Inside New Horizons

Chris Hersman posing with a model of New Horizons during our interview (F. Bernardini)

Chris Hersman posing with a model of New Horizons during our interview (F. Bernardini)

One of the most exciting missions of the last few years, years that have already been pretty interesting for solar system exploration, is going to unfold with reaching its main objective in the next few days. The New Horizons deep-space probe is about to greatly expand our knowledge of our planetary neighborhood by unveiling many of the secrets that Pluto and its moons kept hiding since Clyde Tombaugh discovered it less than one century ago.

Sending a deep-space probe to Pluto required lots of ingenuity considering also the need to keep costs constrained within given limits, and also trying to reduce trip time, ensuring at the same time that all on-board systems are kept up and running until the main target is reached. But the inflexible rules of space trajectories dictates that the target can only be observed in details for only a few days, a very tiny percentage (around 0.25%) of the whole trip time, in what is known a fly-by passage through the Plutonian system. During 9 precious days most of the science objectives will have to be accomplished. The short, but dense, activity time, paired with the very long light-time interval from Pluto to the Earth (about 9 hours round-trip) makes impossible for the mission control centre to provide real-time guidance. Everything must be programmed in advance for a single shot at everything, without any possibility to repeat the unique performance. All these constraints prompted a clever design of the robotic spacecraft and this article will explore many details of it, thanks to an in depth interview that Spaceflight conducted last May with Chris Hersman, New Horizons Systems Engineer at John Hopkins University – Applied Physics Lab (JHU-APL), where the spacecraft has been designed, built and now it is operated.

Everything in New Horizon is built to withstand a decade long journey to the edge of the solar system and beyond: redundancy and autonomy are two keywords that define the main characteristics of this spacecraft design. Volume and mass and power have also been a major consideration to be managed while keeping cost within budget limits and also making maximum use of the acceleration provided by the launcher. New Horizon has been the fastest spacecraft ever built, and to reach Pluto in the record time of about ten years it also used one gravity assist from Jupiter.

A block diagram of New Horizons (JHU-APL)

A block diagram of New Horizons (JHU-APL)

Click here to download a PDF of the above image.

Propulsion
With reference to the diagram, we can see on the left side the propulsion sub-system. An hydrazine (N2H4) tank pressurized with nitrogen feeds, through a network of pipes and valves, four groups of 0.8 Newton attitude thrusters and four 4.4 Newton orbital adjustment thrusters. It is interesting to note how the main thrusters have two valves in series (both in each engine must be operated to fire it), while the attitude ones have only one valve (for a quicker reaction time), but each group of three has one latch valve in series. Therefore, also to operate one attitude thruster two valves must be operated. The main thrusters also have a common latch valve, with a parallel redundant one to avoid that a single failure hampers using all these thrusters. In practical terms, a single failure in this arrangement will still leave this sub-system operational also because of the clever positioning of each of the thrusters: overall, eight thrusters (6 attitude, 2 maneuvering) constitute a primary set, while the other eight (again 6 + 2) may be considered a redundant set. In a hydrazine thruster (a mono-propellant system) the fuel is fed to a catalyst bed that decomposes the N2H4 molecules. The decomposition produces nitrogen, oxygen and ammonia at very high temperatures and produces thrust without the need of an igniter. Each thruster temperature is monitored as well as the pressure in the main feed lines at the output of the tank and at the beginning of the feed lines network. At launch, the propellant tank contained 76 Kg of hydrazine. After the Pluto encounter phase, that is expected to use 6 Kg of propellant, 30 Kg will be left for the extended mission to a Kuiper Belt object.

Electrical power
On the right side of the propulsion system, the rest of the block diagram shows a very strong redundant layout: particularly the rightmost side which is very symmetrical. However let’s consider first the source of electrical power and its associated electronics.

For the trip to Pluto and beyond, the only possible solution is to use a RadioisotopeGiven the continuous availability of power, a system of shunt regulators is used to control the voltage of the main power bus keeping it very close to the required 30V. There are multiple shunt regulators that are called in according to load requests and this constitutes by itself a good level of redundancy. The low-voltage distribution is kept close to the optimum point of the power curve, that is with a minimum waste of heat. The effects of sudden loads are damped by a charge reservoir implemented with a Capacitor Bank.

