（评论）

（评论）
(comments)

原始链接: https://news.ycombinator.com/item?id=40117599

本文描述了与 20 世纪 70 年代发射的航行者号太空探测器相关的技术问题的令人印象深刻的解决方案。它的作者在几个段落中对问题和解决方案进行了清晰的解释，不受 SEO 的限制。他们的沟通风格反映了军事背景，强调效率和精确性。然后，文本想象了航行者号和当代航天器之间潜在的相遇，表达了对其尽管已经有四十五岁以上的耐用性的敬畏。它强调了由于距离遥远和处理能力有限而继续更新软件所面临的挑战。此外，它还思考美国宇航局是否考虑过深入太空冒险的影响，质疑先进文明是否可能已经发现了我们。最后，本文推测了太空旅行的潜在改进，暗示了太阳系内部核推进和观测研究的进展。

The substance of this news release is incredible, but its style is also admirable. The authors managed to convey in seven paragraphs a concise and comprehensible explanation of how the team resolved the technical issue.

That's what happens when you're freed from "SEO-optimized content". It's also a culture thing I'd probably put under military philosophy. I've worked with ex-military engineers, and you can tell from how they communicate. Writing technical reports and memos is a skill.

There's something very beautiful about Voyager's journey so far.

I hope one day when we're a true interstellar species we'll still keep tabs on it. The data may not be useful anymore, but it would be cool to imagine a year 3000 society with a little "Look at where Voyager is now :)" tool that you can see its path and where humans have colonized by comparison.

I think a 19 year old yobbo, whose dad is a successful interstellar logistics businessman, who out of a guilty conscience for never having had time for his son and having bought him an overpowered spacecraft, will either put grafitti on it or misjudge his afterburner and half burn it while trying to fly a very tight corner around it, in order to impress the 2 girls he has on board.

There is no destination; there is only the voyage.

There was no beginning; there is only the voyage.

In life, we voyage together, and in death we shall voyage alone.

Now, then, and forever voyaging.

I’ve imagined a scene playing out, in sci-fi or for real in the distant future, where astronauts test out a new propulsion system by flying out towards Voyager 1 and catch up to it with ease. As they approach, they see the ancient probe grow larger and larger in their window until…

If my math is correct, we would already need to build a spaceship that can travel at 1/10th the speed of light to reach Voyager 1 within one week. It will be quite an engineering challenge for the future.

Ah yes, the first Michael Crichton novel I read - and also the book that first introduced me to binary code (https://imgur.com/N4IjIYq)! And after watching (and then reading) Jurassic Park some 10 years later, it took a while until I realized that it was from the same author...

Yes, I too had the same disparity in recognizing the author as a young 'un, reading CONGO and SPHERE and so on .. Mr. Crichton sure had his finger on the pulse of the technological world we live in. What a wild series of stories he has created .. he was my favourite author until Messrs. Stephenson and Gibson came along ..

Staying on subject, I wonder if we will see a new adaptation of ANDROMEDA STRAIN some time. As a story it seems topical and relevant.

The plutonium 238 decays according to a curve, and the thermocouples are degraded as well by heat and radiation according to a curve. So the power output drops rather predictably: "The radioisotope thermoelectric generator on each spacecraft puts out 4 watts less each year." [1]

The Voyagers will soon no longer have enough power to operate any of their instruments. They'll have enough power to continue operating the transmitter (which serves as a science experiment of its own) into the 2030s. The power of the signal will drop before the electronics and control brown out (if it works as designed), and it the signal might become too weak to detect before the probe completely stops operating. Such a fate befell Pioneer 11, who may yet still be warbling away at low power no longer pointed at Earth; its carrier was last detected in 2003.

[1] https://voyager.jpl.nasa.gov/frequently-asked-questions/

> [1] https://voyager.jpl.nasa.gov/frequently-asked-questions/

Also:

Even if science data won't likely be collected after 2025, engineering data could continue to be returned for several more years. The two Voyager spacecraft could remain in the range of the Deep Space Network through about 2036, depending on how much power the spacecraft still have to transmit a signal back to Earth.

That FAQ covers a lot of interesting ground (though it talks about 2020 in the future tense).

