The ISEE-3 Reboot Project: a dream SDR application

Puerto Rico is host to many interesting sights: lush greenery covering rolling hills as far as the eye can see, emergency first responder vehicles that always have their lights flashing even when not on call, the best crab turnovers, and the Arecibo Radio Observatory – the world’s largest single-dish radio telescope. As luck would have it, multiple Ettus Research USRP N210s are now connected to the telescope for the purpose of communicating with an all-but-forgotten 36 year old NASA space probe known as the International Cometary Explorer.

Artist's rendition of ISEE-3
Artist’s rendition of ISEE-3

The space probe was originally launched on August 12, 1978, as part of a group of three craft that were destined to reveal secrets about our solar system. Its name at launch was the International Sun/Earth Explorer 3 (ISEE-3, later renamed to ICE), and its unique purpose was to study the interaction between the solar wind and the Earth’s magnetic field. It accomplished some notable firsts, being the first space craft in a halo orbit at Earth-Sun L1 (the ‘Lagrange point‘ where the gravitational effects of the Earth and Sun cause the probe to be drawn along at the same point relative to the two large bodies). It was also the first space craft to pass through the tail of a comet (in this case Giacobini-Zinner in 1985). It hosts an impressive array of instruments for science data collection: thirteen investigations were underway in such areas as solar wind plasma, cosmic rays and gamma ray bursts.

Manoeuvres

ISEE-3’s official mission came to a close when NASA decided it was time to pull the plug – literally. Funding dried up, and the equipment that was used to communicate with it was removed from the Deep Space Network ground stations, effectively relegating the probe to the cemetery of space. Nevertheless, ISEE-3 continued on its relentless trajectory back toward Earth for its next lunar flyby (a waypoint built into its mind-boggling orbit). A testament to the longevity of NASA (and sub-contractor) engineering, the probe’s solar power and transponder systems were still functioning and its lonely RF carriers were detected once more by the DSN on September 18, 2008.

Thus the ISEE-3 Reboot Mission was born, and Software Defined Radio would make its inevitable debut as a means to rescue the old bird. Formed by Dennis Wingo (Skycorp Inc) and Keith Cowing (Space College/Spaceref) in April of this year, the mission’s aim is to evaluate the health of the probe, capture it by performing a series of thruster manoeuvres to bring it into a high Earth orbit (pending NASA approval) and re-enable the science instruments making this the first publicly-accessible space craft for STEM education. As this is a private enterprise, RocketHub was used to crowdfund the project and is therefore supported by the community at large.

RocketHub

There are two factors that impose an important time constraint on the schedule of the project: the trajectory of the probe, and the amount of fuel remaining on board. As the probe closes in on the Earth, there exists a window to fire the thrusters and adjust its course. If the opportunity is missed, the probe will not be in our neighbourhood for another 15 years. Luckily there is still some fuel left with which to perform the manoeuvre, however as the probe draws nearer, more fuel will be necessary to perform the same trajectory correction. The lunar flyby looks to occur on August 10, and thrusters should be fired in July for the mission to proceed comfortably. This has brought about a sense of urgency in assembling the necessary talent, software, hardware and radio telescope to get in touch with ISEE-3.

Arecibo's Dish

The Arecibo radio telescope provides up to roughly 75 dB of gain at S-band. Putting that in context, planetary RADAR astronomers use this to image asteroids with a dual 1 MW klystron arrangement that provides 20 TW EIRP. This gain makes it ideal to close the link with a space probe (at the time) over 15 million kilometres from the Earth. The remaining unknown was what equipment (MODEM) could be used to send commands to the probe as well as decode its telemetry in order to evaluate its health. The Reboot Project made initially contact with Matt Ettus, who forwarded the opportunity to my former-colleague John Malsbury and I. We jumped at the thought of this unique application and agreed it was the perfect situation to demonstrate the flexibility of Software Defined Radio. Within one night, John had already assembled the first implementation of the uplink modulator as a proof-of-concept flowgraph in GNU Radio. The Reboot Project then bought several N210s, shipped them, along with me (and John briefly) to Arecibo for an attempt at first contact. On the Skycorp side, the Arecibo Away Team was led by Dennis, and lead engineer Austin Epps.

