least 1.5 years. This choice was made to improve the spectral resolution of the spectrometer by using the known property of BGO that its light output per unit energy deposited in the crystal increases with decreasing temperature [Melcher et al., 1985]. Although this increased light output is accompanied by an increase in the light emission time of the crystal, and hence by an increase in the pulse time width, concomitant widths for temperatures near −30 C are not a problem for lunar applications because useful counting rates are always less than about 10 khz. The temperature-dependent light output of the LP BGO crystal and the resultant enhancement in BGO spectral resolution with increasing light output for temperatures between +40°C and −40°C are shown in Figure 5. The linear regression of these measurements yields a 0.7% change in amplitude per degree Centigrade change in temperature at −30°C. Inspection shows that reducing the temperature from +40°C to −40°C improves the spectral resolution from 14.5% to about 10% full width at half maximum (FWHM), for the 137Cs 662 keV gamma-ray line. Measurement of the FWHM of the 0.662 MeV 137Cs gamma-ray line at a variety of temperatures between +40°C and −40°C is shown in Figure 6 and of a variety of isotopic gamma-ray sources at −30°C is shown in Figure 7. Inspection shows that the spectral resolution of the flight BGO detector was indeed limited by photon statistics because the measured FWHM follows a 1/√E law.
 Although in plan, the temperature of the GRS should have been maintained within ±2° of −28°C by the combination of a multilayered insulation (MLI) blanket and a thermostatically controlled heater element surrounding the sensor, it did not work out that way in practice. The albedo from the sunward face of the Moon overpowered the insulating capability of the MLI so that the GRS temperature rose above −28°C. This temperature peaked at the lunar pole that was reached after the portion of those orbits that passed over the sunlit side of the Moon. At maximum, the GRS temperature reached −17° when the orbital plane of LP was parallel to the vector between the Sun and Moon at the beginning of the mission. One year later, after the reflecting surface of the MLI degraded somewhat, this temperature reached as high as −11°C. The resultant temperature difference between minimum and maximum (17°) corresponded to a decrease in the BGO gain that amounted to about 12%.
 The ACS of the GRS provided two functions. The first was to tag and eliminate penetrating charged-particle events from the accepted gamma-ray spectrum. The next was to detect neutrons using the boron content of the BC454. Events due to low-energy neutrons are tagged by the coincident detection of charged particles in the BC454 with a 478 keV gamma ray in the BGO, both resulting from the 10B(n, α)7Li* reaction. Detection of such an event results from a thermal or epithermal neutron. Detection of a time-correlated pair of events signifies a fast neutron if the first of the pair is a BC454 interaction only (which corresponds to proton recoils in the BC454 if the incident radiation is a fast neutron), and the second consists of the charged-particle-recoil component of the 10B(n, α)7Li* reaction [Feldman et al., 1991a]. Both counting rates provide sensitive measures of hydrogen [Feldman et al., 1991b].
 The LP GRS is, as a function of energy, 2 to 8 times more sensitive than the Apollo GRS but several times less sensitive than a high resolution, germanium spectrometer. It has provided maps during the 1.5 year lifetime of the Lunar Prospector mission that have distinguished distinct surface compositional units of Th and K through detection of gamma rays from their radioactive decay, and of Fe, Al, O, Si, Mg, Ti and Ca through detection of gamma rays induced by neutron capture and inelastic collisions.
3.2.2. Functional Overview of Front-End Electronics
 The front-end electronics (FEE) of the GRS were designed to recognize compound events consisting of two interactions in the BC454 within 25.6 μs and to measure the elapsed time between them (the time to second pulse, TTSP). A simplified block diagram is shown in Figure 8. Pulse heights for each interaction were digitized using analog-to-digital converters (ADCs) fed by the BGO and BC454 analog electronics. Whereas the BGO ADC had 12-bit resolution for energies ranging up to 9 MeV (17.6 keV per most significant 9-bit channels), the BC454 ADC had 8-bit resolution for energies up to 2.55 MeV equivalent-electron energy (10 keV per channel).
 Only four of the possible combinations of prompt and delayed BGO and BC454 interactions were recognized by the FEE for transfer to the experiment microprocessor (located in the Spectrometer Electronics Subsystem (SES) box on the spacecraft (S/C) bus) for formatting into an output data packet. Packet accumulation times were 32 s. The four categories of pulse combinations for which data were transmitted to the ground are as follows:
 Category 1: A single BGO interaction unaccompanied by a coincident (within 100 ns) BC454 interaction (which corresponds to an accepted gamma ray).
 Category 2: A single coincident (within 100 ns) BGO and BC454 interaction (which corresponds to an interaction rejected by the ACS or to a moderated neutron).
 Category 3: A prompt BC454 interaction having energy less than about 2.55 MeV followed by a single, delayed BC454 interaction having energy less than about 640 keV (equivalent to pulse-height channel 64) and occurring within a 25.6 μs gate window beginning about 400 ns after the firing of the constant-fraction discriminator (CFD) fed by the BC454 analog electronics (which corresponds to a fast neutron).
 Category 4: A prompt BC454 interaction having energy less than about 2.55 MeV followed by a coincident (within 100 ns) BC454 and BGO delayed interaction, having energies less than about 640 keV and 1125 keV, respectively (equivalent to pulse-height channels 64 in both the BC454 and BGO ADCs), and occurring within a 25.6 μs gate window beginning about 400 ns after the firing of the CFD fed by the BC454 analog electronics (which corresponds to a fast neutron).
 A logically possible fifth category, corresponding to a prompt BC454 interaction followed either by a single BGO delayed interaction or by no delayed interaction within 25.6 μs, (1) was recognized by the FEE logics, (2) had its contribution to detector dead time (DT) recorded, but (3) did not have its digitized data (consisting of three pulse heights and the time-to second-pulse, TTSP) transferred to the microprocessor.
