The Kepler spacecraft is in the final moments of its life. NASA isn’t quite sure when they’ll say their last goodbye to the space telescope which has confirmed the existence of thousands of exoplanets since its launch in 2009, but most estimates give it a few months at best. The prognosis is simple: she’s out of gas. Without propellant for its thrusters, Kepler can’t orient itself, and that means it can’t point its antenna to Earth to communicate.
Now far as spacecraft failures go, propellant depletion isn’t exactly unexpected. After all, it can’t pull into the nearest service station to top off the tanks. What makes the fact that Kepler will finally have to cease operations for such a mundane reason interesting is that the roughly $600 million dollar space telescope has already “died” once before. Back in 2013, NASA announced Kepler was irreparably damaged following a series of critical system failures that had started the previous year.
But thanks to what was perhaps some of the best last-ditch effort hacking NASA has done since they brought the crew of Apollo 13 home safely, a novel way of getting the spacecraft back under control was implemented. While it was never quite the same, Kepler was able to continue on with modified mission parameters and to date has delivered so much raw data that scientists will be analyzing it for years to come. Not bad for a dead bird.
Before Kepler goes dark for good, let’s take a look at how NASA managed to resurrect this planet hunting space telescope and greatly expand our knowledge of the planets in our galaxy.
A Steady Gaze
To understand the problems Kepler ran into, it’s important to understand how Kepler searches for planets. The telescope watches a section of the sky and carefully notes the dimming and flickering of individual stars. While on Earth the stars appear to twinkle due to refraction as the light passes through our atmosphere, in deep space the light from stars should be constant unless it’s physically blocked by something. Operating under this principle, Kepler looks for disruption in the light from a star which can indicate that there’s a planet in orbit around it. With careful observation it’s possible to determine the size and number of planets around each star; allowing us to virtually image distant solar systems.
As you might expect, for this to work Kepler must be able to control its orientation in space very carefully. The stars need to remain relatively stationary from the perspective of the telescope to minimize false positives. As anyone who does astrophotography here on Earth can tell you, there are ways to compensate for drift and noise to get clearer images of the sky. But for the best results the camera really needs to be locked onto the stars as closely as possible.
To maintain its orientation, Kepler was outfitted with thrusters and four reaction wheels: flywheels that are used to store angular momentum and apply torque on the spacecraft. Thrusters are ideal when large changes to the spacecraft’s orientation are required, with the reaction wheels reserved for small and precise adjustments. Unfortunately, in 2012 one of Kepler’s reaction wheels started having problems. This left three remaining, which was enough to continue on with the mission, but in 2013 another wheel shut down. With only two functioning reaction wheels the spacecraft was unable to precisely align itself during observations, effectively ending its original mission.
At this point the mission had already accomplished its scientific goals. Even if Kepler never looked at another star again it would have still been a huge success. But as the spacecraft was still largely functional, NASA started looking for a way to utilize it with a revised scientific mission the agency referred to as K2.
Perfect Balance
NASA realized that with only half of the reaction wheels operational, there was no way for Kepler to equally apply torque in all dimensions. Trying to use the two remaining wheels would simply cause the spacecraft to tumble. What NASA needed was a way to apply some sort of pressure on the craft which the remaining reaction wheels could push against. The solution came from a rather surprising place: the sun.
The fix NASA came up with exploits the fact that photons striking the spacecraft exert a constant, if slight, force. By carefully maneuvering Kepler to face the sun in the proper orientation, the two remaining reaction wheels can be used to apply torque in opposition of the solar pressure. Once equilibrium is reached, the spacecraft is balanced well enough that it can continue making observations.
It’s not a perfect solution. Positioning Kepler with this method consumes more thruster propellant than would normally be required, and there’s been a considerable drop in sensitivity due to the fact that this careful balancing act isn’t quite as stable as when orientation was being controlled by all four reaction wheels. But even still, the Kepler K2 mission has managed to collect invaluable scientific data over the last four years from a spacecraft that many wrote off as dead.