The Power Distribution Unit contains the switches and current limiters to provide power to every sub-system and unit. Command signals for the switches come from the MIL-STD-1553 avionics interconnecting bus and from the Command Decoder that is wired through a bi-directional serial line directly to the dual-redundant Uplink Receivers that are permanently switched on. Each Receiver has a Critical Command Decoder that translates basic commands from ground into instructions for the Command Decoder that performs the switching. This arrangement, typical of most satellites, enables the possibility to re-configure units and sub-systems even in the event of major problems, and without the intervention of the on-board computer.

The power sub-system houses also the watchdog circuitry that monitors the activity of one C&DH. During the long cruise, for instance, only one C&DH is switched on, along with one Receiver and one USO (Ultra-Stable Oscillator). If the C&DH does not regularly reset the watchdog within a given time interval, the latter logic reconfigures the avionics and, for instance, re-boots everything from the redundant C&DH. While a typical solution in every spacecraft, in New Horizons it is interesting that the logic is implemented in the PDU itself which acts a little bit like a lower-level “brain” function.

Thermal control is also accomplished by means of heaters controlled by the PDU. Dedicated heaters are allocated for the catalyst beds for each thruster, while other heaters manages thermal control according to the operating mode (spinned or three-axis). Radioisotopes heaters have not been selected because of radiation concerns with the on-board electronics and also to simplify the approval cycle before launch. On the other hand, the wasted energy of the shunt regulators can be used also for thermal control, introducing an important power saving measure. Overall power available for New Horizons systems at Pluto will be around 200W and its budget must be carefully balanced.

The whole power distribution system has been developed in-house by the Applied Physics Lab using very conventional electronics, without any new technologies. This is an important tradeoff when balancing performances with reliability, the latter supported by heritage data.

Avionics
The bulk of the avionics is housed in two boxes called Integrated Electronics Module. The use of IEM-like solutions is slowly increasing in the satellites world and the compact implementation in New Horizons might as well lead the way for other missions. Using a well-known commercial backplane standard like PCI (Peripheral Cards Interconnect) most of the avionics functions can be implemented on single boards. Two dedicated DC/DC converters, one for the digital functions and one for the radio frequency ones, are implemented within an IEM.

Each IEM contains a primary Command and Data Handling (C&DH) processor, that assumes control of the MIL-STD-1553 bus. A separate Guidance and Navigation (G&N) processor, running a 25 Hz processing cycle, performs the related auxiliary tasks off-loading the main processor that has to manage all the platform, the instruments and, most importantly, the rule-based autonomy that takes control of the instrument both in cruise and during the encounter phase. Each processor has its own Boot PROM Assembly (BPA), a dedicated permanent memory containing the bootstrap software to restart the computer in case of need. Another use of a commercial technology is the I2C (Inter-Integrated Circuit) serial bus used to exchange data with Remote Input/Output (RIO) cards that expand the C&DH reach adding interfaces for collecting housekeeping data and other uses. The I2C implementation is performed directly into a telemetry ASIC (Application Specific Integrated Circuit) hosted on the C&DH card.

On the IEM backplane it is also connected the Solid State Recorder (SSR) that is used to store all the mission science data particularly during the encounter phase that is entirely performed without direct contact with Earth. The SSR is Flash EEPROM based and has a capacity of 64 Gbits organized in 16 banks of 500 Mbytes per bank. The very low-power architecture of the SSR enables to switch on only the EEPROM chips required. During the encounter phase both SSRs will be available and the most important data will be stored in both units. The SSR is also used to store compressed and packetised (in CCSDS format) data before downlink to Earth. The allocation of memory is managed with a “bookmarking” system that index data sets and permits to associate priorities. The resultant memory organization that applies during the encounter phase is as follows: each SSR will allocate 40 Gbits to raw science data and 24 Gbits to compressed/packetised data. Of the 40 Gbits allocation, 20 Gbits will be redounded (the science data will be stored twice on the two separate SSRs), while the other 20 Gbits will be used as separate banks. Therefore the total memory allocation for science during the encounter will be 20 (redundant) + 20 + 20 = 60 Gbits (i.e. 7.5 GBytes). The maximum writing speed into the SSR is 13 Mbit/sec.

