Hacker News Reader

Comment by samuell | Hacker News Reader

> "This proves electronics built around 50 years ago, 12 years before Voyager 1, and far before microprocessors and integrated circuits are still capable of working in the hostile environs of space."

I think in some aspect, it might be even more so (robust), than today's technology: Simpler, more to-the-point technology, leading to greater robustness.

I think today's tech and IT industry has something to learn there, and I'm really happy about the "back to basics" trend in IT, going back to compiled, to-the-point languages like Go, closer-to-the hardware cloud systems etc.

View on HN

Recently I've been working my way through the PPrune megathread on Concorde: http://www.pprune.org/tech-log/423988-concorde-question-77.h... , in which (somewhere) is a discussion of the control systems. Analog fly-by-wire! A complex set of "synchro" and "resolver" systems connected to op-amps, giving it autostabilisation and autothrottle all the way up to Mach 2. The pilots would just push the throttles full open on the runway and leave them there until it was time to come down.

AFAIK Concode only uses ~80% thrust on takeoff due to vibrations engines create and risk of overheat. Also fuel gets pumped around to change CG, it is a fascinating piece of machinery.

It's all in that thread :) The 80% limit was only on one engine, due to vortex issues. The others had to be at "100%" and full afterburner to get off the ground. The actual volume of fuel flow for a particular power level was also determined by an analogue computer measuring the inlet air temperature and pressure.

http://www.pprune.org/tech-log/423988-concorde-question-4.ht...

Relevant distraction: Nice in cockpit vid of pre takeoff CG and fuel transfer checks, take off, and after checks. 10m.

http://www.liveleak.com/view?i=16a_1231096566

Low light takeoff, because pretty: https://www.youtube.com/watch?v=l7mt6AKKhq4

NSFW warning for the liveleak vid, if you're at work the site contains porn ads.

> NSFW warning for the liveleak vid, if you're at work the site contains porn ads.

For you maybe.

You gotta start signing out of your google and facebook account when searching certain things... retargetting will get ya! ;-P

Someone should write a book based on that thread, seriously.

I just returned here from that thread. Fascinating and time consuming. Well worth reading.

The F-16 used analog fly by wire, and the F-16's analog computer was very well thought of.[1] Here's what it looks like.[2] All those discrete parts! Note the quadruple redundancy.

[1] http://www.f-16.net/articles_article13.html [2] http://duotechservices.com/pitch-1-board-repair-of-the-analo...

I think the main lesson would be that making your transistors and switching currents enormously large is good for longevity in space, but that's nothing new.

Well, I was also thinking about (but didn't write) that most mission-critical systems aren't written in javascript :) ... but rather something "ancient" like ada or pascal.

I kinda doubt that any satellite from this era ran a high level language.

Why? The first high level languages were already existing in 1942, so it’s not that far fetched.

EDIT: As I can’t answer to comments due to the retarded "you are submitting too fast" rule, here is a source: https://en.wikipedia.org/wiki/Plankalk%C3%BCl

.. on room-sized computers. VLSI hadn't been achieved yet, so Apollo flew with a suitcase-sized computer made from discrete gates and core memory programmed in assembler by Margaret Hamilton and others. https://github.com/chrislgarry/Apollo-11

I doubt the satellite bothered with a processor.

Not trying to be pedantic or a dick, but you should refer to that language as assembly, not assembler. It is a non trivial difference as assembly refers to the language and assembler refers to the software written to parse assembly commands into machine code (binary strings defined in the given machine's instruction set that are meaningful to the processor).

Otherwise it's similar to referring to the language C as Compiler, which is just silly :)

http://stackoverflow.com/questions/1176138/assembly-vs-assem...

Also, to your point, agreed. It took several years before we could actually fly high level, compiled languages in space, the first being HAL/S on the Shuttle (completed in 1975).

https://www.cs.toronto.edu/XPL/hal.html

Cheers

This is the most pedantic thing I have seen in 2016.

I dunno how other !English languages handle it, but in German both are simply called Assembler. It would also seem that this distinction isn't generally made, due to the human ability to deduce meaning from context.

You must be new to HN.

I'm gonna need a source on that.

The first actual formalised and widely used programming language was FORTRAN in the 50's.

Which just had its 60th birthday a couple weeks ago.

From the article you linked:

> the first compiler for it was implemented in 1975

Although a few high-level languages existed at the time, compilers back then were not capable of code optimization. The only way to write code tight enough to operate a spacecraft would have been Assembly language.

Nonsense. Fortran I (happy 60th birthday!) was an optimizing compiler, and would have failed had it not been.