The Final Chapter
Had all four reaction wheels remained operable, Kepler probably wouldn’t be running out of propellant right now. Ironically, the increased propellant utilization necessary to keep Kepler positioned relative to the solar wind has, in a way, hastened the end of the mission. But considering the alternative was to have shut Kepler down in 2013 when it first started tumbling through space, it was a supremely successful hack.
As it stands, NASA isn’t 100% sure how much propellant Kepler still has in the tanks. Believe it or not, there’s no way to tell with the sensors onboard. They can estimate based on how much was in the tanks when it left Earth and how many burns they’ve done, and that tells them they are getting down to the wire. But until they command the thrusters to fire and nothing happens, there’s really no way to know for sure if the tanks are dry. Accordingly, NASA is limiting thruster usage as much as possible.
When the thrusters start sputtering and NASA knows Kepler is in the final moments of its life, they will command it to point its high-gain antenna to Earth and make one last broadcast of all the data its collected before powering down forever. Kepler’s orbit is far out enough that it will never return to Earth or get close enough to anything else in the solar system to be a problem. It will likely spend the rest of eternity as a deep space monument to human ingenuity and our insatiable need to explore.
Maybe it’ll encounter another alien space probe (Tan Ru’), repair one another, and seek out perfection.
K’pler.
or it may get shot down by the klingons :P
… or smashed up with little hammers by the Vogons like a scintillating jewelled scuttling crab…
That’s some impressive lateral thinking hacking indeed. What I’d like to know is how NASA handles major updates like that, given that if you change something that would crash the onboard computer, or just make it non responsive to communications, how do they set it up so as to allow trying again? I know they do as much as possible on the ground first, but…
Does this stuff need some sort of “OK, you made it, keep the newer version” signal to the bird to make it not revert automatically after a timeout? Seems like an “interesting” remote support problem.
Particularly compared to what I’d guess the average hacker here is doing with linux in an SBC or something in an arduino you can just wipe and replace at will (even if it’s a pain – it’s very possible when the thing is on the bench).
they regularly put two computers on these things, both equally capable of controlling the spacecraft. a significant enough software problem would drop the spacecraft into various emergency modes, and sooner or later one of them would switch over to the other computer.
they also do things like run the software in simulation, and/or on a stripped-down copy of the flight hardware.
I sadly cannot find the white paper anymore but I know at one point I read a paper by Nasa on this subject.
They actually will put multiple processors on any probe/satellite (10 was recommended in the paper). Each one is a carbon copy of the other and each one allows for firmware updates. When 1 processor goes out of whack the others send it the updated firmware and the processor is “restored”. The idea being that as long as the capability to rewrite firmware exists you can still repair the chips and once that fails the processor is “voted off” from the list of approved processors and this process continues until only 2 processor exist at which point when 1 goes out of wack the processors can no longer self validate.
This problem is so heavily researched by Nasa because in space, even with radiation shielding, bits are frequently flipped within the processor. Kind of like a side channel attack but from the sun.
that sounds like an extremely aspirational sort of paper. i worked on a number of missions, with design/construction dates ranging from the 80s to 2010, and none of them had 10 computers voting on things.
maybe they did that for human spaceflight equipment, but i seriously doubt it.
The Shuttle had 5. So if one goes, you shut another down, and then you’ve still got 3 for majority voting.
This. It’s always a good idea to include redundancies in applications where something is very likely to break. SpaceX also does this with their rockets.
The Harrier Jump Jet used 9 68000 uPs in 3 boxes for its engine controllers.
For software uploads, you can use checksums to check they’re correct. Then in general use, it almost certainly has a watchdog timer. If that fails, it can go to backup mode, either point it’s antenna at Earth and await a new program, or perhaps just use a limited subset of it’s abilities. You could set it to revert to the previous program, if it’s still in memory, or to a basic recovery program.
All long-established techniques in software controlling stuff.
Here”s an idea I just came up with. Somebody like Gates, Musk, or Bezos could fund it. It’s basically what I call Kepler’s TowTruck (or lorry for our Brit friends). It essentially is a disposable “band-aid” type spacecraft that makes the trip to Kepler’s solar orbit some 9 light minutes out there trailing Earth. It uses a remote docking ring to attach to rear of Kepler. The docking frame has a sophisticated 3D printer-like stepper motor array with tethers to reorient Kepler from commands from Earth.
