Hacker News new | past | comments | ask | show | jobs | submit login
Starlink – SpaceX’s broadband internet system (starlink.com)
533 points by mulcahey 26 days ago | hide | past | web | favorite | 359 comments



So, part of Musk's long-term play here has got to be, eventually, to include a Starlink terminal into every Tesla, right? Probably without even charging any subscription. You and your passengers get free broadband internet in the car, and Tesla get a high-bandwidth channel to every vehicle on the road, allowing them to stream live, high-fidelity telemetry data back to their HQ and push out software updates as and when they need to.


It's far more ambitious than that.

In the developing world, we've seen a leapfrog effect in telecoms - they skipped over fixed-line phones and went straight to mobile phones. It's far easier to blanket a rural area or a dense slum with 3G signal than to run wires to every house.

That poses a problem when you need fast fixed broadband - there's only so much mobile spectrum to go around, you can't really cheat the Shannon-Hartley theorem, so you end up with tight data caps and unreliable performance. Rolling out fixed infrastructure can be prohibitively expensive, so a lot of areas are stuck on mobile-only.

Low-orbit satellite broadband potentially breaks this dilemma. The low orbit provides tolerable latency (within 30ms of fixed broadband), there's an abundance of spectrum in the Ku and Ka bands and you can steer a tight beam with a phased array. Starlink and OneWeb could potentially offer a satellite broadband product that is competitive with fixed-line broadband on cost, latency and bandwidth, but that is available almost anywhere on earth. Becoming the default broadband provider for half the world's population is Kind Of A Big Deal - maybe enough of a big deal to bankroll a Mars mission.

I think there's obvious potential for this to be Kind Of A Big Deal on a purely personal level. Is your local cable monopoly absolutely terrible? Within a couple of years, it might not be a monopoly any more. Huge amounts of rural real estate becomes a heck of a lot more attractive if literally everywhere has good broadband. This might just be one of those once-in-a-generation technologies that meaningfully changes human geography.


I am part of a group setting a tiny house community. We found being more than a hour from any major city center was not a problem for most of the people interested in our setup.

But limited mobile phone service and no high speed internet limits a lot of people. StarLink hopefully comes along by the time we start our planned future expansion.


How big is the project? Have you looked at running your own fiber? I imagine the residents wouldn't mind not having to manage their own WiFi...


We are at South River Ontario Canada.

At present we are three owners each building own house, we have 5 different people interested right now joining us to build their homes and our planning is no more than either 50 quarter acre or half acre lots which will let us leave about 15 acres untouched.

The real short term is just Wifi, with ethernet cables for the further homes. Once we get more people, we will look into fiber for everybody.


How can I find out more about your project? I am in the area, and this sounds very appealing.


How did you get together to build this community?


Point-to-point Wifi is powerful and is being deployed as an alternative to fixed lines in some places. E.g. see the Ubiquiti products. https://www.ui.com/products/#airmax


I had to check that the Kind Of A Big Deal wasn’t a GSV.


Obviously, this means that we need to advance technologically to the point that we can fix that problem.

By building a GSV and naming it Kind Of A Big Deal.


.


I want to add that self driving cars, plus high available broadband, might mean a lot of people moving back to rural areas.

With self driving cars you can go out, drink alcohol, socialize, and get back home whenever you want, you can even sleep in the car


Could this lead to low population density cities?


Not a chance. People will still live in the cities. It just means that city which went from 5,000 people to 2,000 people in the last 20 years might go back to 5,000.


sleep in the car and not pay hotel tax or property tax? travesty.


Not for long


live in a country without hotel, property and all the other stupid taxes? What a nonsense...


> you can't really cheat the Shannon-Hartley theorem

No, but you can work within it's limits to achieve some fairly spectacular things.

I predict that if low-orbit sat service starts bringing competition to rural areas, we will see the cell companies start using spatial diversity antennae to dramatically up the available bandwidth.


I'm not sure how solving problems with multi-path and/or LOS helps here. The fundamental problem with rural is that the $ invested : customers served($ out) is drastically lower as population density diminishes.


Spatial diversity does more than help with multi-path issues. It multiplies your available bandwidth by sub-dividing the space around the base-station.


Bandwidth doesn't help you with HAAT[1] and the fact that you need to build out towers every couple miles or more depending on your terrain.

Towers are not cheap to build or maintain, if you only get a couple hundred people per tower it may not make sense to do(and why you don't see a lot of community WISPs except in unique cases).

[1] https://www.fcc.gov/media/radio/haat-calculator


> Low-orbit satellite broadband potentially breaks this dilemma. ... Starlink and OneWeb could potentially offer a satellite broadband product that is competitive with fixed-line broadband on cost, latency and bandwidth.

But it doesn't, despite the marketing of Starlink as a "broadband" network they simply can't service high population density areas.

So it might be a backbone to a 4G tower and possibly usable by consumers in rural areas, but it won't be installed on the roof of a car nor offer connectivity to individual consumers in urban regions.


SpaceX are planning on launching a lot of satellites - 2800 Ku/Ka-band satellites at 1150km and 7500 V-band satellites at 340km. Both will be capable of steering a 1.5 degree spot beam.

I'm not saying that you're definitely wrong, but Starlink is designed to provide a truly phenomenal amount of throughput. The V-band VLEO constellation is clearly intended to serve urban areas.

https://licensing.fcc.gov/myibfs/download.do?attachment_key=...


At 340Km, a 1.5 degree beam covers a circle of just under 9Km diameter. Let's call it 50 square Km.

Population density in Boston is over 5000/sqKm, so 250,000 people have to share that satellite. If 10% are subscribers and they use it 10% of the time, that's 2500 simultaneous users.

Starlink has claimed 20Gb/s per satellite, which gives 1 Gb/s per 125 simultaneous users in this high-density area. That's not great, but it's also not awful.

In any less population-dense area, it can be reasonably competitive, especially as the end-user deployment model is expected to be "here's a pizza-box sized antenna: make sure it's pointed in this general direction of the sky, and give it power and ethernet."


> which gives 1 Gb/s per 125 simultaneous users in this high-density area. That's not great, but it's also not awful.

thats per satellite... wouldnt you, in theory, have access to multiple satellites? i though the theory was overlap, etc...


Yes. I seem to recall either Elon or more likely Gwynne Shotwell saying they're shooting for three or more satellites in view for the system to truly work the way it has been designed.