Next to the SSR we find the Instrument Interface card that hosts a number of conventional RS-422 (lower speed) or LVDS (higher speed) serial interfaces that connects the C&DH to the suite of seven science instruments managed by Southwest Research Institute (SwRI), in Boulder, Colorado. SwRI led the mission proposal to its success, and is responsible for all science aspects, through dr. Alan Stern, the mission Principal Investigator, its collaborators, and the many scientists that are part of the science team. The science objectives and the science instruments are better described in the press-kits and other sources. It is worthy saying that the mission hosts one students’ instrument (the Student Dust Experiment, from University of Colorado, Boulder, was added as Educational and Public Outreach opportunity after the proposal) that despite being the last selected, it has been the first delivered and integrated on-board. It must also be noted that the LORRI (Long Range Reconnaissance Imager) camera plays a role in optical navigations to both spacecraft sides).

Communications
Still part of the IEM is the X-band communications sub-system based on one Downlink Transmitter (Tx) and one Uplink Receiver (Rx) with integrated Critical Command Decoder (CCD). The output of both transmitters (one in each IEM) are combined via a hybrid coupler and fed to two different Traveling Wave Tubes Amplifiers (TWTA), each one powered by its own power controller (EPC). Each TWTA uses 30W of DC power and produces 12W of RF power. The switching system after the TWTA permits to use both transmitter at the same type with the same antenna (as they operate on the same frequency, they are operated a different polarizations). There is one High-Gain Antenna (HGA), one Medium Gain Antenna (MGA) and two Low-Gain Antennae (LGA), the latter used for the early phases of the mission. The HGA and MGA requires precise pointing to ensure two-way communications: the HGA needs to be pointed to Earth, while the MGA is used when a coarse pointing toward the Sun is performed in not-nominal conditions. The same RF switching network feeds from the selected antenna one of two Low Noise Amplifiers (LNA) each one connected to one of the two Receivers. Each Rx is based on a digital design (not re-programmable from ground) which is a very low-power one using only 4W, an important consideration given the fact that the receivers are always switched on. The Critical Command Decoder in each receiver is used to send contingency commands to its own C&DH and, as described before, to the PDU. The 30MHz oscillator that provides not only a clock reference for the Tx and Rx, is used in the Rx to generated a 1 pps (Pulse Per Second) signal, and it is a dual-redundant Ultra Stable Oscillator (USO). Each USO can provide clock signals to both IEMs.

Communications with Earth at Pluto distances is a daunting proposition but the precious 12W generated by the transmitters’ power amplifiers guarantees a data rate of 1Kbit/s. By using two transmitters at the same time, a technique demonstrated during the long cruise and not originally considered for the mission, it will be possible to achieve 2Kbit/s, ideally halving the time to downlink all the science stored in the both SSRs after the encounter phase. The 2.1 meters HGA antenna guarantees the minimum downlink performances of 600bit/s at Pluto distance, while the MGA guarantees the possibility to send commands to the spacecraft at up to 50 AU of distance. For monitoring the spacecraft during the long cruise phase, the mission control team relies on on-board autonomy and a simple tone-signaling system that generates at regular interval a “green” tone (to indicate that everything is normal) or one 8 “red” tones, for various kind of problems. The 8 tones are generates as sub-carriers. The ranging system used to locate the spacecraft from ground relies in non-coherent Doppler tracking technologies developed by APL, with a 0.1mm/s accuracy.

Attitude and orbit control system
Not part of the IEM, but connected to the C&DH and G&C processors, are a set of attitude and orbit control sensors strategically located within the structure of New Horizons. One Sun Pulse Sensor is a unit specific to New Horizons and it is used to provide a coarse sun-positioning signal during spin operations. The Fine Sun Sensor is instead used for three-axis operations. A dual-redundant Inertial Measuring Unit (IMU) provide attitude information and lateral accelerations at a rate of 100 Hz. The IMU is based on Ring-Laser Gyro (RLG) technology. A dual-redundant Star Tracker, each one pointing in a different direction, are used to provide absolute, stellar-based, inertial attitude information. The Star Trackers, built by Galileo Avionica in Italy (now SelexES) is probably the only European contribution to this unique spacecraft. The Star Trackers themselves are pretty unique as they must be able to work in two different ways: three-axis and spinned. Usually a spacecraft is designed for a single attitude mode, but in the case of New Horizons, the spin mode is required to keep the HGA pointed at Earth during the long cruise phase, while the three-axis mode is required for all science operations. These are two completely different modes for a star tracker, similar to having two different software applications stored in it. Star trackers are critical during the encounter phase to satisfy the precise pointing requirements for science observations. They generate attitude data (absolute attitude, relative attitude and attitude rates) at a rate of 10 Hz. For attitude control New Horizons uses only the thrusters. Momentum wheels have not been selected because in the long run they always create problems.

