Why satellites are manufactured in clean rooms | Hacker News Reader

Why satellites are manufactured in clean rooms

ohb.de

126 pointsricw44 comments

This article is missing some big details. Dust is a concern for sure, but it's way more complicated than that.

Basically most satellites fall into three camps, communication, optical imaging, and radar imaging. The most sensitive of these are in the optical category. High performance optical satellites (think JWST, Hubble, Kepler, KH-x series, worldview, etc) are absurdly sensitive to contamination. The clean rooms exist in no small part to keep people from using or introducing substances and chemicals that will destroy mirrors, lenses and focal plane arrays (image sensors). By far the biggest threat is anything silicone related as in a vacuum environment, silicone will off-gas and coat optical surfaces fogging them. If this happens on the ground during thermal vacuum testing the satellite would need to be completely disassembled and decontaminated. If it happened on orbit it would effectively kill the satellite. As such volatile compounds that are likely to emit any vapor are carefully controlled during assembly.

The other side of the coin is electrical static discharge. Pretty much all satellites are extremely susceptible to this including the optical sats as the focal plane arrays are absurdly sensitive. The white suits you see in the cleanroom are not only to contain dust but also electrically conductive to ground any buildup of static charge. By maintaining a controlled environment with trained workers you can combat this much more easily along with humidity control to reduce buildup of charges. It's worth noting that the charge levels at work here are far lower than the ones you've experienced on doorknobs or are worried about assembling ie. a computer. As little as a few volts of buildup can destroy or degrade some parts and the normal human body can almost instantly build up hundreds of volts just moving around under normal circumstances.

Finally a big issue is large FOD (foreign object damage) and having a nice controlled environment is beneficial for keeping it under control. Stuff like hair and wire clippings would be common examples. You need to control that stuff as it can either short out circuits or be heated and vaporize to coat optics/controlled impedance circuits. Everything in space behaves like a giant vacuum plating chamber so the less stuff floating around the better.

You're not wrong, especially about optical components. But there has been a general overreaction about ESD and milspec/spacespec components since, I don't know...maybe the mid-1970s budget cuts? And we have a lot more knowledge about cosmic rays and other bit flipping events now.

It wasn't unusual to launch a Pioneer 10 and 11 or Voyager 1 and 2 back in the day both because the launch vehicle wasn't super reliable and also because the overall system and the environment it was going to wasn't fully known.

Now we have Hubble, which initially a disaster but after multiple manned servicing missions, was turned into a glorious scientific instrument. But we also have a couple equivalents sitting around in a warehouse[0] that could have been launched (at least now) for much less money. I'm very concerned that JWST will be the next.

[0] https://www.space.com/16000-spy-satellites-space-telescopes-...

I'm not sure what you mean. ESD is a big deal with sensors, namely optical FPAs and ADCs for radios. Their voltage tolerance is well established and the ESD generated by humans is easily measured.

It's not that this stuff isn't necessary or we don't understand it so we overreact, it's a calculated risk decision that's currently changing with cheaper launch costs. Satellites are starting to get smaller as the components are being miniaturized much like most tech and launch costs are falling. Small-sats are frequently built in much less strict clean environments and instead of massive controlled transporters, are shipped in crates by Fedex. It just depends on the reliability profile of the mission and the type of instruments in use.

ESD is a huge deal with passive component too, and honestly even with components that you never thought could fail in the past like resistors, lately.

Obviously an instrument that is never going to be serviceable deserves the best components and handling, but they will fail eventually and you either have a redundant circuit or another fallback.

I guess my only point was that I've seen both approaches - you either over-spec the components and when it doesn't fail like you think the spec was the secret, or you design for unreliable parts and when they don't fail as expected, you look like a genius.

In addition to the above, the effects of ESD are cumulative. You may think you're getting away with not following correct ESD procedures until one day something stops working and you're left wondering why, because nothing you did has changed.

One of the “equivalents” will be launched in a few years as the Nancy Grace Roman Space Telescope (formerly WFIRST). It isn’t quite an equivalent: it was designed to image the Earth, not distant objects. It has a wider field of view than Hubble, but is less useful for imaging distant galaxies and exoplanets in detail. Unfortunately it also doesn’t deliver the expected cost savings, since the telescopes came without instruments or electronics. Different people at NASA made different estimates, but most showed the NRO telescope as being roughly a wash compared to the earlier WFIRST concept, and it has overrun its cost substantially since then.

