Hubble Advanced Computer Down

Says it's a payload sub-processor that failed, a NSSC-1:

Not the image processing/systems management main computer

The NSSC-1 sounds like a pretty dumb machine, basically a discrete micro-controller made with MSI chips as of whatever the latest revision mentioned in the wiki is:

They probably weren't using core memory even at the time Hubble was built, tho

That's way back in the stone age, pretty sure it's not that. Or even this much more advanced one:

Hubble was designed in the stone age! They started designing the bugger in like 1975 and construction of the main mirror began in 1979, there were tons of schedule slips and cost overruns on that project. They wanted to launch it in 1983 originally...here's a technical paper about re-programming the NSCC-1 of the Hubble in-flight so seems like at least as of 1993 it had one:

This is a fun book:

But as far as we know all the KH-11 mirrors were fine which were essentially the same tech, being designed and built around the same time...so why didn't they just get those guys to do the mirror in the first place? ???

NASA apparently has two "stubby hubbles" in storage they're planning to put up there, someday, though they're configured differently due to their spy-mission so aren't as useful for the farthest observations:

This is the book I'm reading right now:

The Statue of Liberty was closed for repairs the first time I visited NYC, and Space Mountain was closed for repairs the first time I visited Disneyworld. Thanks, corrosion...

Looks interesting. I'll try it.

That was Perkin-Elmer. People walk away from that place with PTSD.

I don't see much about the Hubble lens here:

Out of curiosity, please elaborate...What causes the PTSD at Perkin Elmer???

I knew a few folks at Perkin Elmer Danbury (where the mirror was made). They all seemed fairly normal. After the big disgrace, the facility got sold off to Hughes, and then Goodyear, then United Technologies, and most recently (2020) seems to have been spun off alone as Danbury Mission Technologies.

That could get seriously old all by itself.

Cheers

Phil Hobbs

I had a boss in the 1990s who had worked at Itek when the Hubble mirror (and optical system) was made. He told the following story:

They had two "null correctors" (optical systems designed to cancel what the system being built does), so a simple test on the combined system of null correctors and newly-made device could be tested as if it were a perfect spherical mirror tested at the prime focus (the centered of curvature).

There were two null correctors made. One was simple, made from purchased optical components. The other was complex and precise, with custom-made optical components. When the Hubble mirror was tested, the two null correctors disagreed.

Itek had proposed to do a full ground test of the whole Hubble telescope optics, for $20 million. This would have caught all problems, not just the mis-made main mirror. NASA declined to exercise the full-test option.

Instead of setting up a third test to break the tie, management decided to believe the full custom null corrector and ignore the simple null corrector.

Turns out that the simple null corrector was correct, and the fancy null corrector had a slight error of construction - a spacer was a millimeter too long or short.

So the main mirror was ground wrong. Oops.

Joe Gwinn

.

.

The error on the Hubble mirror was at the level that an amateur telescope mirror grinder would detect. Amazingly, the telescope was launched without actually ever imaging anything.

Are you certain the right satellite was launched? I thought it was short-sighted for focus at ground level :-)

Issue is moot know. Even their entire backup computer won't work, gives them the same error- sounds like a CRC check on the data keeps failing. Maybe the Russians will be able to visit it and install a whole new computer system.

Hmmm OK, for a lot of things, I am a 'put it together and test it' kind of engineer. Well, after doing a full review of the system analysis...even more so when it cost $X to do it on earth, and many times $X to fix it in space. Why would an organization offer to do an integration test for an additional $20M, as opposed to having that task as part of the original contract??? Seems like a severe lack of engineering diligence to me. or is this another case of handing the contract to the lowest bidder, good engineering practices be damned?

I've seen plenty of the latter and been involved in red team tech assessments after a system failed....It is a great money maker for the contractor and the USG actually enables this stupid behavior....Ohhhh, we think you might want to do an integration test even thought it wasn't part of the original proposal...sure, we can do that...for another $20M please......

I don't think you understand how government contracting works. Everything you do is totally visible and stated in the contract. The bidding process involves a certain amount of adjusting the statement of work to optimally suit what the government is looking for. It is up to the government to indicate what testing is required. They opted out on that test and it bit them. It wasn't the first time and won't be the last. We all make mistakes in the interest of economy. Often such efforts work out, sometimes not.

Well, with govt contracts, one is forbidden by law from doing more than what the contract required. The contractor does have an implied duty to warn the customer of the omission. So the standard approach is to offer the forgotten and/or well-advised items as options, with an explanation.

Joe Gwinn

"We'll put it in the overruns."

Cheers

Phil Hobbs

The attitude on Hubble was "the optics is so good that there's no reason to use unworthy tests on the ground"... like looking at a star.