On-board operations
During the long cruise one, C&DH has been always on, acting as Bus Controller of the MIL-STD-1553 avionics bus that interconnects most of the on-board equipments. This operating mode, loosely called “hibernation”, was not really like that also because one Receiver, and one USO, were always switched on. A transmitter and a TWTA were also called for action during the weekly report of the spacecraft status, with basic tone-signalling. For the encounter phase, both C&DH are switched on, one acting as Bus Controller, and sending commands to the other, configured as Remote Terminal, like the other units. A re-configuration is always possible with a roles reversal and/or with a complete management by a single C&DH. The role of the C&DH are decided upon their MI-STD-1553 role: the one which starts as Bus Controller is the master while the other, started as Remote Terminal, is the slave.

Data management during the encounter phase foresees the storage of priority data for a quicker downlink to ground as soon as the encounter is done. These priority data include both science data of significant relevance and outreach data that will be required for public updates. Two different compression algorithms will be used: a loss-less one, for science usage (it takes more bandwidth) and a lossy one for public outreach data. After the encounter these priority data will be preceded only by so-called “first look” data that will be used to drive the selection of subsequent, not priority, data for downlink.

The design of New Horizons is based on a spacecraft autonomy concept that the Applied Physics Lab developed with along with missions like TIMED, NEAR, Solar Probe and MESSENGER. The encounter phase autonomy is based on the concept of solving a problem and then keep going. A rule-based engine, running every second, is a technology advance from that developed for TIMED launched five years before New Horizons. Autonomy software and fault-protections schemes that takes advantage of the redundant architecture of the spacecraft, pairs with the monitoring role of the PDU and the availability of the tone-based signalling scheme to create a very robust system. Different safe states, one Earth pointing, and one Sun pointing, are also available for worst-case conditions, ensuring communications with ground in all foreseeable contingencies.

The critical part of the encounter phase last 9 days, of which 7 before the closest approach and two after that. Once loaded in Flash EEPROMs, and started, the sequence is not changed because the 4.5 hours one-way light time (OWLT) makes any real-time impossible. The sequence has been carefully crafted in great advance and it takes into account more than 250 possible contingencies. The sequence itself can be shifted in time to take into account last minute optical navigation measurements performed before the sequence started, this taking care in Pluto position uncertainties. Once started, because if the continuous attitude maneuvering required, the spacecraft will never be back in contact with Earth. These will be very tense 9 days for the mission control team, the science team, the spacecraft designers and all the public interested in this one of a kind event. The sequence, that is timed in synchronization with Pluto’s six days rotational period,  provides also some built-in alternate trajectories that may be called in action (before the start) as debris avoidance maneuvers, in case high-exposure observations performed with LORRI, identify any such threat.

Most of the encounter sequence has been established years ago and fully tested in 2013 (only final details will be inserted in the operational one). The test of the sequence has been performed not only in one of the two engineering models of the spacecraft available on ground, but also on board the spacecraft. For this test, fake ephemeris data have been loaded in the G&N processor, forcing the on-board systems to think that Pluto was much closer than in reality. In this way it has been possible to test the sequence finding the same star patterns, required by the attitude sensors, and also experience the same dynamics that will happen during the encounter phase. The only part not tested on the spacecraft has been the post-closest-approach one, the last two days, because of concerns related to pointing instruments at the Sun (some observations are to be made while the Sun is occulted by Pluto). For this test only a limited amount of data has been downloaded, just to verify milestones in the sequence.

Conclusions
The design of New Horizons may well be part of a course in spacecraft system engineering for its ingenuity and also the reliance on both commercial standards, heritage components, and in-house developed technologies. It belongs to a line of spacecraft that JHU-APL successful produced through the years, with constant advancements in technology. We wish New Horizons and its engineers and scientists the best of luck.

Acknowledgements
The author wishes to thank Mr. Hersman for finding unscheduled hours to discuss with us details on New Horizons, with no previous warning of our visit.

F. Bernardini, FBIS

Be sociable; support the BIS!