 Category event processing logic in the FEE and SES were programmed using field-programmable gate arrays (FPGAs) to categorize and histogram each event immediately after its front-end analog processing was completed. The definition of a first interaction was decided by the first BGO Constant Fraction Level Discriminator (CFLD) and/or the BC454 CFD trigger that was detected after the generation of an end-of-process (EOP) trigger (which denotes the finish of analysis of the last detected event). The EOP, in turn, resets a busy gate to its low state, thus arming the FEE for detection of the next event. In the case of detection of a category 1 or category 2 event (signaled by detection of either a single BGO prompt CFLD trigger or a coincident BGO CFLD and a BC454 prompt CFD trigger, respectively), the time required for a complete analysis (consisting of amplitude digitization, event categorization, and incrementation of event counters) was fixed at about 10 μs. Detection of a category 3 or category 4 event (signaled by the detection of a single prompt BC454 CFD trigger corresponding to a nonoverload BC454 interaction) followed by either a single BC454 delayed CFD trigger or a delayed coincident BC454 CFD - BGO CFLD trigger, requires a variable processing time that depends on the TTSP. For these category types, event-processing times can range up to about 36 μs. The total processing time in each 32-s accumulation interval (dead time) was recorded by summing every second positive edge from a 20 MHz clock (gated by the busy logics gate), in a 16-bit scaler. This scaler was fitted with an 8-bit prescaler to yield a digitization uncertainty of 25.6 μs.
 A “truth table” that allows visualization of the total number of different possibilities of GRS event category types is given in Table 1. Information that was digitized by the front-end electronics for transfer as data words to the SES data acquisition electronics for packetization into an output telemetry format is summarized in Appendix A.
Table 1. GRS Event Category Types
III, IV, V
see delayed pulse
 The repetitive cycle time of the GRS was 32 s. Its share of the Lunar Prospector telemetry downlink was 690 bps. This allocation was apportioned to events and their pulse-height spectra (PHS) in the four different categories of detected events. All pulse-height spectra were accumulated using 16-bit counting registers but were compressed to 8 bits/register after successive 32-s accumulation periods using the algorithm in Appendix B. The composition of the GRS-portion of each 32-s data packet, formatted appropriately for transfer to the spacecraft data bus, is given in Appendix C. Main components that were used to provide primary information relevant to element composition are (1) the category 1 BGO pulse-height spectrum that provided the primary information for determining elemental abundances from the radioactive decay of K, U, and Th, and neutron-induced gamma-ray emission from the major rock-forming elements, (2) the category 2 simultaneous BGOW = BC454W = 1 events that recorded the number of low-energy neutrons at orbit and hence provided some information regarding the abundance of hydrogen (see definitions in Appendix A), and (3) the prompt pulse-height spectra of Categories 3 and 4 events, which provided the energy spectrum of fast neutrons, from which we derived the average atomic mass of surface soils. Other information in the GRS component of the spectrometer data packet provided valuable information regarding the calibrated operation of the GRS and its ACS.
3.3. Neutron Spectrometer
3.3.1. General Overview
 Achievement of the NS scientific objectives requires separate measurements of thermal, epithermal, and fast neutrons that leak outward from the lunar surface. The Lunar Prospector Thermal/Epithermal Neutron Spectrometer (NS) consisted of two identical Reuter Stokes model RS-P4-1808-225 3He gas proportional counters mounted at the tip of S/C Boom 2 on a bracket that also supported the Alpha-Particle Spectrometer (APS). Each sensor was 5.7-cm in diameter, had a 20-cm long active length (a 29.0 cm total length), and was filled with 10 atmospheres of 3He. One of the sensors was wrapped with a 0.63-mm thick sheet of Cd and the other with a 0.63-mm thick sheet of Sn. Whereas the first responded only to epithermal neutrons (energies greater than about 0.4 eV), the second responded to both thermal and epithermal neutrons. The difference in counting rates between the Sn- and Cd-covered sensors yields a measure of the flux of thermal neutrons (energies between 0 and 0.4 eV in the moving frame of the spacecraft).
 The cylindrical axes of both sensors were aligned perpendicular to both the boom and the spacecraft spin axis. Both counters were mounted on a chassis shared with the APS detector, as shown in Figures 3 and 4.
 Fast neutrons were measured using the ACS of the GRS, as explained previously. The ACS was also used to provide a separate measure of escaping neutron flux distributions that encompass the thermal and epithermal energy range, which was also described in the previous GRS subsection.
 Counting rates of both 3He sensors were sufficiently large that maps of H near the surface were possible within 1 month at the poles after achieving mapping orbit, and within about 6 months elsewhere. Its sensitivity to spatially confined H deposits was enhanced by a factor of 10 by lowering the LP orbital altitude from 100 km to 30 km. In addition, the LP NS has provided maps during the 1.5 year lifetime of the LP mission of Gd+Sm through detection of anomalous reductions in the leakage fluxes of thermal neutrons.
3.3.2. Functional Overview of Front-End Electronics
 The two 3He counters of the Neutron Spectrometer were biased by separate 2 kV high voltage power supplies (HVPS), each having 256 levels of high voltage control. The high voltage power supplies were contained within the SES box mounted on the equipment shelf of the spacecraft bus. The front-end electronics consisted of a pair of charge-sensitive preamplifiers that were mounted just behind the sensors at the end of the boom, and a pair of 8-bit ADCs mounted in the SES box, which cover the energy range up to 1 MeV. Digitized pulse heights for each detected event were used to set a one-bit indicator depending on whether (1) or not (0) the pulse height was within an energy interval that brackets the 765 keV peak in the 3He pulse height spectrum stemming from the 3He(n,p)T reaction. Channels needed to define the pulse-height windows for the neutron counters (HECDW, HESNW) were set by ground command. This indicator was, in turn, used to increment a 16-bit counter for each sensor, whose counts were related (through on-board calibration), to the number of detected epithermal (EPI) and combined thermal plus epithermal (TPE) neutrons. These counters were read out and compressed to 8 bits using Appendix B [see also Lawrence et al., 2004] every 0.5 s, and then reset and restarted.
