NASA To Repair Hubble By Remote Control 53
Matt_dk writes "NASA says it plans to fix the Hubble Space Telescope by remote control this week.
The Hubble stopped beaming information to Earth about two weeks ago, when a data unit on the telescope completely failed.
Scientists on Tuesday said they will bypass the failed unit and switch to a back-up system to restart the flow of information.
The computer glitch forced NASA to postpone a shuttle mission this month to repair the Hubble.
That shuttle mission has been postponed until next year."
Update - 10/15, 17:45 by SS: Readers have pointed out further details from Spaceflight Now and the NASA press release.
They're gonna need (Score:5, Funny)
Some really fresh AA batteries to make this work.
Re: (Score:1)
... or a really hot cup of tea.
MST3K (Score:3, Funny)
So, a better summary... (Score:5, Insightful)
NASA will flip a switch and kick in the backup system.
The story is pretty light on details. It reads like a 6th grader wrote it.
Re: (Score:1, Insightful)
Re:So, a better summary... (Score:5, Informative)
Actually, the part that failed has a backup, it's almost like a two-sided cassette tape, and that second side needs to be loaded into the device that is responsible for transmitting data.
The broken part right now stops any and all data transmission from Hubble. Performing the fix (flipping the tape) will restore all performance, with no degradation, other than no backup.
When the next shuttle mission launches to repair the Hubble, they will be replacing this part so that there is, again, redundancy.
It's pretty amazing that these parts are still working after...how many years?
Re:So, a better summary... (Score:5, Funny)
It's pretty amazing that these parts are still working after...how many years?
Maybe NVIDIA needs to have a chat with NASA about quality assurance.
Well, that and Tang, of course.
Re: (Score:2)
Maybe NVIDIA needs to have a chat with NASA about quality assurance.
I hope NVIDIA won't start producing O-rings any time soon.
Re: (Score:2, Interesting)
Re: (Score:1, Insightful)
I think it has to do with the level of cost of a failure, its one thing to have a system fail, but its more complicated then the trip alone costs 10+ million dollars
my guess is 5min to send the commands, 1.9weeks to diagnose what exactly failed, and how to repair it...
Re:So, a better summary... (Score:5, Insightful)
Because the point is to avoid catastrophic failure. Imagine it was something like say a solar flare, backup systems online and poof goes the backups too. Or there is some form of short circuit that'll fry the backups too. Or just figuring out what the failure state is and how to best handle it. To take an example, say a Mars rover wheel is busted. How? Is it stuck? Has it lost the drive? Can it turn? Is the wheel itself torn? Is it just a sensor malfunction? You need to be 100% sure what state Hubble was in when it failed in order to be sure to recover properly. On earth it's really easy to throw a lot of real redundancy into things, in space it's still one device more or less and you try to figure out if the right side is safe when the left side is on fire. Most anywhere else it's either functioning correctly or you'd kill it and replace it with something that does. Trying to salvage half-borked systems only happens when they're really expensive or really hard to reach, and I think the Hubble qualifies on both.
Re: (Score:2)
It's a workaround not a repair, and being careful (Score:2)
Re: (Score:2)
But what they don't know .... (Score:3, Funny)
Fixed! (Score:2, Funny)
Hubble
There, fixed that for you, NASA.
Stupid to call a car repair man to fix your stereo (Score:2, Informative)
The Shuttle program is ending. We can't just keep sending astronauts to Hubble like we could have in the 90's. Now we either do it all a
Re: (Score:3, Insightful)
Shuttle pilot: Good to go, sir
Mission specialist: Hoo-ah!
Boss: Right! You're scheduled to launch...
[Underling comes rushing in.]
Underling: Mr Houston, we've had a problem.
Boss: What sort of problem?
Underling:The AE-35 unit on the Hubble just went to 100% failure.
Boss: How long will it take to prep a replacement?
Underling: Let's see... A week to order the part... three to five weeks of testing... decontamination and cle
Finally! (Score:4, Funny)
They're going to get some use out of that old Atari joystick that's been sitting in the office!
I'm curious... (Score:5, Interesting)
A satellite is the ultimate inaccessible device running SW. Any task that goes wrong has the chance of bricking a device that cost many many millions, so they *must* practice and check all commands sent to it when things go wrong.
Do they have several mock ups?
A complete computer model of the whole thing, emulated right down to hardware and software?
How are reboot/reprogram sequences like this handled/practiced/tested?
Even at design stage I imagine failure modes are extensively analyzed and multiple redundancy built in.
My company builds stuff that goes up masts and is generally quite inaccessible and we always attempt to prove these things first, but we had fast serial communication, low level boot loaders under all the SW and if the worst comes somebody can climb the mast.
Anybody know how space tech is handled?
On a kind of related note, google for "expensive software errors" - most of the top ten are space related...
Re: (Score:2, Interesting)
Back during the Space Race, yes, they had full mock-ups, right down to the last screw. Many (probably not all) were themselves space-worthy.
Nowadays, I kinda doubt it. My hunch would be computer simulation including hardware, but its hard to cite "my gut" as a credible source.
Re: (Score:1)
Re: (Score:1, Informative)
A ton of IV&V is done on the hardware before it is sent into space, some is on the actual hardware or a mockup (utilizing either equivalents or actual orbital hardware) of the hardware. Much of what's done today is done using model based integration and test procedures. This can be done either via a complex computer program, which simulates all the hardware and software, as well as inputs, outputs, triggers, etc, or on a representative hardware suite with simulated inputs.
