50 thoughts on “Kepler”

  1. On some satellites, they can use thrusters to compensate for losing too many reaction control wheels. Usually, this significantly shortens the satellite’s operational life due to increased propellant consumption. I don’t know if this is an option for Kepler. It’s possible the thrusters are just too coarse to maintain the high pointing accuracy Kepler needs. It’s a shame the mission won’t last as long as hoped. It was returning interesting data.

    Reaction control wheels seem to be a fairly high failure rate item. It seems like an area that should be getting more R&D.

  2. My first question is why spacecraft reaction wheels are engineered in such a way that there are any friction-capable points of contact at all? Why aren’t reaction wheel axles suspended in magnetic bearings at each end? This would seem a natural approach for any rotating machinery operating in hard vacuum. Perhaps there’s some engineering showstopper with which I’m unacquainted in terms of continuous energy requirements, but magnetic bearings are a mature technology. Why aren’t they used in something as uber-expensive and unreachable-after-the-fact as a spacecraft?

    2. Given ferrite rods to dampen rotation by interaction with ambient magnetic field, I’m thinking of rings filled with a conductive fluid driven by an MHD setup as a reaction wheel. The trick is finding a decent conductor that won’t freeze in orbit. Mercury, with a freezing point of -38.83 °C, might not stay liquid cold enough.

      Magnetic bearings are very nice, but the ones you want use superconductors, requiring a refrigeration system, something else to wear out.

      1. “Magnetic bearings are very nice, but the ones you want use superconductors, requiring a refrigeration system”

        I hear it is very cold in space….

      2. It’s been done. Hughes tried it back in the late 60’s or early 70’s. Took too much power.

    3. They’re not used for the likely reason that they “aren’t used.” In other words, it’s the Catch-22 that engineers aren’t going to use something with no flight experience, and the only way for “new” stuff to get flight experience is to fly, but they can’t fly because they have no flight experience, etc. etc. Hopefully, more frequent and cheaper access to space will also lower the risk of innovation.

    4. A flywheel on magnetic bearings makes sense. However, these devices must not just operate in orbit, they also have to survive thrust transients during launch, particularly liftoff – it perhaps requires some sort of retractable launch cradle to keep the wheels immobile during launch. This is definitely something that needs to be developed, especially since flywheels can double as energy storage.

        1. Great find, Bart! Poking around the Teldix website I find claims of their units being on almost 350 satellites over a 30-year span. Only geosync comsats get launched in numbers like that so I’m guessing that’s their main market. They claim their biggest wheels can handle up to a 7 tonne bird. Kepler only masses a bit over 1 tonne so it’s well within the envelope of the Teldix product line. I also note that, while Teldix is part of Rockwell-Collins, it seems to be an acquired division with home offices in Germany. One hopes that “Buy American” mandates and/or Not-Invented-Here chauvinism by NASA and/or Ball Aerospace played no role in frictionless reaction wheels being passed over for inclusion in the Kepler design simply because they were from an offshore source, but I confess to dark suspicions on this point.

        2. I wouldn’t be too sure. With at least 1,000 of these puppies manufactured, that counts as Model-T-level mass production by aerospace standards. Be nice to see their price list and compare it to whatever was bid for the bits that actually went into Kepler though.

          1. I doubt the “30-year span” is for magnetic bearing wheels in particular. Teldix has been making wheels of many varieties for that long.

            A colleague informs me there are stability issues relating to magnetic bearings which have been a disincentive in their use. My take is that there are rarely silver bullets. Every technology has its advantages and disadvantages. I expect there are reasons that magnetic bearings have not been used extensively in space.

  3. Even if the mission is unfortunately cut short, it was still a stunning success. Hopefully Kepler will pave the way for a much more capable system that can do a full sky survey.

    As an aside, I wonder if it makes more sense to have robotic observatories planted near the north and south poles of the moon rather than free flying satellites. For these types of systems it would probably be nice to be firmly attached to a nice big rock which moves in precise and predictable ways.

    1. The core capability of Kepler-like observatories is the ability to observe a field of view of stars continuously, every second, for years. That’s diametrically opposed to something like an all-sky survey (though other types of planet detection are amenable to that sort of thing).

      1. That’s why I suggested the lunar poles. Each hemisphere is continuously overhead for observations for as long as you care to look.

  4. By the way, if we had the capability to send a manned mission to do a repair of a spacecraft that was in a heliocentric orbit tens of millions of kilometers away from Earth and to do so cost effectively then we wouldn’t need to. Because such capabilities imply that we would simply have a huge fleet of Kepler-like spacecraft, and launching new ones would cost a tiny fraction of what it does today.