After Voyager 1 took its last image (the "Solar System Family Portrait" in 1990), the cameras were turned off to save power and memory ...

I didn't realize that was the last image.

... it is very dark where the Voyagers are now. While you could still see some brighter stars and some of the planets with the cameras, you can actually see these stars and planets better with amateur telescopes on Earth.

> The two Voyager spacecraft could remain in the range of the Deep Space Network through about 2036

It would be quite depressing if it was us failing to receive a fainter signal, rather than Voyager failing to send it.

> the cameras were turned off to save power and memory

Since it’s powered by an RTG, how does the power get “used up”? I assume that this refers to the available power budget at a given moment versus some sort of expendable power reserve.

It's the first question at the top of that ^ FAQ page. One of their comments is : "Mission managers removed the software from both spacecraft that controls the camera." Makes me wonder if that unused RAM came in handy lately!

> the transmitter (which serves as a science experiment of its own) into the 2030s.

One of the longest-running scientific experiments, too. It's already about as old as the Queensland pitch drop experiment was when Voyager I was launched.

> when we're a true interstellar species

If we can harness all of the energy and mass available in our solar system, we [1] can likely compute more than several galaxies full of classical humans. We might even begin to test the edges of physics.

Maybe we don't need to go anywhere at all. Maybe we [1] have all we need right here to become literal gods.

[1] Our digital descendants. Humans are very much fit to the gas exchange and metabolism envelope of our gravity well.

> compute more than several galaxies full of classical humans

Unfortunately the simulated classical humans in your Matrioshka Brain will want to mine Bitcoin, which means that our digital descendants will have to become a true interstellar species anyway in order to convert the Laniakea Supercluster into coin-mining computronium.

> If we can harness all of the energy and mass available in our solar system, we [1] can likely compute more than several galaxies full of classical humans. We might even begin to test the edges of physics.

I grant that. But why would that keep us from pressing on?

If we have more resources in general, we will also have more resources for interstellar adventures.

Even if you construe this phrase as unequivocally negative, you can't scrub negative emotions and intentions. If you were able to actually scrub some word or phrase from the language, another would take its place, like with all the racial and intellectual disability slurs.

Voyager has been an inspiration for generations of engineers.

Bless you all that worked on it. Thank you.

Recently I designed in a Voyager inspired secret Easter egg into the surgical robot I designed. I put a gold (plated) plate with everyone's signature engraved on it. Gave everyone one as a surprise Christmas gift.

It's the Maestro from Moon Surgical, it's done over 200 surgeries, all successful. So far (no whammy) its had 100% reliability with the first 6 systems we built. We designed it (hardware wise) with only 3 engineers, including me, and we hand built the first ones right here in San Carlos, CA. The company is based in Paris and has a whole interesting history there as well.

No, but I worked on one of those too. I worked on J&Js ottava, its designed to compete with the one doing the grape surgury, Intuitive's Da Vinci. I've had the pleasure of using one (Da Vinci XI) to play around with, it's truly amazing.

How they manage to squeeze all the resources of the probe and keep it working year after year is an astounding achievement, pleasantly mind-blowing.

It is important that all the know-how about this type of maintenance never disappear. I hope the designs in electronics that this team would have wanted to have available in the probe are implemented in the new designs.

I should see whether there's documentation of what they moved and what they replaced. I imagine there's "plenty" of room to do that (in the sense that there's probably some programs that are no longer mission-relevant because they controlled systems that have been shut down), but I'd love to know what got tossed.

Heck of a job.

> With some amount of luck, Voyager might last ten more years beyond that:

Oh, good, it won't have to suffer a year 2038 problem :)

Also: what's with that last link? Definitely didn't expect something to make my browser slow to a crawl.

What a sweet film. Thank you for the link. This whole time when I heard about work on the Voyager mission I assumed there was a larger team, with fewer single points of failure.

> A radio signal takes about 22 ½ hours to reach Voyager 1, which is over 15 billion miles (24 billion kilometers) from Earth, and another 22 ½ hours for a signal to come back to Earth.

It has taken 46 years to get 22.5 light hours away from Earth!