Re: (Score:3, Informative)
The following is shamelessly plagerised from the following two articles; NASA to reboot Hubble Space Telescope [newscientist.com] and Hubble replacement part has glitches of its own [newscientist.com]
Engineers plan to send commands to the telescope switch over to a backup computer that has not even been turned on since before the telescope arrived in orbit 18 years ago. However there's very little ageing that goes on with an unpowered component in space. It's actually a very benign storage environment.
The errors were found in Hubble's scie
Re:I'm curious... (Score:5, Informative)
A satellite is the ultimate inaccessible device running SW. Any task that goes wrong has the chance of bricking a device that cost many many millions, so they *must* practice and check all commands sent to it when things go wrong
Yes they practice everything on a mockup, but there is no chance that you can 'brick' Hubble. Proper spacecraft design means you have a completely separate system able to intercept telemetry and reprogram the main computer if an undiscovered bug happens to lock it up.
Do they have several mock ups?
Yes, both for software and a full mechanical mockup.
A complete computer model of the whole thing, emulated right down to hardware and software?
Partially, yes. But you can never em ulate the real thing perfectly which is why they usually rely on physical mockups.
How are reboot/reprogram sequences like this handled/practiced/tested?
No simple answer to that. But check out NASAs system and software design guidelines, or better yet take a spacecraft design course at your local college or university if they offer one.
Even at design stage I imagine failure modes are extensively analyzed and multiple redundancy built in.
Yes - very extensively. Anything that goes into space is analyzed for years. You have no idea how expensive and time consuming this can be until you actually work on one of these projects.
Anybody know how space tech is handled?
Yes - expensively and with great rigor. There are no shortcuts.
Re: (Score:1, Redundant)
Do they have several mock ups?
They have actual duplicate examples of onboard units, as well as "breadboard" versions built for easy access to the innards.
A complete computer model of the whole thing, emulated right down to hardware and software?
Betcher sweet ass.
How are reboot/reprogram sequences like this handled/practiced/tested?
Endlessly.
Even at design stage I imagine failure modes are extensively analyzed and multiple redundancy built in.
Yes they are. But before switching in a redundant unit
Re:I'm curious... (Score:5, Informative)
I'm curious, I presume somebody knows this.
I don't work on hubble but I know something about this stuff and I'll try and answer your questions directly
Do they have several mock ups?
Yes, for hubble there are several mock-ups, from ones that are fairly low fidelity that are used by the software developers (maintainers) for code/debug/test to a very high fidelity full scale electrical and mechanical mock-up of the aft end of the vehicle called the VEST, when the astronauts practice the repair tasks on dry land before they move to the pool at JSC to learn to do them in a simulated 0g environment.
A complete computer model of the whole thing, emulated right down to hardware and software?
When HST was built we were still doing spacecraft control simulations on hybrid analog/digital computers. For all missions in the last 20 years or so there are computer models of all the control modes built using products like matlab/simulink long before any metal is cut. I wouldn't be surprised if there is a matlab (or similar) model of hubble now as the hybrid computer is long gone. In fact, I'm almost positive, as I don't see any way they could have come up with the 2 gyro science controller otherwise.
How are reboot/reprogram sequences like this handled/practiced/tested?
Not to be a wise guy, but reboot and reprogramming are handled very carefully. This is why switching to a backup system is going to take 2 weeks, most of which were used for analysis and a formal review before the decision to swap to the B side was made. It probably takes 5 minutes to send the actual commands to switch to the B side of the data formatter, but they will double, triple or quadruple check everything as they go thru the process. Remember, they are switching to hardware that hasn't been used since 1990. They expect it to take about 40 hours to switch to the B side and plan to be done by Friday.
Even at design stage I imagine failure modes are extensively analyzed and multiple redundancy built in.
There are failure modes and effects analysis done at each design step. Before launch there would have been a peer review of the final failure detection and correction design.
A lot of the NASA standards are available to the public. If you go to http://standards.nasa.gov/ [nasa.gov] and click on the public button it takes you to the listing of them.
Re: (Score:1)
I'm not sure why you posted anonymously one of the more informative entries in this thread. Might I just add that if I were responsible for switching over to a redundant system especially if I were uncertain as to the exact failure, I would be going over the software and hardware until I was certain of what was going to happen when I did.
For example: The primary system failed due a malfunction in the breakfast dispenser, the first thing that I would want to do is to ensure that the redundant system doesn't
Cosmic rays? (Score:2)
You have to remember that the Hubble Space Telescope has been in orbit for at least 15 years. (I could look up the actual amount of time but it is unimportant.) The probably has been a lot of exposure to cosmic rays that can damage electronics. The damage can be mitigated somewhat by proper shielding. The only disadvantage is that this is heavy so you would want to keep this to a minimum.
Re: (Score:2)
Of course, that was the gold standard when the Hubble was built. That's how I remember it. It was a unexciting, but guranateed process development practice.
Nowadays (since 1995 when DoD went to better/faster/cheaper mentalityand threw out the MIL-STD standards!), it a watered down version using RAD, XP, Agile, and SOP since software development has gone internet style.
Hence why [space vehicle] failures are up almost 30% (in gov't, more in commercial) since they dropped the standards. Here'
Programming in Hollywood (Score:1)
does anyone else think.... (Score:2)
What about (Score:1)
By RC? THAT'S why my garage door is going nuts! (Score:1)