3 satellites makes for 24 mb/s per person. Except—I think it's actually 20 Gbps per satellite, so 24 mbps per person, so really only 3 mb/s, which is...usable, but slow.


Reasonably competitive? The competition is non existent in most of the inhabited world. In addition I have a friend on a farm in the US who can’t get a reliable internet connection in spite of the Sprint fiber that runs right in front of his house. This would be a no brainer for a lot of the population. Starlink could have more customers than ATT and Vodaphone combined.


Starlink satellites spend most of their time above sea and other uninhabited areas.

If I remember correctly, Starlink could potentially serve roughly 40 million customers globally in far to reach areas. Including ships and aircraft. It's not economic solution for the urban population.


> but it won't be installed on the roof of a car

How come?


It won't be a reliable connection because the car moves around. Stuff like trees, tall building, bridges and tunnels will disrupt the signal. So for cars, a ground based cellular connection will work more reliable.

Antenna alignment could also be a factor, which becomes more difficult if the antenna is attached to a moving object.


The antenna has to constantly realign anyway due to how fast the satellites in low earth orbit move. Given that they're using phased array antennas, which can be steered electronically nearly instantaneously, there should be no issue tracking a satellite from a moving vehicle.


As mentioned earlier by dsr_, at 340km, a 1.5 degree beam covers a circle of just under 9km diameter. Let's call it 50 square km.

If a single satellite can cover a 9km diameter, I think it will be just fine for vehicles. Especially so as the idea is for overlapping satellites. Also, much of the starlink bandwidth will be used for extra cellular / wireless / internet backbone areas. Areas that need long haul fiber optics to have bandwidth suddenly won't, so it will actually improve existing internet service providers as well since their addressable market will increase.


Because it won't work in high-density urban areas.


What will it mean to youtube, facebook, google, amazon, netflix, if all those sparse clients with slow internet become a broadband client... And, even more, a client that is driven away by some sort of capping? Of redirection to another site? Ads from another company?

Because net neutrality isn't exactly a thing everywhere, worldwide.


It'll just be a new monopoly owned by SpaceX. Great


I read an HN comment before that convinced me it's physically impossible for this system to work for more than a couple million devices at VERY low bandwidth (I think the quote was like 0.1mbps). Is this true?

I was all aboard the SpaceX hype train until that comment. I remember looking into his argument, and it seemed like everything checked out.


I'm sure that very smart person thought of problems in 5 minutes in an internet comment that the engineers who just put 60 satellites in orbit with didn't consider about their total addressable market...


I mean if you live in a place where you just dont get internet even 100 kbps can be game changing. Sure you cant stream netflix. But being able to some degree go on the internet is pretty big.


Did the arguement stem from the problems of sharing bandwidth due to free space broadcast (instead of wires where everyone can re-use the same bandwidth)? Because, if so, the spot beams for the Starlink sats down to the ground terminal are only ~15 km in diameter. The uplink back to the sats may be a bit more constrained and won't scale. But downstream should be fine.


Satellite internet is already available in rural areas and it doesn't seem to make that big of a difference. Most people care more about the culture that populated areas provide more than internet speeds.

Personally I pay $30 a month for 15 Mbps. I remember EarthLink or Prodigy being like $15-20 a month for 56Kbps. I remember leaving the modem on all night to download a 30MB demo of Monster Truck Madness in Elementary School. Now speeds are ~250 times what they were and the price has only doubled in the past 20 years? That's f-ing amazing. Prices are not getting out of control if you stick to the bottom end. I'm not even sure what I'd do with gig internet.

For areas going from no internet to satellite internet it's probably going to be amazing, but going from high latency satellite internet to low latency satellite internet will not be a huge game changer.

Adoption of http2 would probably help the quality of existing satellite internet more and be cheaper than this endeavor. As everyone started moving to HTTPS, the handshakes take much longer on a high latency connection and multiplexing would help alleviate that issue.


>Satellite internet is already available in rural areas and it doesn't seem to make that big of a difference.

It absolutely makes a difference. Rural/isolated communities are being left behind in terms of access to what are now basic staples of modern life: inexpensive, high-quality VoIP and video telecom, diverse sources of streaming video and audio media, real-time online gaming, telecommuting, etc.

The satellite internet available to most consumers today is cost-prohibitive garbage. Most options are doing satellite downlink paired to dialup uplink, and even the stuff that does two-way satellite links provides abysmal performance. All of it usually has a $100+ USD per month cost. Compared against 3G mobile, the latter is a far better deal.


Just to be clear, I don't think anyone is not moving to a rural area because of the internet speeds.

I'm not sure if you're talking about the U.S. or not, but my parents have satellite internet and they don't even notice a difference unless the dish is full of snow, which I'm not sure is a problem Starlink will solve. They stream, use Skype, and wifi calling. I don't notice much of a difference when I visit. If I try to log into a remote terminal, sure, but I'm not doing that. Internet might be more expensive, but rural properties are cheaper.


> I don't think anyone is not moving to a rural area because of the internet speeds.

Hi, I'm your counterexample. I held out moving to a rural area until the internet situation got better. It's great out here with a yagi on the roof.


What type of RF service is that? I'd appreciate any specifics you care to share. Thanks.


>I don't think anyone is not moving to a rural area because of the internet speeds.

Definitely not true. It is a big decision point for a lot of people.


Even within cities a check of "does this have fast internet" is a check for many people before they rent or even buy.


> I'm not even sure what I'd do with gig internet.

We're going to find out and that's the most interesting part.


Agreed, but at this point in time I can't think of any "killer apps" for it. It seems silly to pay so much for something I wouldn't be able to use today.


I couldn't imagine replacing TV with streaming HD video or downloading free multi-gigabyte sample packs just to see what's in them before before multi-megabyte connections and affordable half terabyte hard drives. Most new or improved technology is speculative. No one knows what people will use it for 10 years later. Guesses rarely pan out.


That's a good point, but as gig internet is already widely available, I'm not sure the technology is what's holding anyone back.


Widespread gigabit is a recent development, and it's still expensive. Whatever technology takes advantage of it is a few years out. I had megabyte cable ~5 years before YouTube launched.


> gig internet is already widely available

I don't have it available and I live in an urban area. Hell, I don't even have a viable alternative to my cable company. There's a ton of bureaucracy around all of it.