> The team started by singling out the code responsible for packaging the spacecraft’s engineering data. They sent it to its new location in the FDS memory on April 18. A radio signal takes about 22 ½ hours to reach Voyager 1, which is over 15 billion miles (24 billion kilometers) from Earth, and another 22 ½ hours for a signal to come back to Earth.

Talk about a slow feedback loop! And I get frustrated when I need to push code to a repo to test things in CI...

Posting a question in a forum has its benefits though. A bunch of drive by folks end up picking up information they would never have gone through the trouble of researching themselves.

Pretty incredible feat of engineering (both back when it was launched and now). Does anyone know its current purpose? I’m curious if there’s anything we’re actively using it for or if it’s just a matter of “look for surprises”.

The magnetometers and charged particle detectors are still operating. So they are measuring the galactic magnetic field, and cosmic rays and the gas in interstellar space. The results are more or less what was predicted, though the exact boundary of the sun's influence was only discovered when Voyager 1 and 2 crossed over it in 2012 and 2018 respectively. Beyond that, yes, they're basically assuring us the sun is still there and that space is very empty. I don't think anyone expects the interstellar gas to vary in density on the timeframe that the Voyagers will be able to observe it but, I guess we'll find out!

Edit: I spoke way too soon. It varies, as discovered with Voyager measurements recently: https://www.jpl.nasa.gov/news/as-nasas-voyager-1-surveys-int...

Yep, 45 years old hardware, still getting software updates. Hey Apple, JPL is close to you, can you get someone to bike over and see how they do it? Thanks!

Planned obsolescence [1] in the problem. Apple could certainly create modular devices designed to be used and piecemeal modified/updated indefinitely. But it's way more profitable to sell you a brand new phone every couple of years than it is to sell you a modular device that you could update as desired, especially if the parts/connections are standardized meaning you could foster a [normalized] third party market for batteries, screens, etc.

Devices would last way longer, prices would be much lower, and it'd be unimaginably better for the environment. But Apple wouldn't make as much money, and the government's GDP growth figures wouldn't be as high. So clearly it's a terrible idea.

[1] - https://en.wikipedia.org/wiki/Planned_obsolescence

The average vape has more processing power than Voyager, and the iPhone is orders of magnitude more complex. With that said, it takes skilled engineers to squeeze perfectly crafted code into such a tiny platform from the 70s.

A Pinecil (digital soldering pen) is probably a better example. BL706 MCU, "a low-power, high-performance IoT chip that supports BLE/Zigbee wireless networking, ... BL702 has built-in RISC-V 32-bit single-core processor with FPU, the clock frequency can reach 144MHz, has 132KB RAM / 192KB ROM / 1Kb eFuse storage resources, supports external Flash, and optional embedded pSRAM."

The Voyager had a custom-designed processor (well, several) that were basically computers made out of basic logic chips (74xx series); see details here https://www.eejournal.com/article/voyagers-1-and-2-take-embe...

Either way, it's clear that we (well, JPL) can build extremely powerful and sophisticated systems with relatively small computers, suggesting that resource constraints can sometimes be a source of stability and creativity.

>The Voyager had a custom-designed processor (well, several) that were basically computers made out of basic logic chips (74xx series);

This was not unusual at the time: many early arcade games were made exactly the same way. They've even been emulated by the MAME project.

Good lord could you imagine the meltdown HN would have if Apple had taken this option to solve the old-batteries-support-lower-peak-current physics problem?

“Your device battery no longer supports the camera. Or the backlight on the top third of the screen. But it runs at full speed otherwise!”

In 2016, Apple released the original iPhone SE and Google released the original Pixel phone.

The Pixel stopped getting security updates at the end of 2019. The OG iPhone SE is still getting security updates eight years later.

Maybe Google should try seeing how Apple does it?

The Google Pixel 8 series receives 7 years of updates. And unlike Apple they are upfront about it, which is important. Maybe your iPhone gets 8 years of updates, maybe 6 noone knows.

The current iOS 17 is compatible with the iPhone XS, which is from 2018. That's 6 years for a piece of tech that the majority of people replace after < 4 years...

Also to nit pick, Apple is based in Cupertino (northern CA) and JPL is in Los Angeles - so it'd be quite a bike ride lol

I have an iPad Air from 2014 that hasn’t been able to get updates since iOS 12.5, so 2019. The electronics are fine, a browser update would be awesome. Would people not replace phones in < 4 years if they could have current software? 4 years on a $800 thing doesn’t seem to be a good deal for the majority of us.