USRP N210

Upon arriving at Arecibo, I knew I was in the right place when I walked outside the rear of the Visiting Science Quarters and found a dipole antenna in the backyard. This was in fact a riometer experiment, and the data acquisition board and laptop were left on my room’s table. While exploring the main facility itself, one truly finds themselves in RF heaven. From the first glimpse of the top of one of the three towers supporting the platform that is suspended above the dish, to riding the cable car up to the platform itself with the dish appearing in a slow reveal, to jogging around the perimeter of the dish, the scale of the place is incredible. This, along with stories of birds being cooked by the S-band RADAR, and only 1 dB of loss across the RF waveguide that extends from the 2.5 MW (peak power) 430 MHz klystrons next to the control room, across the cat walk, and into the Gregorian Dome, conveys the seriously large (and tiny) numbers the science conducted at Arecibo deals with. The Dome itself is a relatively recent addition in the history of the telescope, and is an amazing structure in itself. It contains three levels: the top is the transmitter floor where the klystrons are housed (as well as our power amplifier), and below that is the receiver floor that actually contains a rotating floor which brings the selected receiver or transmitter to the focal point of Dome. Underneath the receiver floor is another level where the feeds of those same receivers and transmitters can be seen. As the floor rotates, the appropriate feed is ‘selected’ for the experiment (Arecibo is half-duplex) and is pointed at the tertiary reflector, which points upward. The secondary reflector is above it on the ceiling of the Dome, and the primary is of course the enormous one outside, far below (visiting the platform is not for the acrophobic). The Dome itself is suspended from the Azimuth arm, and tracks along its length. The Azimuth arm itself rotates around the middle of the platform, and so a source can be tracked as it moves across the sky without having to move the dish itself (it is a spherical dish, which, unlike a parabolic one, means that there is no single centre of curvature).

Underneath the receiver floor the feeds are visible
Underneath the receiver floor the feeds are visible

Secondary reflector on the Dome's ceiling
Secondary reflector on the Dome’s ceiling

On the platform looking down at the Dome and the dish
On the platform looking down at the Dome and the dish

Cable car heading up to the platform, and looking up at the platform and Dome from the centre of the dish below
Cable car heading up to the platform, and looking up at the platform and Dome from the centre of the dish below

The three towers that support the platform
The three towers that support the platform

There were still some issues to resolve before we could begin transmitting to turn telemetry back on. Unfortunately ISEE-3’s uplink frequencies are outside what the existing Arecibo transmitter supports, so a custom 450 W power amplifier was ordered from Dirk Fischer, a German HAM (DK2FD), who built it in short order and delivered it via international shipping. We also needed approval from the authorities to transmit on the uplink frequencies. Then there was figuring out how to patch the USRPs into the labyrinth of RF plumbing that Arecibo allows you to connect to.

The custom-built S-band Power Amplifier
The custom-built S-band Power Amplifier

The IF patch panel
The IF patch panel

The staff were very accommodating when it came to installing and testing equipment. Dana Whitlow, the resident RF and microwave guru, was most helpful in helping set up the transmit chain. The TX USRP (with an SBX to output the raw S-band signal) was placed in one of the equipment rooms, and the RF was sent up to the Dome via an analog optical fiber link (the link has a total bandwidth of 12 GHz, luckily much more than we needed). It is a very conscious decision to keep all digital equipment out of the Dome – RFI is a very real and constant problem, both from external interferers, such as wide-area wireless Internet providers, and any local emitters such as mobile phones (there are signs to switch these off at every turn), copper Ethernet cables (fiber is used everywhere), and more generally anything that contains a digital clock (such as a laptop or a USRP).

The TX USRP installed in one of the equipment rooms
The TX USRP installed in one of the equipment rooms

The TX USRP connected to the fiber link, and calibrating the output level
The TX USRP connected to the fiber link, and calibrating the output level

For the receive side, we had two USRPs with WBX daughterboards connected into the IF patch panel at the rear of the control room. Arecibo’s RX chain is a complex beast, but in our ‘S-band wide’ configuration we had our frequency of interest (the two transponder downlink frequencies) centred on a 260 MHz IF. Phil Perilatt, who lives and breathes the telescope, and operates it for us during a pass, has been instrumental in setting this up, tracking the target and analysing data.

The RX USRPs for each transponder's downlink frequency
The RX USRPs for each transponder’s downlink frequency

Having sorted out the particulars and being allocated time on the telescope (this is very hard to come by as the schedule is almost fully booked – we would rely on free moments during the afternoon telescope maintenance period), it was finally time to track the source and find those lonely carriers. In fact we did detect them, although they were very weak – much weaker than previous detections. I had developed a GNU Radio flowgraph to perform synchronised recording of the two RX channels (the USRPs had their external references connected to the local Hydrogen Maser clock), and after running the samples through an FFT with very low RBW, the faint Doppler-shifted carrier could be seen. The signal was so weak, as it turned out, because the ephemeris that the telescope was using to track the craft was stale. Phil would run some ‘spider searches’ of the sky to improve these numbers and our C/N jumped from something almost indistinguishable to almost 50 dB (0.5 Hz RBW). Another curiosity is that the space craft is actually spin-stabilised, with a nominal spin-rate of 19.75 RPM). As it spins, so do the antennas that transmit the downlink signal back to Earth. If you view the carrier at a small RBW, you can also see the Doppler shift the spin induces on the carrier.

First detection of the downlink carrier
First detection of the downlink carrier

Downlink carrier over time
Downlink carrier over time

Improved carrier power with better pointing
Improved carrier power with better pointing

Spin evident from shift in downlink carrier
Spin evident from shift in downlink carrier

Turning telemetry back on meant we would exercise the other side of the link. Having installed the power amplifier on the transmitter floor of the Dome, and having calibrated the power levels of the fiber link, we were ready to send our photons into the aether, hopefully to be collected by the probe’s Medium Gain Antenna and correctly interpreted by the on-board decoders. It is important to note that the probe does not have a rad-hard processor – a pocket calculator has many more gates than ISEE-3. The probe could effectively be reduced to a flying collection of shift-registers and accumulators. Despite this, there is still much magic and mystery to the probe, and the use of this type of ‘large’ discrete logic has almost definitely contributed to its impressive life span.