Political nonsense aside, we rarely see the use for something when it first comes out. That includes really useful things. e.g. all of the super important inventions that were straight up accidents.

Sure, it could take years, but I can't imagine having faster communication between more people not having a huge effect.


No, that's not the plan. To work with these low-orbit fast moving satellites, the antennas would be then size of a model 3 hood, pretty fancy/expensive and probably more than a little power hungry.

The cars will use LTE, although the backbone of those base stations could very well be Starlink.

Maybe future iterations will have extra hardware for terminal friendly applications, but I'd guess you'll see level 4 autonomy in a model 3 before they can upgrade Stsrlink to do that.


There’s already an LTE modem ok every car, so in populated areas, you’re covered — no fancy satellites needed.

Never have needed a faster Internet in a Tesla and all passengers will be carrying their phones anyway.


Cars are able to drive outside of populated areas, so the satellites might be useful for that purpose


Teslas aren't really able to drive far outside populated areas.


???

People have driven them around Australia. Can't get much further away from populated areas than that.

e.g. https://www.topgear.com/car-news/electric/woman-drove-her-te...


watch cpg grey drive on a desolate road in Nevada. can't get much further outside of population in the lower 48 https://www.youtube.com/watch?v=_naDg-guomA


Neat video, really drives home how essential super chargers are to the usefulness of electric vehicles. Sitting for 12 hours in a trailer park to get another 200 miles of range is insane.

Also, this guy has some serious driving PTSD or something, he's so scared of cars that he thinks its dangerous to drive on an empty highway at night. Yikes.


You can only drive about 1h30m if you want to make sure that you are able to come back from where there are no chargers...


That's really a weird way of looking at it. Most people won't drive in one direction and turn back and refuse to use a super charger.

https://i.imgur.com/leKbJ4k.jpg (Map from April this year)

Most super chargers are within about 120 miles of each other, so at 65mph, you have about 2 hours between them. They've done a pretty impressive job building out the infrastructure


I thought that most people wanted to avoid having to stop their trip for charging while driving because it takes too long (I don't even like to stop for gas and it is much quicker)... it's weird that you would not choose the best solution to your problem (and spend more money while doing it), but I guess some people don't care about time or money.

Also, what are those white spots on the map? Unpopulated areas with no charger in a 2h radius?


Your idea of what a tesla is like is far off. My 2012 tesla can drive over 250 miles. If I really gun it in the winter in the mountains its over 200. Then 30 mins charging at a supercharger. The new S can go over 370 miles without charging. 60 miles an hour * 4 hours, people stop to eat or pee.


I'm talking about a round trips of course... so you have to cut that in half if you can't or don't want to charge while on your trip (if you want to be able to come back home).

Most of the car trips that I do are round trips... and I would guess that yours are too?

Makes more sense?


I see what you are getting at. I don't have to go round trip very often and worry about charging because I have enough range, but also because because there are chargers everywhere and I can charge on the way back if needed. Most trips I take are within round trip range of my car. of course it's more convenient not to have to charge.

plugshare.com is a good site that shows all the chargers. supercharge.info is a 3rd party site that shows tesla chargers only, plus where the ones under construction are. tesla.com has one too.

I go skiing a lot, and there's one thing you learn - you use a lot of energy going uphill. An ev has much more accurate energy usage so you notice, gas cars of course use more energy doing more work too. But the huge difference is going back, going downhill, your electric car basically glides down hills for free, and because you are recharging the battery when you do moderate breaking, I can drive 20+ miles back and end up with more range than I started out with.

In Seattle I can easily go round trip to the 3 ski areas around town.

Basically I don't worry about it because I could charge on the way back if I needed it, and in 6 years of day trips I only charged once or twice on the way home - the one I remember was when I went to mt rainier for a week with no charging.


> I don't have to go round trip very often

I'm willing to bet that you are an outlier


So.... you can drive 500 miles without charging?


Me, I can approach 250 miles at 60 on the freeway, but 200-220 is easy with skis, 4 people, a bunch of equipment.

The latest model goes 370. not 500.


You may not, but they may ;)


One day they could possibly teleop cars autonomously (streaming sensor data to servers through the starlink terminals, then processing and sending output data back to the cars through the starlink terminals). Obviously latency is a huge factor in this, but it could be possible.


Interesting possibility. I don't think it would make sense for teleop to be the normal mode of operation, since it would consume a ton of bandwidth, but maybe they could use human teleoperators to handle occasional edges cases between "level 3" and "level 5" autonomy, in particular for the autonomous ride-sharing network.

They could colocate the teleoperators at Starlink ground stations to minimize latency, and prioritize teleop traffic.


My initial reaction was to make a joke about this, but the math is on point.

At the speed of light latency to/from the satellite (3.5ms) and a vehicle speed of 60mph, that's about 4 inches of vehicle travel. Sounds pretty great-- probably better than almost any human driver.

At terrestrial latency (30-60ms), thats a few feet, maybe nearing 10. You're still in the margin for human reaction times.

Doubling that to account for compute latency and such is actually still in the noise for car brakes available today.

Cool how far things have come.


I think starlink is aiming for latency on par with terrestrial solutions (e.g. sub 30 ms rt).


I would also like to see low orbit shielded datacenters. These would have relatively high mass and streamlined design, which would allow for radiation shielding, and reduce the necessity of boosting orbital trajectory. They would also have very high bandwidth line of sight laser communications. This would allow them to migrate the state of virtual servers to the next satellite over. So applications consisting of a cloud of server processes can incrementally migrate to virtually "hover" at very low altitudes above the earth.

Assuming Starlink also works, I suspect the first company to make this work would make an absolute killing.


> I would also like to see low orbit shielded datacenters

Why? Build on the ground with a fibre, far cheaper and far lower latency.


Can't be served a warrant. Arguably superior physical security. Can have line of sight comms with the entire region, which are both more direct and have faster signal speeds than fiber.

For certain global applications, a low orbit cloud consisting of many datacenters has some advantages. The virtual servers could virtually migrate to "hover" wherever they are needed, or to wherever would minimize latency.


Do you have any idea how orbits work?


Do you have any idea how orbits work?

Yes. But your reading comprehension is spotty. Look at the cousin comment:

The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.


That's not how orbits work.


That's not how orbits work.