 The 8-bit addresses of each detected event were transferred to the SES data acquisition electronics where they were accumulated into two 32-channel pulse-height spectra (using FPGAs programmed to use only the 5 highest-order bits). The resultant contribution to the bit rate is summarized in the NS portion of Appendix C. Main components that were used to provide primary information regarding the H and Gd+Sm abundances of surface soils were the two 8-bit scalers that gave the epithermal and [thermal + epithermal] neutron counting rates. The pulse-height spectra from the Cd and Sn gas proportional counters provided diagnostic information regarding the proper operation of the NS as well as the signal to background ratio.
3.4. Alpha-Particle Spectrometer
3.4.1. General Overview
 The APS searched for gas release events and mapped their distribution by detecting alpha particles produced by the decay of gaseous 222Rn (an early daughter in the 238U decay series, half-life = 3.8 days), and solid 210Po (a late daughter in the 238U decay series), half-life = 138 days. The latter is present on the surface for many decades because of the 20 year half-life of its immediate parent nucleus, 210Pb.
 The APS sensor consisted of five pairs of 3 cm by 3 cm square ion-implant silicon sensors, each pair placed on one face of a cube as shown in Figures 3 and 4. Each sensor was fully depleted to a depth of 55 μ, which is sufficiently thin to reduce the proton background in the prime energy range of Rn-decay alpha-particle lines (between 4.1 MeV and 6.6 MeV) to manageable levels. They were covered by thin, aluminized (2000 Ä thick) polypropylene (48 mg/cm2 thick) foils to eliminate direct illumination by sunlight, and collimated to a 90°Field of view (FOV), full-width at half maximum (FWHM). The combination of foil thickness, detector dead layer, and electronic noise gave a spectral energy resolution of about 100 keV at 5.5 MeV. The combined FOV of the 5 faces provided about 3.5π sr coverage, the only blind spot being in the direction of the spacecraft bus (which was blocked by the body of the spacecraft).
3.4.2. Functional Overview of Front-End Electronics
 The outputs of each of the ten Si ion-implant sensors fed individual AMPTEK A250 charge sensitive preamplifiers that were mounted just behind the sensors on each of five removable faces of the sensor cube. The outputs of the two preamplifiers on each of the faces were summed and buffered sufficiently to drive a shielded twisted pair cable to the APS front-end electronics board in the SES box. Here, the analog signals from all ten sensors were summed into a single amplifier having 50 μs shaping time. Computer simulations and laboratory measurements showed that this design achieved better than 25 keV FWHM noise resolution from the summed output of each pair of ion-implant sensors. When added in quadrature to the broadening in alpha-particle energy due to penetration of the light-tight foils at angles ranging between 0° and 45° and the straggling in the front-end sensor dead layer (39 keV), a FWHM energy resolution of about 100 keV was achieved. A capability was incorporated into the FEE (through use of an FPGA) to disconnect (using an 8-bit Enable-Disable word, APSED, which could be reset by ground command) one or more of the sensor analog chains before they activate the 50 μs shaping amplifier, to prevent paralysis of the system due to a noisy sensor. This command was contained in a one-bit identifier to set the threshold for acceptance by the shaping amplifier to either 1.5 MeV or 4.06 MeV. This capability was indeed needed during the mission to prevent paralysis by a light leak that developed after launch during separation of the LP spacecraft and the second-stage rocket.
 A single bias power supply (having a single voltage level of −25 V) generated by the SES low-voltage power supply was routed to the APS sensor boom to provide bias for all 10 sensors.
 Analog signals output from the 50 μs shaping amplifier were digitized using an 8-bit ADC, spanning the nominal energy range between 1.5 MeV and 6.6 MeV (corresponding to 20 keV per channel). Counting rates in the upper half of this energy range, which covers the 222Rn information-containing alpha-particle energy lines (5.3 MeV for 210Po, 5.48 MeV for 222Rn, and 6.0 MeV for 218Po), totaled no more than 10 s−1 for the sum of all 10 sensors during quiet conditions. Such a low rate allowed data transmission to the ground using an event-mode (EM) format. One digital window discriminator (APSEWL) was used at the output of the ADC to define a one-bit identifier (APS Event Strobe) to choose events having energies in the range between APSEWL (nominally equal to 4.06 MeV) to 6.62 MeV. The total number of detected APS events in each 32 second accumulation interval was captured in the 16-bit EM counter and included in the APS portion of the telemetry packet. Information that was transferred to the microprocessor for further processing is summarized in Appendix D.
 Components of the APS data rate are summarized in the APS section of Appendix C. Main components that were used to detect the presence of Rn and two of its daughters were the event-mode data in the output telemetry packet, composed of a maximum of 294 events per 32 s measurement cycle. The pulse-height spectrum provided diagnostic information regarding the level of solar energetic particle backgrounds.
3.5. Spectrometer Electronics Subsystem
3.5.1. General Overview
 The Spectrometer Electronics Subsystem was an 80C186 microprocessor based, central electronics box located on the spacecraft bus. In addition to the microprocessor board, it contained a low voltage power supply, high voltage power supplies, data acquisition electronics and the spectrometer electronics. The SES was the primary interface to the spacecraft for distribution of power and commands to the various elements of the LP spectrometers. The LVPS, in addition to providing regulated low voltage power within the SES, provided regulated low-voltage power for distribution to all three spectrometers. The SES housed separate electronics boards for the (1) analog BGO/NS FEE, (2) analog fast neutron FEE, (3) analog APS FEE, and (4) digital logics/GRS ADC functions for all three spectrometers with digital interfaces to the SES data acquisition electronics. In addition to distributing low voltages, the SES housed two dual programmable (using 8-bit D/A converters) 2 kv high voltage power supplies for distribution to the GRS and NS for sensor biasing. The 80C186 microprocessor with its control software programmed in programmable read-only memories acquired spectrometer data from the data acquisition electronics along with digitized housekeeping monitors. The control software was synchronized to the spacecraft telemetry frame strobes to control which science and housekeeping data was to be acquired by the spacecraft. The microprocessor board contained FPGAs to implement serial command and data interfaces to the spacecraft. The data acquisition electronics were contained on 2 boards. FPGAs along with double buffered memories were used to acquire and histogram all of the spectrometer data from the FEEs and to provide a digital bus for the microprocessor to acquire the data according to the fixed telemetry format. A block diagram of the SES is shown in Figure 8.