    1. You’re making what I like to call the “space program fallacy”. Having a crew fix Kepler could indeed be cheaper than launching a new one, so long as fixing Kepler is not the only reason why the crew is out there.

      1. Hardly. I’m saying that if we lived in a world where a manned interplanetary trip of repair men were commonplace and supremely inexpensive then merely launching a new one is likely to be even less expensive. Maybe under some conditions the repair would be cheaper, but I’d imagine an unmanned retrieval would be cheaper yet. In any event “deep space capability” on its own is orthogonal to being able to cost-effectively repair or replace a spacecraft like Kepler.

        1. It’s an interesting thought; certainly having a whole constellation of Kepleroids already deployed would both simplify the response when one fails, and eliminate some of the urgency, if any, for replacement — the failed spacecraft’s workload could simply be redistributed until its owners got around to putting up another.

    2. How fast is the star field moving when you’re not on a planet? I would assume the scope would be giving the same slow if any rotation. In which case, wouldn’t a very small puff of gas be enough to make adjustments?

      I have no idea, but would be fascinated to learn from those that do.

      3. Remember the requirements: to keep a pixel accurate field-of-view continuously every second for years (tens of millions of seconds). That’s a difficult trick with any control scheme. Reaction wheels are the best option because they offer very fine-grain control which won’t upset the gathering of science data. You can use propulsive control as well but you’d want something very, very light like cold-gas thrusters, and even then it would probably disrupt science observations anyway (so why bother, if it doesn’t remove the need for reaction wheels/CMGs).

        The major torque on the craft comes from photon pressure and the solar wind as well. Some sort of electrophoretic panels which made it possible to adjust the induced torque from photon pressure could potentially solve the problem but it would be a considerable amount of engineering to make it work well.

    3. Why would you assume the repairs would need to be done via manned spacecraft? Sending a spacecraft with robots to do the work could be done much cheaper as Robin mentions.

      1. You want to do robot replacement of a part that wasn’t designed to be replaced on a satellite that is 3.5 light minutes from Earth?

        1. You’d be surprised at what is currently being worked and developed in just that field.

          1. Possibly. I remember back when the shuttle used to do repairs and the problems they had capturing satellites that were designed for capture. It’s not going to happen so I guess there’s no way for you to convince me.

          2. You’d be surprised at what is currently being worked and developed in just that field.

            Even so, I think your back to the fallacy Robin and Trent discuss. If you’re talking a new satellite to perform a robotic repair that far away and be successful; then you might as well just launch another Kepler. I’m nearly positive it would be cheaper, since you would need: the same booster power to get to Kepler, a guidance gyro better than what Kepler had for replacement, the ability to work with Kepler and maintain attitude for both masses, and then either hope all of this worked perfectly for the price, or pay for lots of testing to prove it or make multiple repair attempts.

            That’s pricey. Just make another big, precise mirror, a large enclosed volume of nothing, and a better control gyros; use the launcher and booster you would have spent on the repair satellite.

    1. Not likely because it takes 3 wheels to have 3 axis pointing control. I also don’t know if Kepler’s sensors are good for the NEO search mission.

      They might be able to use short thruster pulses to maintain control on the 3rd axis but that tends to use a lot of propellant. This Aviation Week article discusses the options:

      Sobeck said there remains a chance the wheel can be restarted by applying more torque, or trying to run it backward. Similar efforts may be made with the other failed wheel, he said. Controllers also may be able to operate the spacecraft in a less-accurate mode, using its onboard fuel to counteract the effects of solar wind on its solar arrays. There is enough fuel on board for several months of operation using the thrusters alone to maintain attitude, and the less-accurate approach could keep the spacecraft arrays pointed at the Sun for “years,” Sobeck said.

      The mission’s engineering team will work on ways to keep using the spacecraft in the weeks ahead, as the science team continues to analyze the data already collected. The mission is spending about $20 million a year, and isn’t due for another review until the spring of 2014, according to Paul Hertz, the astrophysics director at NASA headquarters.

  6. Judging from some of the Ball Aerospace photos, I don’t think the reaction wheels are very accessible, so even if we had the capability to get to the satellite, it doesn’t look like it was designed to be as serviceable as Hubble.

    Since it was already at the end of its initial life (launched in 2009 for a four-year mission) and was producing great results, it would probably make sense to launch a more capable version.