You're not understanding the proposal. The way the "hovering" would work, is that the entire state of one virtual server would be serialized, then sent via line-of-sight high bandwidth comms to the next satellite to go over the region of interest. This way, the entire application's server cluster could incrementally migrate to remain over one geographic area, but do so at a much lower altitude than geosynchronous orbit.


So wait, we're sending entire server states from one satellite to another every few seconds, rather than routing the data to the satellite when needed? Seems like it would be a trade-off that would be worthwhile only for a small number of edge cases where there's a small server size, a fast start-up time, and a huge amount of very local traffic. In which case, put the server on the ground near the traffic? Or alternatively, move just a bit of the data (caches and some processing) around.


Turn every supercharger station into a LTE provider then put a dish on it and you're there. I'm glad spaceX isn't on the public market though. So we'll actually have a chance of seeing some of these ambitious goals come true.


No, they can only service a limited number of customers within a given region. The only users of Starlink, et al. in major metropolitan areas will be backbone providers.


I suggest you count the number of nodes they can support with that many satellites. Stop thinking like an old tech.


"This system, if successful, would provide people in low to moderate population densities around the world with affordable high-speed internet access, including many who have never had internet access before," - Tom Praderio, SpaceX engineer

Source: https://www.scientificamerican.com/article/spacex-launches-t...


What point of that quote leads you to think they also can't provide high-speed access in non-rural areas?


I'm guessing the "low to moderate population densities" part?


He’s emphasizing the use case that matters. That he doesn’t mention something of little interest (yet another broadband provider in urban areas) doesn’t mean it’s not something possible.


Broadband in car all over US and Canada. Yes please!


Broadband outside of cars all over planet Earth; much more important both from an economic standpoint and a geopolitical one.


Very excited about this. What's interesting is the deployment mechanism, where they almost use the momentum of the rocket to "fling" the 60 tightly packed satellites away from the booster, and then after about 3 hours they fire up the Krypton ion thrusters and start the course correction phase. IIRC the only other times this deployment mechanism is used is on tiny CubeSats where they opt for this more conglomerating type of deployment due to the durability of the CubeSats.

Will definitely be getting service for some family members in Northern Africa where incredibly slow speeds are charged high prices if Starlink can compete on price.


Last night at the sat. launch there was almost no fling effect. They slowly floated away in a group, hardly separated. I thought they could have had some mechanical problem.


>where they almost use the momentum of the rocket to "fling"

this is compulsory anyway due to physics.


Not really, the rocket and the payload would have the same momentum unless an additional force is applied between them.


Why? You could also just use a spring to push it out.


And what happens to the rocket from the reaction force from the spring?


Nothing, because you used a spring to launch the other satellite on the other side of the payload at the same time.

Or, you just use thrusters, like the ones that launched from the Shuttle did.


> ion thrusters powered by krypton

> autonomously perform maneuvers to avoid collisions with space debris and other spacecraft

Pretty cool.

Also, the north and south pole are conspicuously avoided by Starlink, confirming my suspicions about the locations of intergalactic space ports.


This sentence reads like science fiction. I had to google “krypton ion thrusters” to make sure it wasn’t a joke. It seems it was not. Very cool.


> Also, the north and south pole are conspicuously avoided by Starlink, confirming my suspicions about the locations of intergalactic space ports.

Are you saying that there will be intergalactic space ports at the poles in the future?


There are various wacky conspiracy theories about Nazi remnant and/or alien UFO bases at the poles.

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


Hollow Earth is another bizarre one: https://www.reddit.com/r/hollowearth


I didn’t realize krypton was a fuel source. I thought Elon put in an easter egg, only to be disappointed by my ignorance.


The krypton isn't being used as a chemical reactant, and in fact being largely chemically inert is a benefit for ion thrusters. I suspect they're using krypton instead of xenon due to the latter being more expensive, even though the higher atomic mass of xenon would be beneficial.


Lower atomic mass gives better Isp, although for power-limited situation you usually want to maximize thrust instead.

In the pre-flight press conference the topic of krypton thrusters came up, and at first he made a superman joke, but then said the real answer was the lower cost of krypton.


Lower atomic mass gives better Isp for chemical rockets, for which the exhaust velocity is proportional to the temperature of the exhaust divided by the square root of the atomic mass of the exhaust product. However, these satellites are using ion propulsion. In ion engines, the exhaust velocity is constant, so a higher atomic mass allows each atom / molecule to carry more energy. This makes higher mass fuels like Xenon better for ion engines. (Radon would be better still but the radioactivity poses problems for various systems.)


Why is the exhaust velocity constant? I would naively expect the engine to impart roughy equal energy into each atom, so that lighter atoms would be faster. But I know little of the specifics of ion engines.


Thrust power = Isp * thrust. For the same amount of input power, increasing molecular size raises thrust and reduces Isp.


What does "Isp" stand for?


Probably meant specific impulse.

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


It stands for Specific Impulse, which is a measure of fuel economy for rocket engines.


Specific impulse. It is a convoluted method of measuring the exhaust velocity, and thus the efficiency, of a reaction engine.


It's not a fuel, it's a propellent (simply put, solar electric is used to accelerate the Krypton out the end of the thrusters)


Krypton is reaction mass. Sir Isaac Newton is the deadliest son-of-a-bitch in space.


Also,

> Starlink is the first krypton propelled spacecraft ever flown.

So they're using non-flight-proven technology (Technology readiness level at most 6)... impressive!


They might have gotten launch costs so low that a live test with the possibility of doing it over was cheaper than expensive test runs. Another comment mentioned SpaceX was close to losing some key license. Launching too late was probably going to cost more than the worst-case scenario cost of launching early.


Where do they get more krypton? Or they have limited fuel?


Limited fuel. Intelsat just lost a satellite due to sudden fuel loss. A big part of satellite management is fuel conservation.


Limited fuel, but ion thrusters are at least an order of magnitude more efficient (specific impulse) than the usual chemical propellants. The downside is low thrust, but these satellites can be patient when maneuvering between orbits.


Limited propellant (not fuel - they're not burning the krypton for energy, they'll have some electrical power source, presumably solar panels).


Fuel doesn't need to imply burning though.


Fuel implies a chemical reaction that releases energy, which is almost always burning. (Some people extend the term to cover non-chemical reactions, but in any case Krypton is not used that way either).