3.5.2. Data Handling, Commands, and Control of the Spectrometers
 The classification of each event detected by the three spectrometers, as described in sections 3.2, 3.3, and 3.4, was accomplished using the three analog and one digital electronics boards housed in the SES. Digital information about this classification was used by the data acquisition electronics within the SES to construct and store histograms of the data to form pulse-height spectra. These spectra were initially stored as 16-bit words, which were compressed to 8-bit words using the algorithm in Appendix B every 32 seconds. Also stored in microprocessor memory were four 16-bit scalars generated by the GRS every 32 seconds, and one, two, and one 16-bit scalers generated by the GRS, NS, and APS every 0.5 seconds, respectively. A last data product generated by the spectrometers consisted of event-mode GRS and APS information, which were stored in fixed-length buffers in coded form that was detailed in sections 3.2 and 3.4 (see Appendix A for details). All data were placed in a single data packet and the spacecraft C&DH acquired the data packet from the SES every 32 seconds for transmission to the ground. The contents of the data packet are summarized in Appendix C, yielding a total science data rate of 920 bps.
 Operation of the three spectrometers was very simple. There was only one mode, which was on or off. However, the type and quality of the data returned from them depended on how the spectrometers were configured in the data processor. This configuration was specified through a series of five pulse and 24 serial commands. These commands are summarized in Appendix E. Note that one of these commands, APSED, could configure the APS detector so that each of the faces could be turned on or off independently. The definitions of the information content of the serial commands were given in sections 3.2, 3.3, and 3.4. These configuration data were encoded with the science and engineering data into the total spectrometer data packet.
 These science data were supplemented by engineering data that provided a continuous update of the state of the health of all three spectrometers. They consisted of temperature sensors placed on all three spectrometer sensors and inside of the SES box. They also contained monitors of two representative low voltages (+5 V and +12 V monitors in the GRS electronics), the levels of the four 2 kV high voltage power supplies, the −25 volt bias on the Si ion-implant sensors, and the total current drawn by the SES electronics box. Whereas the four temperatures were monitored directly by the spacecraft and downloaded separately in its portion of the telemetry string, the remaining eight engineering data words were included with the science data in the total spectrometer data package. Definitions for the spectrometer engineering data are summarized in Table 2.
Table 2. Spectrometer Engineering Data
Number of Bits
+ 5 V monitor
+ 12 V monitor
− 25 V bias
IELE, spectrometer current
 All the foregoing information was coded into a spectrometer data packet that was transmitted to the ground every 32 seconds. A summary of the format of these data is given in the “Lunar Prospector Science Data Interface Specification,” which is part of the Lunar Prospector level 0 archive in the NASA Planetary Data System. The total bit rate amounts to 928 bps.
4. Calibration and Representative Performance in Space
4.1. Gamma-Ray Spectrometer
 Four isotopic sources of gamma rays were used for the preflight calibration. These sources, their strengths on the day of calibration (6 November 1996), their distance to the center of the BGO crystal, and the rate of gammas incident on the crystal (area = Ad) are given in Table 3. The calibrations using the 60Co and 88Y sources were conducted with the GRS symmetry axis oriented perpendicular to the source-detector direction, and those using 137Cs and 22Na were oriented at angles spanning −90° to +90° to the perpendicular at 15° intervals.
Table 3. Gamma-Ray Sources Used for Calibration of the LPGRS on 11/6/96
Source Strength (S)
3.695 × 105 s−1
3.928 × 105 s−1
1.30 × 104 s−1
2.875 × 105 s−1
 Samples of accepted (left-hand panels) and rejected (right-hand panels) accumulations by the ACS for 60Co and 22Na at (from top to bottom) −90°, 0°, and +90° are shown (using open square symbols) in Figures 9 and 10, respectively. Note that the +90° axis was pointed toward the south pole of the Moon before spin flip. Also shown as an overlay are calculated response functions for these sources using a computer-simulated model of the Gamma-Ray Spectrometer, which was used as input to a Monte Carlo computer code written specifically for this purpose [Prettyman et al., 2002]. The fits are seen to be excellent. This model and computer code were then used to calculate the full energy-angle response function for the LP GRS, which was then input to our data reduction code used to determine absolute elemental abundances on the Moon [Lawrence et al., 2004; T. H. Prettyman et al., manuscript in preparation, 2004].
4.2. Neutron Spectrometer, 3He Gas Proportional Counters
 The cadmium-covered 3He counter (HeCd) responds only to epithermal neutrons and the tin-covered counter responds to both thermal and epithermal neutrons. Because the response functions of both bare counters were matched, the difference in their counting rates (HeSn-HeCd) provides a measure of the thermal neutron flux. Simulations of their respective efficiency as a function of incident neutron energy are shown in Figure 11.
 The general character of these response functions was demonstrated before launch by calibrating both counters at a large pile and a small pile at Los Alamos. The large pile consisted of a 1.5 m wide by 1.5 m thick by 2.5 m high rectangular block of reactor-grade graphite that was fitted with a central hole that housed an 8.1 × 105 n/s Am-Be source of fast neutrons. The hole was buffered by a cylinder of reactor-grade graphite so that most source neutrons were thermalized before they emerged into the laboratory. The whole rectangular block was clad in a 0.75 mm thick sheet of cadmium, so that when it was fully covered, only very few residual nonthermalized neutrons emerged. The calibration consisted of removing one portion of the Cd sheet on the front side of the pile and then centering the NS on the exposed graphite 50 cm from the exposed surface. Spectra were recorded for 640 seconds with both the Cd cover removed and in place. The difference in counts then gave the response of the NS to thermal neutrons with the room background removed. The results are shown in Figure 12. Inspection shows that the Cd counter recorded virtually no counts while the Sn counter gave a robust response.