If these satellites were designed to point at Earth, why didn't they donate them to NOAA, instead of NASA? Or why isn't NASA giving them to NOAA now?

A condition of the NRO donation was that the telescopes must not be used for Earth observation.

I spent the better part of a summer internship scrubbing vacuum chambers with nasty solvents at SSL. They were in the process of retrofitting their testing chambers, but something in the new equipment was off-gassing. After a lot of trial and error (and scrubbing) we narrowed it down to the spacers on the new thermal equipment. Wasn't quite how I had hoped that summer would have went, but was interesting none the less.

Since anti-satellite missiles could cause debris issues for everyone, would it be feasible for an adversary to launch a silicone payload at an enemy satellite, with the intention that any optics would be rendered unusable through some electro-static attraction between the satellite and the silicone?

There are probably better methods, such as using electromagnetic weapons, or attempting to capture the satellite.

https://en.wikipedia.org/wiki/Sokol_Eshelon

https://www.esa.int/Safety_Security/Clean_Space/Setting_a_sa...

Not the worst idea really. There's far more worrisome stuff than silicone, like a small can of black spray-paint which would simultaneously obscure the optics and solar panels while also degrading thermals causing the sat to overheat itself if it still managed to function.

FOD... It would be pretty depressing to launch a multi million dollar optical imaging satellite, only to discover an 300 mile long eyelash obscures every photo.

It wasn't contamination related, but something similar happened to Hubble in the 1990s.

https://www.nasa.gov/mission_pages/hubble/servicing/index.ht...

> You need to control that stuff as it can either short out circuits or be heated and vaporize to coat optics/controlled impedance circuits.

You don't even need to be working on satellites to observe this. For example, car headlight bulbs have a much shorter life if you touch the glass bulb with bare, oily skin during the installation.

This came to my mind as well. I've worked on satellites in clean rooms and have also had cause to change the bulbs in my car's headlights. The packaging on the bulbs made it clear that touching the glass was not good.

They aren't satellites and they haven't made it even close to orbit yet, but I recently had a chance to make it down to Boca Chica, TX recently where SpaceX is building rockets.

Granted the incredibly complex engines are built off-site and only installed at Boca Chica, but the entire facility is a gigantic dusty, oily, noisy construction site with guys in hard hats and driving lifted trucks. My kind of place, tbh.

In college I worked at a physics lab that built an instrument that was on the Cassini probe to Saturn. The amount of effort that went into the supply chain and maintaining even jellybean components under a constant nitrogen atmosphere (to prevent any oxidation) was crazy. Meanwhile, Ingenuity[0] is flying around Mars with running Linux and literally off the shelf components[1].

[0] https://mars.nasa.gov/technology/helicopter/

[1] https://www.zdnet.com/article/to-infinity-and-beyond-linux-a...

> Meanwhile, Ingenuity[0] is flying around Mars with running Linux and literally off the shelf components[1].

Ingenuity was built in a clean room. It uses some off the shelf components mated to space qualified components. You're describing it like it's a drone from Best Buy that was duct taped to the rover at the last minute.

Just to be clear, the article says only one para about the heading "DUST REDUCES THE RELIABILITY OF SATELLITES".

Other than that, it talks about cleanliness and standards.

thanks for the TLDR

That was so annoying! It was all I wanted to know.

Yep, me too. It read like a damned advert for clean rooms towards the end.

> Once launched into space, satellites can no longer be serviced and contaminations increase the probability of malfunctions occurring during the planned lifetime. Even a single dust particle can interrupt a circuit.

How does a single dust particle interrupt a circuit? (I know nothing about this; it's interesting.)

If it's conductive at all it could bridge contacts on a circuit board.

Due to https://en.wikipedia.org/wiki/Whisker_(metallurgy) I believe leaded solder is still the norm for aerospace applications.

It is, also in my experience boards get a conformal coating to minimize baby risk of FOD being an issue with the electronics.

Yeah they do a lot of individual things to bullet proof satellites because you can't repair them and they cost millions (generally, cheaper satellites are starting to be used more and more).

They do, but for small scale projects like satellites we tend to push con-coat until we have a good confidence the unit is working. Removing conformal coating to replace a failed component or change a resistor can be quite tricky, especially for epoxy based coatings. So there is a long time without any coating (up to spacecraft assembly).