    1. Very few satellites were designed for on-orbit maintenance. About the best you could hope for on a vehicle like Kepler would be to attach an attitude control system to the base of the satellite and use that to control the pointing. Such a system doesn’t currently exist and designing one wouldn’t be easy. Unless you had a manned system to fly it there (which we don’t and won’t have for years), you’d need for the unit to fly itself to the Kepler, rendezvous and attach itself. It’d need its own TT&C, attitude control and propulsion systems. It’d have to be able to operate with the new CG. Counting R&D and launch costs, it’d cost several hundred million dollars and take many months/years to get built and launched using normal NASA procurement. You’d have to keep Kepler in safe mode until the unit arrives.

      1. The unit you describe would have utility beyond the Kepler mission. You’ve basically described a small space tug.

        2. While many of the requirements are the same as a space tug, this particular application would need much more precise attitude control. However, once you’ve designed a unit suitable for the Kepler mission, there’s no reason why you couldn’t also use it for other things including a space tug. In other words, your average space tug isn’t likely to be able to do this job but this system is more than adequate to be a tug.

    2. I could see doing that, but I wondered about closing the control loop between Kepler’s fine pointing system (I assume it uses an optical pickoff following a guide star), and the real-time signals for the external reaction wheel. The loop could perhaps be closed by having Kepler transmit the information (if possible) to the ground, then have the ground relay it back up to the new reaction wheel/sidecar satellite, but that would inevitably include a large transmission lag. Of course with smoothing, predictive algorithms, and slow tweaks, it should probably work okay, but as you said, such a mission probably wouldn’t cost much less than Kepler itself.

      In future missions, if there is any remote possibility of maintenance, it would probably be wise to include an external jack that carries the required control signals, or a way to dock a dedicated control satellite to the old one (if the old one has really heavy and expensive optics) and have the new one completely take over as the attitude control and power system.

      1. “In future missions, if there is any remote possibility of maintenance, it would probably be wise to include an external jack that carries the required control signals, or a way to dock a dedicated control satellite to the old one…”

        Or build it that way to begin with…meaning the control is done by a docked module to begin with.

      2. We’re talking minutes of delay, increasing with age of the mission. The bandwidth of such a control would be so low that, even in the quiescent environment, I don’t think it would do much good.

    3. Yeah, build and deploy a better Kepler Mk 2. The reaction wheels aren’t the only problem this bird has. Part of its CCD sensor array is also kaput and the signal-to-noise ratio of the thing is appreciably worse than intended, at least in part due to noise originating from insufficiently shielded on-board sources. This results in it requiring data to be taken for several times as long as it otherwise would to be sure one has a real planetary transit observation and not just a bunch of coincidental noise spikes. Bird was supposed to provide data with 4-sigma confidence, but only manages a bit over 2-sigma in service. Mk2 Kepler needs a much faster data link also. That means bigger solar arrays and a bigger antenna. I say start over with a clean sheet of paper.

      1. Orbital is adamant that the mission went well beyond its required lifetime of 3.5 years, and was hugely successful.

    1. It’d be a pretty huge fee, easily more than the value of the satellite. Kepler was not designed for on-orbit maintenance or capture, and it’s 40 million miles away.

      1. I was kidding, but at least they might be able to do it. NASA’s got nothin’.

        1. Sorry, hard for me to tell sometimes. Anyway, a rendezvous requires an earth escape orbit and about a year travel (depending how much delta V you’re willing to spend getting there).

  8. Are reaction wheels so expensive and/or complex that you can’t include multiple backups? You need 3, so Kepler had 4. IIRC the IIS has 4 as well, but only “needs” 2 or 3.

    Since these things seem to fail faster than we’d like, would it be feasible to add 5 instead of 4? (I realize you can get silly quickly by saying “if 5, why not 6,” but let’s not go there.)

    2. The ISS doesn’t have reaction wheels, it uses control moment gyros (gyrodines). A reaction wheel is just that, a wheel, it exerts attitude control via spinning the wheel one direction or the other (or speeding up or slowing down the wheel, as applicable). Reaction wheels become “saturated” when they are spinning at the maximum rate in one direction. A CMG is always spinning at high RPMs, to exert attitude control the axis of the CMG is moved, which exerts a gyroscopic torque in response. A set of CMGs can run into a “singularity” when the axes of too many CMGs are too parallel to each other, and it’s no longer possible to apply torque in a certain direction (this is the same phenomenon as gimbal lock for sensing gyros). The disadvantage of CMGs is that they generally require desaturation more often, the advantage is that they can generate a lot of torque using far less power than you could with a reaction wheel.

  9. What should be done is a follow up mission. Oh and NASA please point the telescope to nearby stars next time. It is interesting to map extra-solar planets in general but why these missions keep not targeting the nearby stars boggles the mind.

Comments are closed.