 We repeated this calibration using a much smaller pile that consisted of a 36 cm wide by 36 cm deep by 46 cm high block of reactor-grade graphite. This block was completely covered by a 2.5 cm-thick veneer of polyethylene. It also contained a central hole to house a 2.5 × 105 n/s 241Am11B source of fast neutrons. Sample results for the HeSn and HeCd sensors are shown in Figure 13. The reason why the HECD response to neutrons from the minipile is a larger fraction of that of the HeSn response is that a smaller fraction of the total neutrons emerging from the minipile are fully thermalized than emerges from the big pile.
4.3. Fast Neutron Spectrometer, ACS Shield
 The fast Neutron Spectrometer (the ACS) was calibrated using unmoderated sources of fast neutrons emitted by 241Am11B and 252Cf isotopic sources. Only the ACS response to the 252Cf source will be reported here. At the time of calibration, the activity of the 252Cf source was 5.58 × 104 s−1. It was wrapped in a 0.318 cm thick sheet of Pb and placed 100 cm from the ACS along a line 90° to the symmetry axis of the ACS. After an accumulation time of 2080 s, the delayed counts spectrum associated with a delayed pulse detected between 20 and 25 μs after detection of a prompt pulse, was subtracted from the prompt spectrum associated with a prompt pulse detected between 0.4 and 5.4 μs.
 We report here only a cursory analysis of our measured counts spectrum in order to demonstrate that the ACS fast neutron detector worked according to design. Channel to equivalent-electron energy conversion, Eee, was made using the Compton edges of four gamma rays at 0.511 and 1.28 MeV from a 22Na source, 0.662 MeV from a 137Cs source, and 0.898 MeV from a 88Y source, and from the 93 keVee charged-particle recoils from the 10B(n, α)7Li* reaction. We then followed the prescription developed by Byrd and Urban , who calculated the neutron response of a 10 cm diameter by 10 cm long borated plastic scintillator to mono-energetic fast neutrons. Although their calculated response functions are broad compared to that of standard scintillator-based gamma-ray spectrometers, they are narrow compared to the very broad energy spectrum of neutrons emitted by an isotopic fission source such as 252Cf. We therefore approximate this response function by a delta function of energy and use their energy-to-light conversion function, given by [Byrd and Urban, 1994]
where Eee is the electron-equivalent light-output energy in MeV and E is the neutron energy in MeV. The known neutron energy spectrum of 252Cf,
was then converted to a counts spectrum for our calibration:
Here, ɛ(E), the energy-dependent efficiency of the ACS to fast neutrons, was taken from Figure 3.1 of Byrd and Urban . It was reduced to correct for the fact that fast-neutron events were only accepted by the GRS ACS if the 10B(n, α)7Li* reaction was detected between 0.4 and 5.4 μs after the neutron entered the ACS. The resultant energy-dependent efficiency can be well-approximated by a power law between 1 and 8 MeV, given by
Other variables in equation (3) include ΔE/Δc = ΔE, which was estimated from Figure 5.2 of Byrd and Urban  and our channel, c, to Eee calibration using the foregoing set of isotopic gamma-ray sources. The detector area, A, was chosen to be 10 × 12 = 120 cm2, ΔT = 2080 s, and R = 100 cm. Choosing Eo = 1.3 MeV [Lorch, 1973], we obtain the open diamond points in Figure 14 for the predicted Counts/E from equation (2). The measured spectrum is included in Figure 14 as the black dot points. Comparison of both the solid lines that give the best fitting exponentials shows that they closely follow one another with a constant offset. This agreement verifies our procedure and indicates that the calculated efficiency-area product for the LP ACS needs to be multiplied by the constant factor, 1.30. If this factor is applied to the area alone, then the effective area of the ACS to fast neutrons incident at right angles to the symmetry axis of the ACS is 120 × 1.30 = 156 cm2.
 Samples of histogram data averaged over a 12 hour period (6 orbits) on 10 April 1998 that illustrate the ability of the Neutron and Gamma-Ray Spectrometer sensors to uniquely identify thermal, epithermal and fast neutrons are shown in Figure 15. The response of the BGO sensor to gamma rays gated on detection of an interaction in the ACS having pulse height between channels 5 and 15 is given in Figure 15a, and the response of the ACS to the capture of a neutron by 10B gated on detection of an interaction in the BGO between channels 20 and 30 is given in Figure 15b. The isolated, clean peaks that are well above the continuum backgrounds in each of the histograms attests to the use of the coincident detection of charged-particle recoils in the ACS having an energy of 93 keVee and a gamma-ray in the BGO having an energy of 478 keV, for a unique identification of the 10B (n, α)7Li* reaction in the ACS.
 The histogram of time-to-second pulse, TTSP, for coincident interactions in the ACS is given in Figure 15c. The horizontal line is a fit to the measured counts between channels 200 and 250 (which corresponds to time differences between 20 μs and 25 μs) represents the chance coincidence counting rate, Ro. The diagonal counts and linear fit at the left to the measured spectrum of TTSP after subtraction of the chance coincident counting rate, give the probability function for capture by the 10B in the ACS, which corresponds to, A exp(−t/τ), with τ = 2.2 μs. This time constant is consistent with (NσV)−1, where N is the number density of 10B atoms in the BC454, which was loaded with 5% by mass of natural Boron, and σV = σoVo is the product of neutron absorption cross section by 10B and the speed of the neutron, which is a constant, independent of neutron energy.