Limited fuel. Thing is that ion thrusters have very very high specific impulse so you get a lot of oomph from them, especially if all you are doing is LEO stationkeeping.


I think they have limited fuel. According to the site, at the end of life, the satellites de-orbit. Maybe they will be recovered and recycled, too? Either way, the Starlink grid needs to be replenished with new satellites from time to time.


> the satellites de-orbit. Maybe they will be recovered and recycled, too?

Deorbit is a fancy word for burn up in the atmosphere.

They are within earths gravitational pull, so if they have a hard failure or run out of fuel to the point they can't "push back" they get vaporized on reentry. It's a clever fail safe.


LEO satellites deorbit due to friction with residual atmosphere, not because of gravitational pull. Orbiting already means you are in an equilibrium with gravity.


> They are within earths gravitational pull

Every satellite going around earth is "within earth's gravitational pull". The starlink satellites are so low that they are inside the rarefied upper layers of the atmosphere, meaning they have low but constant drag pulling on them. When they stop thrusting, they will eventually fall out of the sky.


> the satellites de-orbit. Maybe they will be recovered and recycled, too?

They definitely de-orbit. Half of the reason their orbit is so low is to ensure that they de-orbit quickly once they are out of fuel (the other half is latency).


Future versions will burn up in the atmosphere, as required by the FCC as SpaceX couldn't guarantee the satellites could be contained to sea landings.


Those are actually where the secret entrances to the Lizard people's land are at.


Yes and flat Earth conspiracy about the south poll and North Pole


This is an amazing achievement. The pace at which SpaceX is doing things is nothing short of a miracle. Congratulations team SpaceX. I can only feel envy at the super cool engineering being done there.


From the page source:

     <meta name="author" content="Elon Musk">
Elon runs the show!


Now I'm imagining Musk hand-crafting the HTML...


What are those 'og:' and 'twitter:' meta tags for?


Open Graph tags. They facilitate the thumbnails/descriptions/etc. that automatically show up when you post the link to various social networks. http://ogp.me/

Without them, Facebook/Twitter/etc. have to guess what the most representative image and introductory paragraph describe the link. They often miss.


open graph or original gangster. You decide.


OG is for Open Graph http://ogp.me/


Too bad there's no /humans.txt


oh wow. that's pretty meta.


I see what you did there and I like it.


That's amazing,

As a foreigner living in China, I'm wondering if such network can be technically prevented from being used in covered land/countries? How cool would it be to finally get an unrestricted access to the Internet here.

But I'm probably just dreaming.


Most probably, to use Starlink, they would have to abide by local laws and regulation. So in the case of China, I don't think the connection would be any different from what you can find consumer ISPs today. Or Starlink simply skips China.


AFAIK foreign satellite phone operators don't modify their service over China, but satellite phones are illegal to possess in China. I'd expect Starlink terminals would also be illegal to possess in China.


There is existing satellite internet coverage over China. It is just illegal to own a device that can communicate with it without explicit permission.


There is existing satellite coverage over most of the world, it’s just expensive and with high latency (and relatively low bandwidth).


I don't think anyone was doubting either of those points.


It depends on how conspicuous the transmitter/receiver is. The authorities could just roll up and confiscate your antenna if such is easily visible and uncommon. Or deploy a jammer in your area. I think it would be about as safe as operating a pirate radio station.


What if the whole group of satellites overhead acted as one huge phased array? If that could be made to work, then it could be very hard to detect the uplink and downlink. This would be fiendishly difficult to pull off, however.


It's a flat box the size of a pizza box. If you have a building with a flat roof it should be pretty much undetectable visually.


But it'll also be a device that will most likely blast RF all over the place.

I can't imagine finding a transmitter capable of reaching LEO being all that hard for a country.


That makes me wonder if the signals from such satellites can be jammed by nation states.


Unless SpaceX has implemented high speed frequency hopping (which would 10x the cost of end user devices), they can easily be jammed. It isn't hard to overpower a single from space with a ground based transmitter.


Even if you implement frequency hopping it'll do nothing for China blocking it. The end users need to hop to the same frequencies to receive the signal, which means that a jammer can do the same.

This sort of thing works for militaries (e.g. for GPS) because you can assume that the receivers are secure, i.e. they're guarded by people carrying guns.

That's not really a viable plan when a hostile state can just buy a receiver with the built-in hopping algorithm from a retailer.


It works for (the US) military because they assign unique NET codes to groups and unique keys to transceivers. The frequency schedule is seeded using part of the key, the NET, and the day. The system is actually designed with the assumption that transceivers will fall in to enemy hands.

Starlink could implement something similar where the schedule is seeded using a key stored on the device, the day, and the receivers serial number (or other unique value). Supporting a million receivers is the same as supporting one, it is basically just one more input to a hash function.


It is also easy to control the direction your receiver see it's signals from, to properly jam a satellite you need to fly the jamming equipment high in the air on a plane/balloon. A ground based jamming will have to be very powerful or close to block a receiver that is pointed to the sky.


how about overpowering jamming signal aimed at the sat? genuine question, no idea about the physics here


Probably feasible, but would create international tensions, in worst case retaliation - many countries jamming foreign satellites or taking other hostile actions. Yuck.


aye. still better than shooting the bird down.


China can both jam the satellite signals or (as Musk said during a Q&A) blow the satellites up. Selling directly to consumers will not be Starlinks main line of business, it will be backbone connections to local providers. So even if Starlink is unwilling to do any filtering, the local ISPs will do the filtering.


> Selling directly to consumers will not be Starlinks main line of business

It absolutely will be.

What they have said is that it won't be Starlinks main line of business at the beginning, until the receiver costs have been driven down and the constellation is much more dense. At that point, they very much intend to provide internet access directly to millions of people.


I remember Neil Stephenson once compared the cutting of undersea fiber optic cabling to the use of nuclear weapons: once one nation does it, every one is going to do it, so no one ever does it.

So I think China will innovate and find ways to block access to starlink, without attacking the satellites.


> blow the satellites up

!

They don’t just sit over China. There’s no way they could just start shooting them down without repercussions.


China is the poster-boy of doing things without repercussions. They could shoot down any commercial satellite and suffer barely any consequences.


This first batch of satellites doesn't have the planned laser links for communication between satellites, so service in China will only be possible with a ground station in or near China. (The satellites just launched also only operate on two out of the three planned radio bands for connections with the ground.)