Sorry about the distance thing. I’m from Philadelphia, so all of that (vaguely waving west) has got to be bikeable. But I’ll remember the /s for next time.

Yep... days worth of work to get Python code I wrote 3 years ago working again from all of the 'bitrot.' Can't imagine how much work it must be for them to produce new binaries to update these old systems from modern computer hardware.

Although I suppose it could actually be easier depending on how the code works- perhaps it's just simple bare metal assembly without the approx 10^99999 libraries a modern python stack has.

It's probably a lot easier in a way, because they don't have to worry about external dependencies changing at all. Modern code is a real PITA that way.

What's hard about their work is 1) it's really, really slow to communicate with, so you can't iterate quickly, 2) the tech is really old and unlike today's stuff, so it's very specialized domain knowledge.

Nice one! Voyager carries the hopes, aspirations and fantasies of many of us space romantics.

On the technical side of things, there are also other companies doing live patching, like the Ericsson telephone exchanges. Their code can be altered “live” while operating, in order to fix or enhance the software and with zero downtime ;-)

All I can say is wow! This probe refuses to die. Despite being built with almost 50 year old technology. Amazing engineering by the people that designed and built it. Even more amazing is the people that continue to debug software problems from 2 light days distance.

> from 2 light days distance

The distance is currently 22.5 light hours (and increases by half-hour per year.) But it is indeed about 2 days of round-trip time for debugging.

Although that’s one of the many possible explanations of the Fermi paradox [0], I prefer to think that the real reason we haven’t discovered (or we haven’t been discovered) is the fact that we’re limited by the speed of light.

The distances are so vast, almost unfathomable, that we need Faster Than Light means of traveling. Perhaps I’m being naive or romantic, but I prefer to think this is the real reason :-)

[0] https://en.wikipedia.org/wiki/Fermi_paradox

The speed of light and great distances are indeed a limiter, but the universe has been around for billions of years. Even with the great distances, an interstellar civilization that's been around for millions of years would have had plenty of time to find us by now.

Must be difficult debugging a system with a 45 hour round trip each step of the way. And here I thought debugging a system on customer premises was tough. Hats off!

Worse than the round-trip is that there's no second chances in some scenarios. If you mess up the wrong part(s) of the system, it's bricked with no way to recover it.

The Voyager project is an amazing feat for humanity. But I wonder, does NASA take into account the repercussions of sending a probe deep into space? I know space is big but I can only think of the dark forest hypothesis. What's the plan for further space exploration?

Voyager 1 is currently 2.4e10 km from Earth. Trisolaris - sorry, I mean Proxima Centauri - is 4e13 km away [2], so in the (almost) 47 years since its launch, Voyager 1 has covered a whopping 0.06% of the distance. And it's not even headed that way.

Meanwhile, radio, TV and radar have been advertising the presence of a new technological civilization on Earth for more than a century. That means any entity worth worrying about within a 100+ light year radius - a distance 44+ times longer than to Proxima - already knows about us. And that sphere keeps growing in all directions at the speed of light, or roughly 18000 times faster than Voyager 1 is moving.

If you really want to worry about the Dark Forest, it would be more justified to ask if your local radio station takes into account the repercussions of sending commercials and TOS reruns into deep space.

[1] https://theskylive.com/voyager1-info

[2] https://en.wikipedia.org/wiki/Proxima_Centauri

Inverse square law. With the power we're transmitting at, signals become just background noise relatively quickly. They're nowhere near strong enough to be detectable at e.g. Proxima Centauri. This is what makes the radio signals we do detect, like fast radio bursts, so interesting. So for instance, the furthest signal we've detected is called FRB (fast radio burst) 20220610A, and that millisecond length signal came from a source with output energy equivalent to decades of the Sun's entire output.

This "law" is only for point (or spherical) sources, i.e. those emitting evenly in all direction - the area is increasing as a square of distance and thus signal power drops accordingly. With lasers, directional antennas, phased array antennas [1] the signal won't decay that fast. For instance with lasers it will be just a matter of alignment of internal elements to obtain a parallel light beam which doesn't lose power over distance (obviously there are other factors - atmosphere, particles in the vacuum, et al which will result in diveregence anyway).