 Pulse-height spectra of ACS second interactions recorded at early times, 0.4 μs < Δt < 5.4 μs, and at late times, 20 μs < Δt < 25 μs are given by the dashed and thin solid curves, respectively, in Figure 15d. The difference between these two spectra giving the response of the ACS to moderated neutrons alone is given also in Figure 15d. Note the prominent peak at about channel 10, which corresponds to detection of the charged particle recoils produced in the 10B(n, α)7Li* reaction without detection of the coincident 478 MeV gamma ray. The bump centered at about channel 25 corresponds to the coincident detection of charged-particle recoil and the Compton edge of the 478 keV gamma ray.
 Pulse-height spectra measured using the tin-covered 3He counter (HeSn) and cadmium-covered counter (HeCd) are shown in Figures 15e and 15f, respectively. The solid histograms were measured while the spacecraft was in orbit about the Moon and the dashed histograms were measured during a 12-hour period during cruise in transit between Earth and Moon. The prominent peaks in both pulse-height spectra correspond to detection of the charged-particle recoils resulting from the 3He(n, p)T reaction in the 3He gas proportional counters.
4.4. Alpha-Particle Spectrometer
 The ten Si ion-implant sensors of the APS were calibrated before launch in vacuum using 241Am (5.48 MeV), 252Cf (6.11 MeV), and 244Cm (5.80 MeV) alpha-particle sources. Both alpha-particle pulse-height spectra and angular acceptance response functions were measured during these calibrations. Due to an inadvertent oversight, the pulse-height spectrum of 241Am was also continuously monitored throughout the mission. In order to match the thickness of the light-tight foils with the thickness of the detector dead layers to obtain the best pulse-height resolution for each of the five pair of alpha-particle sensors of the APS package, the thickness of each foil was measured before assembly using an 241Am alpha-particle source in transmission. Enough of the 241Am from the source that we used sputtered onto the foils that their alpha-particle decay was detectable above background throughout the mission. Spectra from one of the APS faces perpendicular to the spin axis of the spacecraft are shown in Figure 16. The reason why these 5.48 MeV peaks are slightly broader (and also lower in energy) than those measured during laboratory calibrations is that the 241Am nuclei observed during flight were spread all over the foils and therefore were not collimated by the sensor housing. Although this happenstance provided positive confirmation that the APS detector worked well throughout the mission, it posed a difficulty for reducing the data because the 5.48 MeV 241Am peak overlaps almost exactly the 5.48 MeV peak expected from 222Rn. However, the intensity of the “calibration” peak was not high enough to prevent detection of 222Rn from the Moon because the spin-stabilized operation of LP allowed a subtraction of the peak detected when the APS sensor was pointing away from the Moon, from that detected when it was pointing toward the Moon (Lawson et al., submitted manuscript, 2004).
 The angular acceptance of one of the APS sensors to energetic alpha particles for incidence angles, ϕ, relative to the sensor normal in the spin plane of the APS is shown in Figure 17. The expected angular acceptance can be estimated from the line drawing shown in Figure 3 using the width dimension of each of the square detectors, w = 3 cm, and the height of the top of the square collimators above each of the detectors, w/2 = 1.5 cm. The resultant angular acceptance is therefore w2 [1 − 0.5 Tan(ϕ) cos (45°]2. Inspection of the measured and calculated angular acceptances normalized to unity at normal incidence shows excellent agreement.
 Our knowledge of the absolute efficiency of the APS was checked by comparing its response to a series of Solar Energetic alpha-particle events with that reported by the Energetic Proton and Alpha-Particle Monitor (EPAM) aboard the Advanced Composition Explorer [Gold et al., 1998], which was located at the inner libration point of the Sun-Earth system. A sample of time series data that covered these events is shown in Figure 18. Conversion of measured APS counts to alpha-particle flux (in units of cm−2 s−1 MeV−1 sr−1) was made using the known sensor dimensions and the collimation geometry of the sensor head. Specifically, the sensor head consisted of two Si ion-implant detectors on each of five of the six faces of the cubical sensor head at the end of one of the three detector booms. The total detector area per sensor face was 18 cm2. The collimator for each detector defined a square viewing solid angle that was limited to 90°Full width at half maximum. The solid angle of particle acceptance of each detector is therefore π cos(45°) = 2.221 sr, and the fraction of the sky not blocked by the Moon at an orbital altitude of 100 km is 0.6627. The time interval per individual counting rate sample was 32 s and the acceptance energy range of the detector spanned 2.56 MeV. Putting all of these terms together, the ratio of measured counting rates to an assumed isotropic alpha-particle flux in units of cm−2 s−1 sr−1 MeV−1 is 10,854. Inspection of Figure 18 shows that the absolute correspondence between the alpha-particle flux measured by the LP APS and EPAM is generally excellent. The one divergence occurred at the highest counting rate between day-of-year 112 and 114, when the dead time of the APS was significant.
5. Summary and Conclusions
 Lunar prospector was conceived, designed, and implemented to demonstrate that a simple and inexpensive way to derive cutting-edge information in a planetary mission is feasible. Accordingly, all three spectrometers were designed to apply existing technologies in new ways to achieve many of the current scientific goals for lunar research that have been documented by several standing NASA planetary-science committees. All three spectrometers performed sufficiently well (the GRS and NS performed perfectly and, other than a light leak in one of the APS detectors, the APS performed adequately) to achieve all of the scientific goals mentioned in section 2 of this paper. Analyses of these data resulted in 27 manuscripts published in the refereed planetary literature, and the delivery to the Planetary Data System of high-level data sets that are sufficiently complete to provide high spatial-resolution maps of the elemental composition of the lunar surface that covered all significant rock-forming elements (Fe, Ti, Al, O, Si, Mg, and Ca), two important radioactive elements (Th, and K), as well as several important trace elements (H, Gd+Sm). This achievement was ground breaking in that the LP spectrometers provided the first global overview of the elemental composition of any planetary body.