It doesn’t tow Kepler anywhere. It just allows for something else to aim Kepler without Kepler going “bingo” or out of fuel.Kepler doesn’t need to aim it’s com-link array toward Earth either as the TowTruck is stationary and always aimed at Earth. It repeats Kepler’s data and telemetry like a TDRSS satellite does. .The Truck could just dock the frame and leave it with a power/control tether dangling so any future Truck could dock with it. The frame would not use Kepler’s power. Also any maneuvering of the Truck would not impact Kepler’s attitude as they would not be hard-docked, only a power/control tether. Go ahead with the criticism, I don’t mind…
http://www.upl.co/uploads/kepler-towtruck1533055560.jpg
That would be great for like Hubble space telescope since fuel is going to be it’s issue to
tylerhoot – I’m just brainstorming my idea… what about inertia with my remote control docking frame? If a stepper motor tether pulls the spacecraft from one axis will the craft move? What will stop the move? An picking of the opposite stepper motor? Everything in outer space is different due to lack of gravity and such. The Kepler uses fancy inertial “wheels” to re-orient. What happens when you need to aim Kepler (or Hubble) 180° out of phase? Will my TowTruck need to pull it into alignment with its ion jets.It gets complex after thinking about it. Only NASA rocket scientists can handle this one I guess.
Yeah that is where the dellemma comes in how would attaching a device to the spacecraft affect it’s balance/ properties when it tries to move or do something look at how old Soho is and is still working ok with no gyros but it’s around the L1 Orbit
But also TESS it’s replacement is now fully in science mode now so it’s becoming a new dellemma of is it worth even trying figure out a way to keep on using keplar.
tylerhoot – I love the Keplar idea and it has a good exoplanet track record. Too bad NASA is thinking of letting it go. So much money! I say send Valkyrie (https://goo.gl/H7oDfS) out there to fix it with some sort of external pack like Matthew is suggesting. She can go out there strapped to the back of a rocket, no shuttle needed. No EVA, no oxygen, no food, water, bathroom, nothing except electricity and her payload package. Then when she’s done she could just strap herself to the Keplar and draw solar power from her own solar cells to keep her recharged. She would be the local Maytag repair woman. She would be great for Mars Mission and destruction of next N.E.O. Earth-killer too (self-destruct mode after digging into NEO – like in the movie Armageddon).
I think she’s almost ready for Mars Mission. Any recent word on her yet? .
Maybe we can ask #45’s new NASA director of SPACE FORCE. :-/
Sorry Matthew! I meant Greg E. My bad…
It might be cheaper to just build a new Kepler. If my 20 year washing machine breaks I am not going to have it fixed even though it may be cheaper. It’s 20 years old.
Too bad we can’t just send the Shuttle crew out to refuel it.
Like this, the Maytag Repair man but with some Hydrazine fuel:
https://www.nasa.gov/sites/default/files/styles/full_width_feature/public/images/258799main_STS061-098-000K_full.jpg
if our washing machine broke down we would probably fix it ourselves and its older than 20 years i think
money don’t grow on trees you know :)
Matthew Lewis – I would. But I would have to get the round’ tuit tool first… ;-)
argh, sorry Matthew Lewis I did the accidental Report comment on your washing machine comment.
ALSO Hackaday – please!!! use an icon or at least rephrase the “Report comment” link so that it doesn’t start with the letters “Re___” or at the very least have a confirm “you are reporting this post. confirm?” dialog.
Mis-clicks I think are because most other sites have RePLY on the right hand side of a post, and if out of habit you just want to reply my brain at least just sees the “Re___” and autofills Reply before I realize I just clicked “RePORT” instead.
I use the Greasemonkey add-on for Firefox, a simple script hides the “Report Comment”. If I actually want to report a comment I just disable the script (easy to do from the add-on’s UI) and reload the page. Only problem is having to remember to re-enable.
That said, for Hackaday, I always think of a toggle switch with a safety cover–require a short upward drag to move the “cover” out of the way, then click to report.
@JDX I almost did the same thing, but I resorted to change the font color to black instead, and have it change color upon mouse hover. I know that the button is there, but it’s not distracting me to click on it whenever I want to reply.