I'm assuming it's the connectors and/or soldering. Given the pressure difference I imagine you could see a connector blow off or solder joints explode.

I still imagine it's insanely overkill, but when you're paying millions of dollars to send something into orbit, you probably want to make sure it works well first time.

With regards to malfunctions, I imagine this risk would actually be better mitigated with (fail-safe) circuit redundancy. After all, radiation is going to be causing bit flips and all sorts of weirdness.

Space is only ~14 psi of pressure differential from sea level, far less than in your car tires. Nothing is exploding. The issue is mostly off gassing contamination of critical surfaces like optics and sensitive circuit boards.

Added circuitry is not free. Every gram or redundancy is one less gram of mission payload. Added circuitry is also increasing the number of failure modes.

That's not to say redundancy is bad or should never be considered but it needs to be weighed (literally and figuratively) against every other part of the mission. Optimizing for mass and mission capability means manufacturing will get more expensive.

Material science is hard. Conductive contamination will destroy circuits in unpredictable ways. This also makes debugging and future iteration difficult as the satellite cannot be inspected.

https://www.ibm.com/docs/en/power9/0009-ESS?topic=floors-con...

it might also attract other particles to it, growing it in size and potential to obstruct etc...

Not an expert but I'm acquainted with someone that builds high vacuum stuff for physics experiments. Cold welding is an issue. Essentially if two sufficiently clean metal surface come in contact in sufficient vacuum, they weld together without any heat or liquefaction. So if you had conductive metal dust shavings contaminating the sat, once it's in space they may find their way to a spot where they can create a short circuit. Apparently the welds behave pretty differently from ordinary welds like allowing movement.

I also don't get this. Most circuits are usually vacuum sealed already. Hack, I can drop my iPhone in water, let alone dust. Reminds me of old days when people used to insist that PCs must be kept in air conditioned rooms.

That's a bit of an exaggeration for sure. But dust can be an issue in connectors, making the connection not as stable as expected or even break the connection on a pin if there is enough.

There is also the risk of short circuit on non-coated PCB, increased ESD risk. And beyond PCB, there are a lot of other risks mitigated by clean room environments: dust on sensors, dust on bearings or gears, dust on optical cameras... And any deposit of dust will float around everywhere once in orbit, especially after being shaken and vented during launch.

Could you make a satellite in your garage without a clean room environment? Yes sure, and you would even have a decent chance to have a working satellite once in orbit if you know what you are doing. Would you accept a 50% failure rate if you are flying a payload worth a few millions and need to spend a similar amount on launch cost? Probably not.

All this sounds insanely expensive. Just makes me double down on the fact that it is cheaper to make disposable satellites than make them last forever. Especially if it is cheaper to get them into space as well.

The problem is that high performance optical systems are expensive.

There are physical limits that forces them to have big mirrors and other expensive and relatively bulky optical elements. Also, there is no mass production due to small batch sizes (typically one or two).

So if they have a two year life span, it will not be economically viable.

It depends on the orbit, the function, sensors, and the size of the satellite.

If you want to go to an L2 orbit with a 10,000 Kg satellite with one of a kind sensors (relatively speaking) for earth observations, disposable doesn't make sense.

Low earth orbits with low weight satellites, pico/nano/micro/femtos - cheap disposable constellations with mass produced widgets make sense, particularly when you can launch them on another mission.

I hope this article ages badly. We're at the tail end of the "mainframe" age of satellites. Expensive, long development times, low-risk projects. Hopefully with launch prices dropping, satellite development can become cheaper and iteration can become faster. We'll likely always require clean rooms for optical satellites but this doesn't have to be the case for other types.

For optical, I'd love to see a comparison between standards for terrestrial use mirrors vs satellite ones. Surely the ones used on earth are manufactured well, but end up with dirt on them anyway? Is the difference because the earth based ones can be cleaned earlier during use? Would a future path to better ability for optical imaging satellites to self-clean be a game-changer?

How does the launch prep, launch, and deployment compare to the assembly of the sat? I was under the impression that eg fairing separation likely is quite messy, perhaps even metal flakes dropping of due to abrasion and such. Silicone off-gassing perhaps not so much of an concern then.

Great article!

Reply on news.ycombinator.com