Why would they not include the inter-satellite linkage capability? That doesn't make any sense, given that was a solved problem in the Iridium constellation 20 years ago. That was a huge benefit of Iridium, in that it didn't have to rely on the "bent pipe" of just relaying data between a client and a ground station within the same coverage area. It was literally, "if you could talk to one satellite, anywhere in the world, you would have service".


They need to meet a FCC (?) deadline in order to avoid losing their licenses; probably that functionality wasn't ready in time.


Iridium does inter-satellite linkage with RF, Starlink intends to use laser comms. The advantage being much higher bandwidth, and not having to license the spectrum.

The laser comms are very much on the roadmap, however they were not ready in time for the first launch, and SpaceX has a hard deadline to meet for getting sats up there, so the first generation is going to just bounce your signal back to a ground station. (Which will bounce it to the next visible sat to push it forwards, unless it has fiber.)


Maybe that's why they need 12k satellites, and it also helps with latency if the signal doesn't traverse the constellation.


> it also helps with latency if the signal doesn't traverse the constellation.

No it doesn't. Satellite-satellite lasers can follow a direct route at the speed of light in a vacuum, much like those microwave towers used by high-frequency traders. Done right it'll be lower latency than ground-based fiber optics (which snake over the terrain and only transmit signals at about 2/3 the speed of light).


It can be legally prevented, although the enforcement mechanisms are murky. To the best of my knowledge they've never made a public claim either way.


I imagine the same countries[0] that don't allow satellite phones won't allow Starlink.

[0] https://www.outfittersatellite.com/Countries-with-Satellite-...


I believe the issue in f.ex. India is not censorship and population control, but rather a slow bureaucracy and in-fights between different government departments and the military. That said, things are changing in India. BSNL were just issued a license for offering IFMC services, including KA-band and L-band services.


Oh I so wish, but satellite communication is restricted (requires special license) in most countries. So this is still out of reach from the masses.


They won't be operating in China, as China will blow their satellites up if they open up service in China. Their govt cant imagine a world where people have access to non-state controlled media. What an Orwellian world we live in.


Destroying the satellites--or enough of them to sufficiently degrade Starlink's ability to operate in mainland China--would be a very expensive bit of overkill with a great deal of international blowback. Destroying one satellite with an ASAT weapon annoys everyone, but the consequences are somewhat limited; the debris from destroying hundreds or thousands in orbit is pretty much a casus belli. China still maintains enough authoritarian elements that it'd be easier to crackdown any dissidents using Starlink instead. Nation states have been tracking rogue radio signals by spies for decades, and they've got the resources to monitor for them.

If Starlink were to pose a problem in China, that's how they'd go about it. Destroying the satellites would be way too flashy. There's also the added benefit--from their perspective--that Starlink's mere presence would be a filter for their domestic surveillance; anyone willing to go through the effort of buying a likely illegal antenna, setting it up, and then trying to disguise it is the exact sort of person they'd want to monitor. Which is a pretty depressing thing to think about.


Ignoring the fact that there are already plenty of Western companies providing satellite internet coverage in China who haven't had their satellites blown up yet...


This is probably the biggest threat ever to autocratic governments worldwide - government independent communication. Whereas conventional broadband and cell requires government sanction (to build towers, control frequency bands, build infrastructure, censor messages) and conventional communication requires physical meeting and distribution, Starlink breaks their grip hold. If people everywhere have unrestricted access to a market driven www without being limited by autocratic governments and dictators, communication and rallying together will proceed unimpeded. Starlink should be setup with absolute transparency.


No, it doesn't. Starlink won't be selling ground stations into countries that don't authorize it. This situation has already played out with satphones in China, etc.


I guess it depends just how motivated and technologically advanced said autocratic nation was. If sufficiently so, they could probably interfere with the operation of the satellites somehow, or just threaten to shoot them down, if they don't "come into compliance".


Conversely, it could be used as an entirely uncontrollable tool to spread dissent and propaganda. The same issue has played out on the current internet.


> > This is probably the biggest threat ever to autocratic governments worldwide - government independent communication.

> Conversely, it could be used as an entirely uncontrollable tool to spread dissent and propaganda. The same issue has played out on the current internet.

No conversely about it. Who wants to control dissent and propaganda? The Chinese government. The Russian government. The US government, with the cooperation of Google, Facebook, et al. Not all dissent is good. Some of it will be downright bad. Some of it will be immensely valuable, however. The truth will out, so long as no one is empowered to decide "the truth" on behalf of everyone.

If the government and big corporations can lock out and stamp out everyone it deems to be "bad people" they can do it to anyone. No one should have that power, no matter how hard they promise they'll follow a motto like, "Don't be evil."


The lie flies and the truth walks =/


Slow and steady wins the race. Better to let the truth win eventually, honestly and organically, than to start fixing the contest.


I wish. The constellation is launched and managed from one of the most power-hungry and corrupt unregulated nations that exists, so the potential for bad is just as high as the potential for good.


Presumably once enough of the network is up, SpaceX will have an "all orbit" telemetry setup using their own network.That would free them from using the NASA network and maybe we'd get to see drone ship video that didn't always cut out at the landing :-).

One of my managers from Sun, Don Hoffman, left and went on as an early employee of Teledesic, a LEO satellite broadband company in 1990 :-). But the prohibitive cost to get to space, and the size of their satellites, ultimately doomed them.

Starlink is a good example of an idea that failed but could succeed now because technology has moved far enough forward to actually do what the engineers envisioned in a more cost effective way.

That said, its a risky move for SpaceX which has the potential to be considered as having a launch monopoly given their costs structure, and using that to get into adjacent markets is something that is the kind of thing that antitrust lawyers go after.


This was leaked a few years ago: https://si.wsj.net/public/resources/images/P1-BZ810_SPACEX_9...

This is the only thing that would allow SpaceX to have non-linear growth and ultimately fund their grander ambitions, ie Mars. Given the potential rewards, seems like a very worthwhile risk.


Isn't the drone ship video problem on the "ground" side? That is, the vibrations and motions the rocket induces ends up messing up the satellite dish's targeting.


The problem, in my understanding, is that the drone ship currently passes the video data to shore via a link to a satellite in geostationary orbit. Because the drone ship's dish has to be aimed within a very tight arc in order to communicate with the satellite, the vibrations from the first stage's landing burn causes the connection to be lost.

