Hibernating instrument on Hubble roused as engineers ponder message problem Ever had a component spew garbage and had to bodge around it? Engineers appear to be faced with a similar issue as they continue work to rouse the science instruments of the Hubble Space Telescope from their Safe Mode slumbers. The good news is that the team might be close to isolating the problem. The bad news is that the …
HST Team raise ServiceNow problem ticket.
Outsourced Support Team immediately ask for more information, like "what is HST? what does it do?" thereby stopping the clock on the ticket SLA. That goes on for a bit, then OST passes problem onto some other team who have absolutely nothing to do with HST, still trying anything and everything possible to not do any investigation of the problem. Then eventually back to the HST team just before the system automatically closes the ticket as it has gone beyond a certain time frame.
Joke icon. But the reality in a company I have worked for
"We've all encountered that one weird component in our projects that occasionally does something strange and we end up having to code a bodge to deal with its funny little ways."
Too true. On an instrumentation project decades ago, we found that the hardware 'full' flag of a dual port memory didn't reliably signal memory full. We'd intended this to allow user substitution of devices with different capacities. To work round it we had to implement a user defined initialisation parameter in the software of the data sender so the system knew what size memory had been installed.
When Nasa built the Hubble telescope they actually commissioned two identical ones, because when you're dealing with taxpayer money, why build one when you can build two at double the cost?
The second one has been in a museum since then. Just dust it off and strap it to a rocket.
There's an interesting story about that: the one in the museum apparently is better than the one in space because they were built by different contractors.
Building a single copy of all your unique instruments right from the get go is a seriously stupid idea though. The actual hardware isn't the big cost (apart from the main mirror, that's a very expensive bit of hardware). For all the other stuff you make 2 and start assembling at least the more basic stuff, pick the best and then integrate all the best choices into a single vehicle for launch to orbit. I highly doubt the museum piece would be fully functional and orbit capable hardware. At the very least it'll be missing a mirror (they only made one. Incorrectly as it turns out)
There was a 'backup' mirror blank made but never polished.
It's in the Smithsonian (or it was last time I saw it)
This is normal practice for large or expensive mirrors - sometimes things happen in the grinding or polishing, a flaw is discovered or an accident takes a chunk out.
Zeiss made a spare 8m mirror blank for the VLT (made 5 used 4) that they were trying to sell off to anybody who needed an 8m mirror - don't know what happened to that, probably got melted down.
Knowing Zeiss, they’ve probably forgotten where it is. A friend has some incredibly good, but definitely not cheap, Zeiss optical equipment.
As they couldn’t come and service it in the pandemic, they sent a 30k machine out to use until the original was serviced, around September last year. We’re still falling over the spare now…
First, there is no need for a new mirror.
Second, all the computer kit, which is where the problem lies, has backups.
Third, they are already running on the backup kit since they brought the Hubble back to operational status in June when it encountered a similar issue.
NSA built few for them self. That one in orbit is pointed at Earth and has a resolution about 30cm to 1 meter. How good it is remains a guess work because its classified on the technical specifications.
NSA donated the two remaining to NASA in 2012.
https://en.wikipedia.org/wiki/2012_National_Reconnaissance_Office_space_telescope_donation_to_NASA
Every time we do this.
(1) The NRO carcasses are not Hubbles: their focal length is much shorter. (2) It is not clear if reusing one is even worth it for applications where wide field is suitable: I believe that NGRST is going to use one of the NRO carcasses, but there was (is?) much discussion over whether it would be cheaper to start from scratch. (3) Mirror is not this huge expense: consider VLT: this has 4 8.2m mirrors and 4 1.8m mirrors, and cost less in total than cost of a single shuttle flight to service HST.
And finally (4) It is not clear to me that a new HST is even interesting now: suspect that for wavelengths it covers ground-based telescopes are now better. Certainly think if asked to choose between 10 VLTs and one new vidible-light telescope in space most astronomers would pick 10 VLTs. Certainly also is the case that there are telescopes that need to be in space – JWST for instance will do things that could not be done under atmosphere, but not clear that visible-light super-HST is interesting. But I am theorist: will defer to practicing astronomers on this (just not to space romantics).
When HST was delayed by Challenger disaster they had to keep telescope in storage for about four years. Keeping it in flight condition cost about $6 million a month, in late 1980s dollars. If this second, incomplete telescope could have flown in 1990 (it could not), then it has been in a dirty, uncontrolled environment full of horrid oxygen and other corrosive gases for 30 yeas since then: it certainly can not now,
We've all encountered that one weird component in our projects that occasionally does something strange and we end up having to code a bodge to deal with its funny little ways
It's called Visual Studio but I stopped finding it's time wasting quirks funny a long time ago.
They're not called "weird components" they're called "users".
Of course it sounds familiar. It not only sounds familiar, this is the mind-set of anyone---and their reptilian-IQ managers---who is (laughingly) referred to as an "engineer", nowadays; to wit:
"Software can fix any hardware problem."
Just ask Boeing about its illustrious use of this "fix", to create---with software---a completely brand-new aircraft, the 737-MAX, out of a completely old airframe...which was---at the time it was designed---designed the old-fashioned way of designing aircraft: by real engineers, using real hardware, created for the design of real, safe aircraft.
Now you know:
Software can fix any hardware problem.
It could have worked fine if they had used 3 sensors and implemented a "voting" system to reject spurious data. The fault wasn't necessarily "fixing it in software". The fault was in doing so badly, ignoring the added hardware requirements and not providing training. Oh, and lying to aviation authorities about the nature and effect of the new systems of course. That's kind of a big issue all by itself even if it hadn't gone wrong.
Airbus has very similar systems in all their aircraft, the difference being that they implemented them properly and pilots are trained in their existence, effects and failure modes.
Lift capability, yes. Crew facilities for space walking, nope.
X37B with remote control - maybe. But if it’s electronics giving up ghost, that’s some awkward cct board replacements to do without a human in the loop.
Hubble is approaching the point of decommissioning while it can still be commanded to do so. No pressure on JWST or anything to take over.
As has been said elsewhere building a bunch of expendable orbital scopes without the cost of the shuttle programme would have been cheaper. But politics versus efficiency is a long standing known issue. And A scope is better than no scope at all. Nobody “Had” to build Hubble.
"We've all encountered that one weird component in our projects that occasionally does something strange and we end up having to code a bodge to deal with its funny little ways."
It has always surprised me how many hardware designers don't allow for metastability*** in logic components when interfacing with asynchronous inputs. It produces glitches that fulfil Murphy's Law with the Sod's Law rider "..at the worst possible moment".
**Metastability is what happens to you when the doorbell chimes - and then the phone rings at the other end of the hall. You pause for an indeterminate moment - not sure which one to service. In logic gates there are input set-up timing constraints for reliable propagation times.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