Who knows, IIRC planes are dedicated C++ frameworks that spit out tons of intermediate representations and then compiled.

Not a satellite, but Deep Space 1 ran on "Remote Agent", which was entirely written in Lisp and even had a remote REPL, which was used to debug and fix a race condition bug while millions of miles away.

That and more about Lisp at JPL are on this awesome little page written by a former JPL engineer: http://www.flownet.com/gat/jpl-lisp.html

While fascinating and true, DS1 was launched in the late nineties, over forty years later. Pretty sure that's a different era :)

Oh, did I misunderstand what he meant by "this era"?

I would assume the proper way to phrase it would be "that era" if he meant the 60's, although he did use "ran" which is the past tense. He also mentioned JavaScript, which didn't even exist in that era, so I guess I'm confused by the entire statement.

Edit: by 'he' I guess I meant 'you'. Which means you did mean to refer to the 60's.

"This era" here means the era in the context of the discussion, not the current era. "That era" would also be correct.

That's incorrect. Commercial avionics are written in C++ these days. Automotive systems are in C or C++. Ada never got any real use outside the US military, and even there it's dead.

Not if you read something like this:

https://www.seas.gwu.edu/~mfeldman/ada-project-summary.html

Sure, the "military" list is the longest, but far from the only one. I also notice four Swedish military systems there (yay :) )

Ada ain't dead. It's still in use for some aircraft systems, not all of which are military.

It's probably all legacy stuff. The avionics stuff I worked on was all C++. I got to see some Ada code, but it was all ancient legacy systems that hadn't been replaced yet. I seriously doubt anyone is starting any new projects in it.

Um... isn't this indicator of further decay of the system? Like a cadaver that fills up with gas and starts emitting sounds.

Imagine the cadaver was just some person who walked on a treadmill of a windmill when the wind wasn't blowing. The person died on the treadmill, and since this was out in the middle of nowhere, the whole scene was left untouched.

The corpse gets wedged into the gearing and the machine stops until the body rots enough that what's left falls free and the gears start moving again when the wind blows.

One thing about tech in hostile environments is resilience to radiation, which is more or less proportional to number of electrons in junction. With smaller fabrication process we win lower energy consumption and much better integration, but at a cost of radiation resilience.

Unlike this one that never worked properly, two other satellites launched in the same series, LES-8 and LES-9 are still functioning as designed. They were powered by RTGs (radioisotope thermoelectric generator) rather than batteries/solar though.

I wonder what kind of lifespan they were expecting out of this thing back when it was launched.

I'd think you'd want a decent return in terms of longevity, considering the size and (I would assume) weight of it.

I'd bet modern satellites are much more disposable by comparison.

I very much doubt anyone would consider something that costs hundreds of millions of dollars to build and fire into space "disposable". It's not like you can buy a new one for cheaper 5 years later.

They are considered disposable because the operating costs usually outpace their value once obsolescence sets in.

Which is why many satellites are resold on secondary markets after their primary mission timeframe has passed.

How about Rosetta?

Well, you sort of can. The majority of cost is R&D, once that's done and you have a template for a satellite, you can definitely build more of them cheaper than the first one.

No they wouldn't launch a spy satellite disposable camera that took one roll of film and was then useless after the film was jetisoned.

https://en.wikipedia.org/wiki/Corona_%28satellite%29

The reverse is the case. Longevity of satellites had increased over time. This is due to a variety of reasons ranging from materials and design improvements to more advanced electronics and superior stationkeeping thrusters as well as reduced reliance on consumable components.

Early spy satellites, for example, relied on film that was returned to Earth, once the film ran out the satellite was useless. Other satellites simply became obsolete or lost attitude control and were no longer useful. Telstar had a service life of only about a year, the first half dozen TIROS weather satellites had lifetimes of about a year or less. Today satellites with similar missions are expected to last decades, which they often do.

Another reason is that nobody is detonating nuclear bombs in space these days. A major cause of Telstar's short life was intense radiation from high-altitude nuclear tests. I'm not sure how long it would have lasted without that....

Hah, well, that only happened once. And even so the overall lifespan of satellites was not very long back then.

Yeah, I definitely get the impression that Telstar might have lasted two years instead of one if not for the nuclear tests, but not decades.

There's a certain level of intuition that one can grasp with hardware as opposed to software. I suspect this might contribute to most robustness in design.

Didn't they used to build navigation systems using analog components so in the event of EMP your systems would have a chance of keeping on working?

Reply on news.ycombinator.com