Appendix A:: Information Digitized by the GRS Front-End Electronics
 Information digitized by the GRS front-end electronics for transfer as data words to the SES for packetization into an output telemetry format consisted of subsets of (1) a 12-bit BGO ADC address, (2) an 8-bit prompt BC454 ADC address, (3) an 8-bit delayed BC454 ADC address, and (4) an 8-bit digitized time-to-second-pulse (TTSP).
 One-bit flags required to identify detected event types by the Digital Acquisition (DACQ) subsystem of the SES correspond to (1) a category 1 or 2 event (CATFLAG = 0, 1, respectively), (2) a category 3 or 4 event (CATFLAG = 0, 1, respectively), (3) a BGO interaction within the BGO pulse-height window (BGOW = 0, 1), (4) a BC454 interaction within the BC454 pulse-height window (BC454W = 0, 1), (5) an early-time delayed BC454 interaction (ET = 0, 1), and (6) a late-time delayed BC454 interaction (LT = 0, 1).
 Three strobes are required to route data from the FEE to the proper DACQ board in the histogramming subsection of the SES. These strobes correspond to (1) a category 1 or 2 event, (2) a category 3 or 4 event, and (3) a category 2 event for which BGOW = 1 and the BC454 pulse-height channel ≤64.
 The portion of this information that was used by the microprocessor to construct an output data packet from each of the four categories of interactions consisted of the following:
 Category 1: The highest order 9 bits of the 12-bit address returned by the BGO ADC.
 Category 2: A set of pulse-height addresses consisting of (1) the top 9 bits of the 12-bit BGO ADC, (2) the bottom 6 bits of the 8-bit address returned by the BC454 ADC, (3) a one-bit indicator (BGOW) reflecting whether (1) or not (0) the BGO address is contained within an energy window, nominally extending from 380 keV (channel 21) to 600 keV (channel 34), and (4) a one-bit indicator (BC454W) reflecting whether (1) or not (0) the BC454 address is contained within an energy window, nominally extending from 20 keV (channel 2) to 135 keV (channel 14).
 Category 3: A set of addresses consisting of (1) the top 4 bits of the prompt BC454 8-bit pulse height (corresponding to fast neutrons having energy less than about 6.45 MeV), (2) the bottom 6 bits of the delayed BC454 8-bit pulse height (corresponding to electrons having energy less than about 640 keV), (3) the 8-bit address corresponding to a digitization of the time difference between prompt and delayed BC454 interactions in increments of 100 ns, (4) a one-bit indicator (ET) reflecting whether (1) or not (0) the time-difference address is contained within a 5 μs-wide window occurring at early times (nominally between 0.4 μs and 5.4 μs), and (5) a one-bit indicator (LT) reflecting whether (1) or not (0) the time-difference address is contained within a 5 μs-wide window occurring at late times (nominally between 20 μs and 25 μs).
 Category 4: A set of addresses consisting of (1) the top 4 bits of the prompt BC454 8-bit pulse height (corresponding to fast neutrons having energy less than about 6.45 MeV), (2) the bottom 6 bits of the delayed BC454 8-bit pulse height (corresponding to electrons having energy less than about 640 keV), (3) the bottom 6 bits of the (high-order 9-bit) BGO 12-bit pulse height (corresponding to gamma rays having energy less than about 1.126 MeV), (4) the 8-bit address corresponding to a digitization of the time difference between prompt and delayed BC454 interactions in increments of 100 ns, (5) a one-bit indicator (ET) reflecting whether (1) or not (0) the time-difference address is contained within a 5 μs-wide window occurring at early times (nominally between 0.4 μs and 5.4 μs), and (6) a one-bit indicator (LT) reflecting whether (1) or not (0) the time-difference address is contained within a 5 μs-wide window occurring at late times (nominally between 20 μs and 25 μs).
 Additional information accumulated for all events regardless of category was contained in four 16-bit scalers that corresponded to (1) the number of category 1 and 2 events, (2) the accumulated dead time in units of 25.6 μs resulting from BC454 first interactions, (3) the number of overload BC454 pulses defined by a minimum energy of 2.55 MeV, and (4) the number of category 3 and 4 events. A fifth 16-bit scaler recorded (5) those category 2 events for which BGOW = BC454W = 1.
 Any event containing either a prompt or a delayed BC454 overload pulse was discarded before transmitting digitized data to the microprocessor, although it was counted in the third scaler as well as in one of the other three scalers depending on the category of event. Otherwise, all digital information developed by the front-end electronics to characterize events corresponding to one of the four acceptable categories just defined, was transferred to the microprocessor immediately after it is generated.
 The contents of the first three 16-bit counters were incorporated into the output data packet by the microprocessor after successive 32-s data-cycle accumulation intervals. The contents of the fourth counter were truncated at its lowest 10 bits and then packetized after successive 4 s intervals, and that of the fifth counter was packetized as an 8-bit word after successive 0.5 s intervals. After readout, all counters were reset to zero and restarted.
Appendix B:: 16-Bit to 8-Bit Compression and Decompression Algorithm
 To carry out the 16-bit to 8-bit compression, the following steps are used:
shift difference right by R, i.e., divide difference by 2R.
concatinate I and J.
Table B1. The 16-Bit to 8-Bit Compression and Decompression Parameters
Worst Case % Error
 The variables are defined as follows: N are the 16-bit values to be compressed, IJ are the 8-bit values resulting from the compression of N, I are the integer indices into table of No and R (in range of 0 to 15), No is the base range, 2R is the interval size, and J is the number of 2R intervals (in range of 0 to 15).
 To carry out the 8-bit to 16-bit decompression, the number IJ is broken up as follows:
To create the original 16-bit number, use the following equation:
Appendix C:: Data Packet Definition
 Category 1:
 PHS of BGO, giving [8 (bits/channel) × 512 (channels)]/32 (s) = 128 bps.