@xrror – you could do like how TacticalNinja does and use @user – in the bottom generic Reply field. To get there quickly just hold down CTRL and press END. For thread continuity you might have to use the blockquote HTML tags to copy & paste in the original message. I know too much like work… ;-)
While you are fixing the ‘Report Comment button”, check your todo list, I’m sure sumwun mentioned an edit button, once, many years ago… I’m sure you’ve probably simply forgotten about it… easily done.
/andyhull scurries off and hides before the troll rat spots him.
It really won’t matter if the sun has illuminated the telescope sensors. That will happen within 9 months of Kepler running dry, unless NASA can finagle a last orientation change so the sun won’t get it right away.
aki009 – I think just like Hubble they have that huge retractable “lens cap” to prevent that. I think if NASA is dumping it then give it to a sister agency like NRO (aka NSDC) and permanently aim it at Earth. But at what target? Any suggestions?
NRO might be able to scare up a black budget to pay for a refueling run? But we’ll never be informed… ;-/
Yeah, they call it “safing” when they put a cover over a lens that might get (unwanted) sunlight into it.
Similar to what I thought of back when the 2nd reaction wheel failed. Build a maneuvering pack to attach to the fixture which mounted it atop its launch vehicle. Command the onboard reaction wheels to shut down then alter the programming onboard Kepler to not do any orientation functions. That would be handled via changes in ground control and having the maneuvering pack take over.
If Kepler’s software can be re-written appropriately, there wouldn’t need to be any electrical interface between it and the addon.
Distant orbiting satellites like this could be built with separable maneuvering systems so that instead of replacing the entire thing when reaction wheels fail and/or fuel runs out, a launch of just part of it could be done. Pop the failed or empty stuff loose and snap on the replacement. The delivery vehicle should also be able to grab the removed part and carry it to an out of the way orbit so it won’t come near the satellite again.
A stepping stone toward this kind of service has been done with the Soft Capture and Rendezvous System mounted onto the base of the Hubble Telescope, where originally nothing was intended to be mounted. The SCRS may be used to attach a booster to raise its orbit, or for a planned de-orbit (which IMHO would be a massive waste), or for a possible non-NASA service mission to extend Hubble’s life.
Gregg Eshelman – That wheel idea is amazing. They have some very smart people at NASA. Maybe we could retask Elon Musk’s mannequin “StarMan” to drive on over there and attach the addon? At least its existence could be justified.
https://brightcove04pmdo-a.akamaihd.net/4221396001/4221396001_5727774196001_5727772953001-vs.jpg
During the space shuttle era, some LEO (low earth Orbit) satellites were
a. designed to be repairable
b. had an “end effector” to allow the satellite to be grabbed.
I found this NASA slide show (.pdf warning) which shows they’ve been thinking about satellite repair.
https://www.nasa.gov/pdf/605316main_carter_8_2_11.pdf
Ren – I noticed Kepler has little boxes on its butt that could replicate ‘end effectors’. Valkyrie (or my TT) could grab a hold of, but she would need to be held stationary by her own station keeping electric jets. She could also be TDRSS via her parabola in her head.She wouldn’t need legs just like the robot already up on ISS. Maybe Valkyrie could use my TowTruck as her torso base like R2D2 using a Star Wars X-Fighter base? Her NASA designation is already R5.
Yes some of this is tongue-in-cheek but hey maybe NASA could glean some of my wry humor to heart? :-)
http://www.upl.co/uploads/valkyrietowtruck1533130186.jpg
Ren – I breathlessly await your humorous reply…
NASA built a robot with boobs?
Gregg – Wow I just noticed that! I guess they were making more room for payload like the power supply (arguably an RTG?) Great place to mount the TDRSS comms parabola too instead of the head turret. Being only 94,000,000 miles from earth, they don’t need a huge parabola. Albeit, it takes 9 minutes for the signal to get to earth. She even has Kevlar for micrometeorites (I think). They are prepping her for Mars.
I would have thought REN would have noticed such a thing Gregg! Shame on you… :-)
BTW – Valkyrie is LOADED with video cameras front and back. Maybe even on the wrists I think.
