A triumph for #Philae and #cometlanding

The ESA comet chaser mission, Rosetta reached a climax last week. The excitement began on Wednesday morning at about 9 o’clock when the comet lander Philae began its descent to the surface of comet 67P. Unlike the frantic ‘seven minutes of terror’ that preceded the robot rover Curiosity’s arrival on Mars, Philae’s journey took place in comparative slow-motion and lasted seven hours.

It would be easy to imagine that Curiosity’s journey was a far more difficult proposition than Philae’s but the fact is that the comet lander had to deal with an entirely different situation, equally fraught with danger. The main subject of my Curiosity blog was reliability in the harsh environment of space and on the planet. The rover only took about six months to reach its destination: Rosetta and Philae have been in space for ten years, three years of which was spent in ‘hibernation’. Ten years of exposure to extreme temperatures and radiation. A few months ago Rosetta ‘woke up’ and began a complex series of maneuvers to bring it into close orbit around the comet. Finally, last week the lander was ejected and pushed on its way by Rosetta towards its target while a spinning ‘momentum wheel’ kept it the right way up (feet down!).

When Philae was designed, the engineers had no idea what the comet looked like let alone what form the surface features would take. The designers of Curiosity had a pretty good idea of the conditions on Mars from several orbiters and three other rovers, one of which, Opportunity had landed just as Rosetta was taking off. On the other hand, nobody knew much about comet 67P apart from its having a tiny gravitational pull and that it consisted of a mixture of rock and ice. Designers would have to guess at the surface conditions and the landing site could only be selected once Rosetta had arrived a few miles above the comet. The weak gravity and lack of atmosphere did simplify things however: no parachute, no heatshield and no retro-rockets were required. Getting down was easy, if slow, but staying down, that was an altogether more difficult problem that had never been tackled before.

Philae was provided with three systems to ensure that it landed without much bounce and then anchor itself so that operation of tools like the drill (SD2) wouldn’t lever it off the surface. The first, an upward firing thruster was found not to have survived the long journey. This was a major blow because its function was to hold the lander down while the second system, ‘harpoons’ were fired into the ground. Without the thruster, the reaction from harpoon firing would kick the lander upwards. This may be what caused the first big upward movement, but telemetry suggested that no firing occurred and a natural bounce may well have taken place. The size of this bounce could be the first indication that 67P is somewhat harder than a snowball. Philae moved upward and eventually came down again but in the meantime, 67P had rotated so the lander ended up in a less favorable, shadowed location. The third anchoring system consisted of ‘ice screws’ in the landing pad on each leg. These don’t seem to have worked either. It’s possible the first contact fooled the system into deploying them while it was ‘airborne’ again or maybe they tried and failed to screw into solid rock rather than the ice for which they were designed. Certainly the pile-driver-like probe MUPUS failed to get very far into the surface. This comet is very far from being as soft and fluffy as scientists believed.

The shadowed location was bad luck, but the possibility that the solar panels might not be able to deliver enough power must have been considered at the design stage. This is why most of the experiments were able to be run in the two days or so of residual battery power and a great deal of data was uploaded via Rosetta.

Those final hours around midnight (GMT) on Friday night were tense and exciting for those of us watching the Twitter feeds. Not only were there virtual feeds from Rosetta and Philae but also from the individual experiments! Observers in the control centre such as Profs Chris Lintott and Monica Grady were pitching in as well; everybody willing the batteries to hold up long enough for the last of the data to be uploaded. Finally at one point someone posted a picture of a monitor screen showing a graph of battery voltage – it had started to fall sharply. Not long after, Philae went into sleep mode and it was all over. ESA had done a grand job conveying the tension and excitement in the control room with a continuous webcast, frequent press conferences and blog posts leading to so much Twitter traffic that #cometlanding top-trended in the UK at least. Mind you, some in that control room were not really aware that they were working in a media goldfish bowl (Type ‘shirtstorm’ into Google). I reckon many people who had never heard of Philae before last week were glued to their computer screens willing it on. Who’d’ve thought that a boring space robot the size and shape of a fridge would seem so cuddly to so many people! You never know, Philae may yet live again.

I only found out last week that much of the computing power on board Philae is provided by space-qualified Harris RTX2010 microcontrollers. This is interesting (for me at any rate) because they are stack-oriented machines designed to be programmed in Forth. As usual they are 1990s technology and ought to be blown away for speed by say, a Microchip dsPIC33, but in fact they can execute up to 32 million Forth instructions/sec! There's no way my FORTHdsPIC an achieve that, even when running on the 70MHz clock dsPIC33E. The RTX2010 is also radiation-hardened thanks to its CMOS Silicon-On-Sapphire construction. Nice device, but I shudder to think how much it costs.