In fact some billionaires [2] invest into using telescopes with fast sampling cameras (in this case IACT [3] telescopes used normally to detect gamma-rays by their interaction with the atmosphere) to detect flashed of extraterrestrial lasers.

[1] https://en.wikipedia.org/wiki/Phased_array (see the animation of the radiation pattern)

[2] https://www.space.com/are-aliens-flashing-laser-beams.html

[3] https://en.wikipedia.org/wiki/IACT

>This "law" is only for point (or spherical) sources, i.e. those emitting evenly in all direction

We're not talking about lasers and directional antenna here, we're talking about humans using radio communications to talk to each other on Earth over the last 100 years, and whether that's detectable from great distances.

>The team discovered that a single chip responsible for storing a portion of the FDS memory — including some of the FDS computer’s software code — isn’t working.

No one thought of having backup computers on the spacecraft?

Most of the computer hardware is duplicated but much of it has already failed/failed earlier - remember these are 50 year old hardware - a lot of the logic is TTL and discrete components - which are far larger than modern equivalents would be.

It's not just 50-year-old hardware: it's hardware that's been subjected to cosmic radiation for 50 years, part of that outside the solar system (so presumably even higher).

sorry, i just have a silly question: what would it take to send new probes out there? voyager 3 and 4 for example to follow the same path (more or less, sans planet alignment) V1 and V2 followed, but with better hardware of course.

You can't. Voyager's launch date coincided with a planetary alignment allowing for gravitational slingshotting out of our galaxy. We have to wait for the next alignment.

There's basically no point scientifically for doing the four-planet flyby again. Since Voyager, we've already done much better at Jupiter and Saturn with years-long orbiter missions. A quick flyby wouldn't get us anything we don't already know at those. Voyager gets the mindshare because it was first and the four-way slingshot is fun to visualize, but the reality is that Galileo and Cassini and Juno delivered thousands of times more data.

We could use more investigation at Uranus and Neptune, but we'd get much more out of extended orbiter missions to those rather than another quick flyby. A Uranus orbiter is currently one of the higher priority missions in planning, and there's a launch window for a Jupiter-Uranus slingshot in 2034 or 2035.

(What I wonder is, how much planning do these things need? Why can't we just launch another copy of Cassini to Uranus and skip all the expensive design? You'd need some changes to antennas and power supply for a more distant planet, but the scientific instruments and computing platform should just be reusable designs.)

Going out further than Voyager might be interesting just for studying the conditions outside the inner solar system. Though I'm not sure how much there is to observe unless you go an order of magnitude faster than Voyager to try to reach the oort cloud

If you mean just to beat them on distance, we could probably launch something capable of it in a decade or two by getting more serious about nuclear propulsion and using some slingshots to pick up speed further.

"The team discovered that a single chip responsible for storing a portion of the FDS memory — including some of the FDS computer’s software code — isn’t working. The loss of that code rendered the science and engineering data unusable. Unable to repair the chip, the team decided to place the affected code elsewhere in the FDS memory."

Just another proof that we may have gained a lot, but also lost something in our pursuit of modernity: on modern systems direct memory access is discouraged if not prevented by the underlying OSes, and this hack would not have been possible.

The modern equivalent to this would be an embedded system with an RTOS, where you do get full control of memory, because you are the OS. We just have nice abstractions on top of that for the most common use cases, since you very rarely need that precise of control over system timings or memory allocation.

Modern systems can automatically detect bad memories and map those hardware pages out. SSDs can do it at the firmware. ECC are also self correcting.

The "hack" wouldn't be necessary, or can be done natively in many modern systems.

What's the paper title? "goto considered harmful considered harmful" I think?

Even in modern C programming goto is still pulling its weight for handling unrolling and cleanups.

The paper was written by Dijkstra, and even he doesn't like how it's become some kind of mindless mantra, instead of a warning against spaghetti code which it was. He never meant that you should never, ever use GOTO.

Next time I bitch about debugging something in a container I'm going to look at this and stop bitching. Great job!

（评论） (comments)

（评论）
(comments)