“uuuh R5? Please initiate a 1° move right please…” – no need to say please though. Who does that with Alexa? Nobody… (/humor)
“End effector” is a robotics team for ‘the thing at the end of the arm that does the work”. I think you mean “attachment point” for an end effector to grab on to.
Seems neat, and suitably hackish, but it strikes me that it would probably be about as big a job to make something like that which would operate with suitable accuracy (as well as being able to travel out to then match orbits with Kepler anyway) as it would to just make another Kepler-like telescope. Maybe one with more reliable reaction wheels or some kind of scoop to gradually gather additional propellant gases from the ultrathin but certainly not full-vacuum regions of space around the Earth (most particularly solar wind…).
NB, Tow trucks are still tow trucks in the UK. A lorry is specifically a rigid or articulated vehicle that’s made for carrying cargo (or if you want to extend railway, and somewhat agricultural terminology, a wag(g)on, particularly if it’s steam powered). One with equipment installed on it instead, including that intended for recovering other vehicles, is a truck instead… which is also a railway term, come to think of it. But it’s probably more derived from the same root as pickup trucks.
Goods vehicles that carry two or more smaller vehicles on their backs, or a drawn trailer, are harder to pin down… they’re not clearly one thing or the other, as the smaller vehicles are technically cargo, but the platforms they ride on are more installed equipment than an actual container. So they just get called e.g. “car transporter” instead, similar to those that hold liquid or gas cargoes rather than solid being specifically “tankers”.
Mark Penrice – I like Greg’s idea to make an “add-on” wheel package and deliver it via a Shuttle Mission. If NASA or ESA cant afford it we could talk some American rich guy to pay for it. Or talk the Chinese, Japanese, or Russians to do it for us.
Or maybe we can talk the new NASA SpaceForce proposed director to convince one of his ancient friends to do it. D’uh!!! :-P
http://cdn2us.denofgeek.com/sites/denofgeekus/files/2017/09/hulu-tv-didnt-know-ancient-aliens.jpg
(Come on Ren that’s worth at least a giggle?)
+2
“(Come on Ren that’s worth at least a giggle?)”
I thought you were being serious!
B^)
Where’s that dilithium crystal when you need it?
That balancing against the solar wind / photon pressure is such a beautiful and brilliant hack that, in comparison, it seems rather weak to just abandon the sat to its fate at this point. Surely it’s not beyond the bounds of ingenuity to fire out a relatively simple craft that could latch on to Kepler and refuel it? It doesn’t even need Kepler to have a fuelling port, or to break through the hull and find the internal tanks to recharge or replace them. After all, it won’t have been made with one-way valves or the like, because it was designed with the intent of operating in what is essentially vacuum… nothing is going to force its way back into the tanks, so the propellant valves will just be simple taps. The refueller can take advantage of that; clamp its delivery hose over each thruster outlet in turn, and, in sync with ground control operating the relevant tap to fully open, pump in fresh propellant, and repeat however many times are necessary. And if all the thrusters use a common propellant tank, you only need to do that once…
And as others have said, you could build in an extra reaction wheel or two that would be in the same orientation as the failed ones once latched on in the right configuration. Kepler would be back up to original launch condition, with the dispatch of a much simpler and cheaper (and more rugged/reliable?) craft than would be needed to replace the entire scope outright.
Trust me if NASA abandons it it will be eligible for space salvage. I can think of a few nation states that would love to fly out and refuel it and make it their own. Elon Musk could probably do it right now if he could get a crew to make the trip. Or send in the robots. Japan can help Elon with that. Its only 9 light minutes away in solar orbit just behind us.
Or we can hope Bridenstine grows a scientific brain and does the right stuff type of thing. In an unrelated note: That’s if he understands NASA climate research is critically important to the world; which he doesn’t just like his Rhodes Scholar boss and inventor of “SpaceFarce” – oops I meant “SpaceFarts” – oh never mind!. :P
This is either citizen science at its best, or some of the nerdiest bike-shedding I have ever seen. In either case, well done!
“The prognosis is simple: she’s out of gas. Without propellant for its thrusters… ”
Is she a she, or is it an it?
We have not been informed as to the preferred pronoun.