Because there'll be a ton of Starlink satellites up there once the constellation has been completed (and they'll be much closer than geostationary), the acceptable arc for the dish to be aimed should become wider, and it will likely become possible for SpaceX to use the Starlink network instead and maintain a good connection the whole time.


As I understand it, both the intermittent telemetry video from the first stage, and drone ships is the challenge of maintaining line of sight from the transmitter to an effective relay.

The antenna for reaching the starlink constellation will look more like the ones you see on the iridium handsets rather than the directional dish that is characteristic of geo-synchronous satellites. Regardless of orientation or vibration, the signals can be effectively transmitted (and received) from ground to orbit.

Now they still have the option of cutting off our access to the video when things aren't going in a way that is favorable to SpaceX of course :-).


Is there any information out there about what the terrestrial requirements to actually use this service are going to be? Is it going to require antennas similar to line-of-sight microwave internet providers like Monkeybrains? How much is that hardware expected to cost?

The satellites themselves, the way they're being deployed, what altitude they'll be in orbit...all of that is really fascinating, but I'm very curious what this is actually going to look like to use.


I don't think they've announced much regarding the pricing or specifics of the terminals, but Musk has said in an interview they will require an antenna array with line of sight, and will be about the size of a pizza box.


I think there was an interview where they talked about aiming to get the cost of a base station down to ~$200 (some number of hundreds, think it was two). That was probably cost to SpaceX not an end user though, and it obviously doesn't include cost of service.


Most people probably won't buy the terminal outright, though, I expect? You'll just buy a Starlink subscription with a fixed initial term and they'll lease you the terminal as part of that.


People pay $1000 for an iPhone. I don’t think $200 upfront is onerous.


People in areas this is meant for don't pay $1000 for iPhones.


Airplanes, cruise ships, ski (and other) resorts, rich hermits, expensive summer camps, consumer trains, etc all do. For lots of commercial activities a thousand dollars is a rounding error compared to the costs of their other equipment too.


No, but they don't need to each buy one - they can pool their money and have one receiver for an entire village, and that may be much better than what they have now, which is either nothing or spotty 3g access


Same as existing terrestrial BGAN-terminals in other words.


So by the end of the year (best case) they will start commercial service after 6 launches. And plan on 2-6 more this year.

This could have a huge impact on people wanting to live out of a van remotely and still be connected.


Or a boat.

When I can get cheap high-speed low-latency internet in the middle of the ocean then I'm buying a boat and going to sea.


I do wonder how well Starlink's antenna will handle roll. Current VSAT require rather large boat, but this is probably different.

I can see price for older boats increase purely due to Starlink. It's already somewhat cheap way of living and I do hope special tax provisions are made for people out in the ocean (I don't use any social benefits for paying my taxes in country X...)


Stationary Starlink antennas will already need to handle aiming at around one degree per second just because of how quickly a satellite directly overhead will move. So the roll of a sailboat in a marina should be no problem at all, and maintaining a connection while underway is probably doable without much redesign.

And that's assuming there's any difficulty to aiming. Phased array radars have shown that merely sweeping the beam around in a predetermined pattern can be very quick. I don't know how much harder it would be to adjust that based on input from gyros.


It’s a phased array antenna. Unless your boat goes sideways it will probably get a signal.


It does go sideways, all sailboats do. But usually less than1 knot when sailing 5-10 knots.

Roll on monohulls is pretty big. Current generation of Youtubers seem to have switched to catamarans, which makes me too envious - I am not sure I can save up that much before responsibilities catch up on me.


I'm far from an expert on this, but I believe the biggest benefit to phased array antennas is that they don't have any moving parts, and adjusting where the signal is pointing is almost instant.

As long as the "pizza box" can track the sats, it should be able to communicate through some pretty intense speed and acceleration changes.


> I do wonder how well Starlink's antenna will handle roll.

I'm thinking some sort of stabilization technology similar to how they manage to land rockets on ships. But not rocket-powered, of course!


Sure just mount it on one of those steadycam gyro gimbals.


Probably don't even need the gyro, just a weighted gimbal.


Or a boat.

Or an anarcho-capitalist hyper libertarian micro-nation perched on top of a spar buoy.


I'm no expert, but I would think that the first customers will be the same people using the very expensive but low-latency terrestrial microwave links. Though now that I actually researched it, I don't know if Starlink will be able to beat 8.5ms round trip of this Chicago-NYC link.[0]

[0]https://www.zerohedge.com/news/chicago-new-york-and-back-85-...


Not likely. The microwave paths they use now are surprisingly close to optimal straight shot. An additional 400 km up and 400 km down is going to blow out the latency budget.


What they could beat is across oceans. Say NYC to Shanghai or London.


Am I missing something? I feel like this website has no actual information on it. It's just pretty pictures and a bit text that doesn't talk about anything new.


The first 60 satellites launched last night.



Does anyone know what that pulsing visible in the mylar of the second stage was? I've watched several SpaceX launches, and never noticed such a long-period (about a second) repeating 'thump' before.

https://www.youtube.com/watch?time_continue=504&v=riBaVeDTEW... is when you can first see it. It's visible in the top view looking at the satellites as well.


I noticed that too! I texted a friend that works at SpaceX asking why does it look like the second stage has a heartbeat. I don’t have an answer but I’ll post back if I get one.


At 18:00 it looks like a membrane inside the satellite stack is also pulsing?


But what is the pricing?

Is it a $8000 base station + $200 a month for a 1 GB capped connection?

Or a $500 base station + $50 a month for 1 TB of bandwidth?

I hope for the latter but fear the former.


I wish they offered sensible guaranteed minimum bandwidth and didn't oversubscribe their links too much and included good QOS/traffic shaping with maybe prioritizing game traffic (or udp) for lower latency gaming.

I'd be more than happy getting say a guaranteed 5 MB/s (~50 mbps) steady with super low latency on the cheap everywhere on the earth. I'm currently getting about 1-2 MB/s on 4G mobile internets for ~25 eur a month but at least it's _real_ unlimited.


Yah I have no need for real unlimited, and would much rather have some better latency than people doing torrents can get more stuff. So basically.. I am all for sensible data limits.


If I had to guess, it'll be $200-300 for the base station and $40-50/m for 100mbps.