After some experimentation with frequency offsets and switching between transponders, first contact was finally made. The two-way round-trip light time at that point was ~103 seconds, which meant that we would start the clock after the end out our last transmission and be held in unbelievable suspense until the expiration of the countdown. The pressure of having come so far (both us and the probe) was palpable. Incredibly, on the 29th of May, a baseband modulated signal originally synthesised by a GNU Radio flowgraph, was converted to RF by a USRP, left the S-band waveguide, bounced off a 305 metre dish, and was correctly decoded by the probe ~52 seconds later. Another ~51 seconds after that, almost like magic, the live unmodulated carrier on our FFT plot suddenly was surrounded by telemetry sidebands – everything had worked. As a bonus, we were able to confirm that RF was actually being transmitted by having a USRP B200 simply listening with a small antenna at the uplink frequency: when a command would ‘leave’ the dish, we could see the modulated spectrum appear on the waterfall.

Verifying that RF is leaving the dish with a USRP B200 (standing in front of the operator's station)
Verifying that RF is leaving the dish with a USRP B200 (standing in front of the operator’s station)

Waterfall of modulated downlink carrier after telemetry was re-enabled
Waterfall of modulated downlink carrier after telemetry was re-enabled

Zooming into the modulated downlink signal
Zooming into the modulated downlink signal

The moment of realisation
The moment of realisation

Celebration locally and remotely
Celebration locally and remotely

The very next day (a week later than we first scheduled), the Away Team returned State-side. We left two USRPs with the Arecibo folks who have been generous enough to allow us to continue communicating with ISEE-3 remotely. The legendary Phil Karn contributed his Viterbi decoder to the effort, which has been used to decode the recordings offline and determine the space craft is doing surprisingly well despite its age. For example, although the batteries are long dead, the solar power generation efficiency is remarkable. The spin rate has also decreased a little (~19.16 RPM from the nominal 19.75), and the theory is that it has slowed over time due to the YORP effect.

The space probe is now 7.3 million km away, and hurtling in our general direction at ~2.78km/s. Since our return, there have been other exciting developments: we have successfully tested the coherent/ranging mode of the transponders, and this has been used by the NASA Deep Space Network to range more accurately the position and distance of the probe. Also I have developed a real-time telemetry decoder that is reminiscent of the original NASA screens, and exploits the Quick-Look-In feature of the convolution code used. This has been important in our passes (or ‘supports’) since returning from Arecibo as the team prepares to send more commands, such as the first thruster firing sequence to ‘spin up’ the space craft returning it to its original rotation rate (this is important for the subsequent manoeuvres). As thruster settings are programmed in, the team can verify the status of the attitude control sub-systems before proceeding to the next step in the firing sequence.

Improved ephemeris after pointing experiments at Arecibo
Improved ephemeris after pointing experiments at Arecibo

Real-time telemetry screen showing state of the Attitude and Orbital Control System
Real-time telemetry screen showing state of the Attitude and Orbital Control System

As passes continue at Arecibo, other sites are coming online to participate in the joint effort. For instance, Morehead State University has installed their N210 and have picked up ISEE-3’s signals (they are a little weaker though as it is hard to compete with Arecibo’s dish size). Being able to simultaneously coordinate between multiple sites during a support will help overcome the transmit/receiver turn-around we currently factor into our communications schedule.

It is amazing to consider the entire project is based around sending and receiving radio waves to and from an object we have never actually seen with our own eyes. This element of intangibility adds to the mystery of the probe. We believe it is there due to the vast amount of documents the Reboot Team has found, and the changes we see on our received spectrum or decoded bit stream.

There is still much work to be done, many more commands for the probe to receive, and both engineering and hopefully science telemetry to interpret. It is also wonderful to consider that the USRP, usually associated with terrestrial communications, coupled with GNU Radio, a community-driven open source DSP framework, underpin this first-ever ‘space hack’ (naturally with the blessing of the authorities!).

UPDATE 1: the first firing of thrusters to perform the spin up manoeuvre was a success! The next (imminent) step will be to perform the actual trajectory correction manoeuvre know that we know the propulsion system works.

The ISEE-3 Reboot Mission Control Team
The ISEE-3 Reboot Mission Control Team

L-R: Austin Epps – Lead Engineer,
Jacob Gold – Systems Engineer,
Cameron Woodman – Flight Director,
Dennis Wingo – Mission Director, Project Co-lead,
Marco Colleluori – Attitude & Orbit Control Systems Engineer (and me).

UPDATE 2: We’re continuing our attempts to fire the thrusters. Here is an example of the real-time telemetry display and graphs that show the state of the space probe during the attempts:

update-accel.png

Graphs on right-hand side show accelerometer readings during thruster firing

update-xpnder-strength.png

Top-right graph shows transponder B RSSI