 Category 2:
 1. PHS of BGO, giving [8 (bits/channel) × 512 (channels)]/32 (s) = 128 bps.
 2. PHS of BGO for BC454W = 1, giving [8 (bits/channel) × 64 (channels)]/32 (s) = 16 bps.
 3. PHS of BC454 for BGOW = 1, giving [8 (bits/channel) × 64 (channels)]/32 (s) = 16 bps.
 4. Number of events having BGOW = BC454W = 1 accumulated during successive 0.5 s intervals, giving [8 (bits/scaler) × 1 (scaler)]/0.5 (s) = 16 bps.
 Categories 3 and 4:
 1. Two sets of PHS (corresponding to early and late times) of (1) prompt BC454 PHS (16 channels), (2) delayed BC454 PHS (64 channels), and (3) delayed BGO PHS (64 channels) for ET = 1 and LT = 1 separately, yielding a contribution to the GRS bit rate amounting to 2 × [(16 + 64 + 64) channels] × 8 (bits/channel)/32 = 72 bps.
 2. The first 412 individual event addresses consisting of (1) a 4-bit prompt BC454 address, (2) a 6-bit delayed BC454 address, (3) a 6-bit delayed BGO address, and (4) an 8-bit time-difference address, yielding [412 × 24]/32 = 309 bps.
 These pulse-height spectra will be supplemented by the three 16-bit scalers (1.5 bps), a readout of one 10-bit scaler every 4 s (2.5 bps), and a 32-bit ID header (1 bps) yielding [128 + 128 + 16 + 16 + 16 + 72 + 309 + 1.5 + 2.5 + 1] = 690 bps.
 1. A 32-bit ID header followed by the 128-channel pulse height spectrum for the summed output of all 10 sensors (contributing 33 bps to the APS data rate).
 2. A data train consisting of (1) a 16-bit ID header, (2) the 16-bit EM counter, and (3) a sequence of 294, 16-bit words. If the number of detected events given by the EM counter value, N, is less than 294, then only 16 × N bits of the EM data block will contain nonzero values. If N is greater than 294, then information for only the first 294 detected events will be stored in the EM data block.
 The APS data rate then totals
 The total data rate of all three spectrometers is 690 + 49 + 181 = 920 bps.
Appendix D:: Information Digitized by the APS Portion of the Front-End Electronics
 Information digitized by the APS portion of the front-end electronics for transfer as data words to the SES for packetization into an output telemetry format consisted of subsets of (1) the 8-bit address of the ADC, (2) the energy-window flag in the form of an event strobe, and (3) a 3-bit face identifier specifying which of the five faces contained the sensor that detected the event.
 After receipt of this information, the microprocessor initiated three operations:
 1. Increment one of 128 pulse-height registers given by the upper seven bits of the ADC address (regardless of which Si sensor detected the event).
 2. Increment a 16-bit counter (to account for the total number of events in each 32-s cycle period that had a detected energy above APSEWL).
 3. Create a 16-bit word for storage in an appropriate portion of the output data packet for every detected event that satisfies the energy-window requirements.
 The words in the third operation were composed of three parts: (1) 6 bits of time relative to the present 32-s cycle period (yielding a time resolution of 0.5 s, which corresponds to 36° of spin phase), (2) 3 bits to identify the face of the cube from which the event was detected, and (3) the least-significant 7 bits of the 8-bit pulse-height address giving the particle energy relative to 4.06 MeV digitized to 20 keV resolution.
 The total number of events that were accepted for storage in the output data block was limited to 294. After filling the 294 event storage limit, no more events were accepted until after the entire data block was transferred to the output APS data packet, the current contents were zeroed, and the current location pointer was reset to one.
 At the end of each 32-s cycle period, a data packet was constructed from the foregoing stored data, which was composed of three parts:
 1. A 32-bit ID header followed by the 128-channel pulse height spectrum for the summed output of all 10 sensors (contributing 33 bps to the APS data rate).
 2. A data train consisting of (1) a 16-bit ID header, (2) the 16-bit EM counter, and (3) a sequence of 294 16-bit words.
 If the number of detected events given by the EM counter value, N, is less than 294 then only 16 × N bits of the output data block contained nonzero values. If N is greater than 294, then information for only the first 294 detected events were stored in the output data block.
Appendix E:: Command Definitions
E1. Spectrometer Package
 Pulse: SES on, off
 Sensors off
 Pulse: On
 1. GRS HVPS1 Level 8 bits (High Voltage Power #1 Supply Level)
 2. GRS HVPS2 Level 8 bits (High Voltage Power #2 Supply Level)
 3. BGOWL 8 bits (BGO Low-end Energy Window)
 4. BGOWH 8 bits (BGO High-end Energy Window)
 5. BC454WL 8 bits (BC454 Low-end Energy Window)
 6. BC454WH 8 bits (BC454 High-end Energy Window)
 1. NS HVPS1 Level 8 bits (High Voltage Power #1 Supply Level)
 2. NS HVPS2 Level 8 bits (High Voltage Power #2 Supply Level)
 3. HECDWL 8 bits (Low-end Window for Cadmium Counter)
 4. HECDWH 8 bits (High-end Window for Cadmium Counter)
 5. HESNWL 8 bits (Low-end Window for Tin-covered Counter)
 6. HESNWH 8 bits (High-end Window for Tin Counter)
 7. Spare
 8. Spare
 Pulse: On
 1. APSEWL 8 bits (Low threshold for Energy Window)
 2. APSED 8 bits (Enable-Disable Sensor Indicator, and threshold for analysis by the shaping amplifier)
 3. Spare
 4. Spare
 Total: 5 Pulse Plus 24 Serial
 We wish to thank R. McMurray for much help in the implementation of the LP spectrometers, and A. Binder for leading the whole Lunar Project Mission. The work at Los Alamos was performed under the auspices of the U.S. DOE with financial support from Lockheed-Martin Missile and Space Corporation, Sunnyvale.