Have they said how traffic will be routed? If I'm in, say, the US, and I'm connecting to a web site in Australia, would the satellite I'm connected to send my traffic to a terrestrial router near me, and let it traverse from there via regular ground links to Australia? Or would my traffic travel via satellite to a terrestrial router near the destination?


My understanding is this first version will route traffic back to the ground and terrestrially, because they don't contain interconnection links.

Future versions will have laser interconnections and route traffic through space.

https://hackaday.com/2019/05/20/everything-we-know-about-spa...


The whole point of the network is to move the data through space because light moves faster in space than through a fiber optic cable, so I'm pretty sure they are gonna try to get it as close to the destination as possible


Not only is this technological amazing, hopefully it will provide competition to areas with monopolistic ISPs. More choices, lower prices, faster speeds!


Considering that the market for this product is people with subpar internet connections, it might be worth considering building a website that doesn't require nice hardware and a fast, stable internet connection to accurately load the page :)


The audience for the product is not the same as the audience for the website.


That's fair, this is probably more brochure to generate buzz, etc. It' still a little ironic, though. I got a chuckle out of it anyway.


Still can't watch 4K YouTube videos. Oh, the humanity!!


Apart from the launch method - how is this going to be different from existing VSAT deployments? They're quite expensive (starting from ~thousand dollars per month):

http://www.melatnetworks.com/price-list.asp

Or those, with prices around 50usd/gb:

https://www.vsat-systems.com/service-offerings/VAR-VNO-Deale...

I doubt every Tesla will have one on board. LTE is so much cheaper.

Quite useful for medium and large marine vessels, or truly remote locations tho.


Their satellites are in Geostationary orbit (about 22,236 miles away). This constellation is at 400 miles.

22,236 in Lightseconds: 0.11936716 (one way) 400 in Ls: 0.0021472775.

(assuming directly overhead)

~250ms round trip for GEO vs 4.2ms for LEO


I believe the primary difference between starlink and other satellite internet is that it will be lower down.


In Ireland we've just agreed a 5 billion euro rural broadband scheme which will involve laying fibre that goes to 540,000 rural homes. 3 Billion is the cost to the state, 2 Billion is the private investment and the private company will own the infrastructure after (I think) 20 years or so. The whole thing is going take several years to become functional and I think 7 years to be fully functional.

So my question is: how likely is it that something like Starlink could provide an alternative in any meaningful timeframe?


If you ask me, not likely. Simply because the Starlink satellites will need to be replaced regularly. And thus will be expensive. While your terrestrial fibre is always there and easy to repair. Launching and building 12k satellites will also be very expensive and time consuming to begin with. But than reusing rocket boosters requires a certain number of launches to be economically feasible (I don't remember the correct number from the ESA (?) study I read a while ago). Just how Starlink will help SpaceX reach that number of launches and whether this results in overall profitability is the real question I have. But I guess this data will never be published...


Starlink will not just serve one country so the cost per country/customer goes down a lot.


But not enough to compete against efficiently deployed fiber in rural areas where you’re not ripping urban streets apart.

It’ll be amazing for mobile users and developing countries though.


SpaceX cost for global coverage is about $2Bil, which needs to be replaced every 5 yrs. So about $.5Bil a yr. Maybe less than 10% that to satisfy Ireland. Can u wait 3 years?


So, technically what do you need to use their offering down the road? A satellite dish on the roof? Just a small antenna as built in in most phones today?

And where do I sign up? :-)


It’s a “user terminal” about the size of a pizza that can be mounted in a roof. Since it’s a phased array the direction doesn’t matter (unlike a traditional dish for a geostationary satellite). Once they are producing them in volume they should be pretty cheap.


> Since it’s a phased array the direction doesn’t matter

Of course it does matter, a phased array isn't magic. A steered antenna will have much better gain when the target is at low elevation, when a fixed phased array will be operating at very oblique angles.


What I mean is you don’t have to point a dish within arcseconds to get a connection. You just have to make sure it’s facing upwards. Per the Starlink Wikipedia page:

> Instead, it will be linked to flat user terminals the size of a pizza box, which will have phased array antennas and track the satellites. The terminals can be mounted anywhere, as long as they can see the sky


Cool! I noticed that it uses a star map to locate itself, like the SR-71 blackbird did before the time of GPS. Is this novel or common for satellites?


Strictly speaking, the star trackers don't provide any position information; what they provide is a very accurate measurement of the satellite's orientation. And yes, they're commonly used on satellites.


Exactly, I remember that Hubble used the GSC catalog and it was the first cd rom that I had at the time: https://en.m.wikipedia.org/wiki/Guide_Star_Catalog


There is no GPS in orbit. You need a point of reference and stars are pretty (99.99x10^zillion)% reliable, and yes, the SRS used the same idea.


who gave you that impression - ofc there is GPS in orbit. it is just that the usual commercial GPS software refuses to deliver coordinates above a certain height or a certain speed. this is ostensible to prevent GPS from being used in non American ICBM and cruise missile weapons.

the GPS birds are in a 20k km orbit and you should be able to use them just fine at lower orbits (starlink sats are planned to be in several fairly low orbit shells 320/550km).

I suspect you can even use GPS all the way out to say the moon if you are a bit creative. (catching GPS sats coming aroud the earth for positioning, or using a special dish antenna to get reflected GPS signals)


> 20k km

A double metric "kilo" prefix like that feels awkward. I know it's common, but I've always felt that we should just say "megameter" instead. (Also, a metric ton should just be called a "megagram".)


Agreed with megameter, but disagree with tonne: the kilogram is the base unit that physical constants etc. are measured against (e.g. Newton's constant is measured in Newton meter squareds per kilogram squared). It can be less confusing to talk about microkilograms than grams when doing physics.


I wonder if I will live long enough to see GPS coverage extended to the full surface of Luna, perhaps with some transmitters in lunar orbit.


How directional are the GPS antennas? They look like the point earthwards, but I suppose you could pick up some signal above them as well.

Wouldn't reflected signals be inaccurate due to additional time of flight?


Well yeah, you can't just duct tape a TomTom to it. But you can buy GPS receivers designed for space.

https://gdmissionsystems.com/en/communications/spaceborne-co...


There is still GPS in leo. It can use more power than some smaller satellites can budget for is a reason it's not always used.


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: