The circles in this map are really deceptive. They're two fixed sizes depending on altitude with no apparent relation to actual coverage.
Here is my attempt at estimating actual coverage, based on a 25 degree angle from the horizon https://i.imgur.com/poj8cae.png
I've included a screenshot of the code I modified so people can play with it themselves. I made a slight simplification in my math (explained in the comments in the screenshot), but I think it should be close enough. Edit: Pictorial explanation of simplification https://i.imgur.com/XtADkmh.png
The circles in the map match what a number of offline (TLE based) standalone Linux and Windows satellite pass prediction tools will show you. They assume that each ground location is on a flat plane with no hills or mountains, and the size of the cone extending from the satellite to the ground is the full coverage of everything the satellite can see (such as if you had a magical laser on the satellite capable of illuminating any ground location visible from the satellite).
Such as if you had a theoretical, nonexistant ground terminal that was capable of communicating with a satellite at only 1 degree look angle elevation above the horizon.
Of course in reality this is not the case and most systems that communicate with LEO satellites will have minimum elevation look angles of 15-20 degrees above the horizon.
The circles do, however, provide a useful approximate scale of how much ground area is visible at any given time from the point of view of a satellite. It's not intended to be perfect.
> Such as if you had a theoretical, nonexistant ground terminal that was capable of communicating with a satellite at only 1 degree look angle elevation above the horizon.
No, no, that is not even close to true. Measure the big circles. They have a radius roughly the same as satellite altitude. From an altitude of 550km, the area the satellite can see all the way down to the horizon is wider than the contiguous united states, east to west.
They're confusing, but I wouldn't say "really deceptive".
The big circles seem to be satellites that are at final altitude, showing range equal to altitude, so slightly lower than 45 degrees. The small circles are satellites that aren't there yet, and the size is meaningless.
I would say really deceptive, or perhaps just plain wrong. The circles are, in fact, intended to convey coverage and left me pretty confused about the huge gaps everywhere.
From the about section:
> Dashed lines show the approximate coverage with very conservative viewing slant (approx 60 degrees above horizon, where 90 means vertically overhead).
> Opinion
> No evidence a working starlink network can offer lower latency, or lower prices, than terrestial
>
> Impact on earth based telescopes has not been discussed
>
> Starlink is a play for pentagon funding to keep SpaceX launch schedule busy?
Author doesn't seem to have any love for Starlink. I wonder why the built the site.
Starlink isn't designed to offer lower latency or lower prices than existing terrestrial wired or wireless ISPs (at least, decent ones). Instead it's designed to offer a low latency wireless service in places where that has never been available at all. Which is most of the land area of the Earth, and all of the ocean. It may also be cheaper and/or faster than some very bad terrestrial ISPs, because there are a lot of those out there, but that's not what it's designed for.
Impact on earth-based telescopes has been discussed endlessly, to the point where it seems like there's a law that you can't mention Starlink on the Internet without talking about it. And SpaceX responded by talking to the astronomers most affected by Starlink and adding a sunshade that should prevent the satellites from ruining their images. Which is more than any other satellite operator has ever done, AFAIK, and there are many satellites much brighter than Starlink in the sky today.
The Pentagon is certainly interested in Starlink, but commercial service has much higher revenue potential. There are several competitors planning to do roughly the same thing as Starlink, so the business plan is at least plausible.
> Starlink isn't designed to offer lower latency or lower prices than existing terrestrial wired or wireless ISPs (at least, decent ones). Instead it's designed to offer a low latency wireless service in places where that has never been available at all. Which is most of the land area of the Earth, and all of the ocean.
It is widely believed that a huge chunk of Starlink’s revenue, if not ~all of it, will come from leasing faster-than-currently-available low latency links to HFT companies for nyc-lon and such. This is why their first orbital planes being launched are for the north atlantic and CONUS.
Providing internet to the underserved areas is secondary; the cash will come from beating terrestrial fiber links (even idealized great circle ones) between markets that are already extremely well connected.
One such private low latency link between Seacaucus and LSEG will probably pay 2-5x what everyone else on the planet combined pays for internet access in underserved areas.
> It is widely believed that a huge chunk of Starlink’s revenue, if not ~all of it, will come from leasing faster-than-currently-available low latency links to HFT companies for nyc-lon and such.
This is speculation. Without inter-satellite links the latency between New York and London on Starlink would be quite disappointing.
> This is why their first orbital planes being launched are for the north atlantic and CONUS.
A LEO constellation can only be targeted at specific latitudes (not longitudes). They're just targeting mid-latitudes because that's where potential customers live.
The data on this map makes it look like they're going for pretty uniform coverage already. Is there a source for the early orbital planes having a specific purpose? The earth rotates underneath each plane, and the times of day it will be available to any given ground path change over the course of a year.
It does seem pretty convenient that the inclination they've chosen lines the satellites up nearly in a straight path from New York to London, but I imagine this would be relevant to many transatlantic communications.
There is a purpose, but it's not HFT... It's orbital mechanics.
You can't have efficient coverage of the whole planet without launching all of your satellites at fairly high inclination (>45°). There will be more satellites per area near the poles and so that's why Starlink will be available at very high (and very low) latitude before it's available at the equator.
> It is widely believed that a huge chunk of Starlink’s revenue, if not ~all of it, will come from leasing faster-than-currently-available low latency links to HFT companies for nyc-lon and such. This is why their first orbital planes being launched are for the north atlantic and CONUS.
True? I'd heard that the initial target market was passenger planes over the ocean, cruise ships, and similar maritime and aerial activity that has to use geosynchronous satellites at present.
Covid-19 has thrown a spanner into the works of that plan, for a while. I'm sure people with money won't be turned away.
Longer term, remembering that Musk hails from South Africa, I thought the plan was to bring internet to failed states or barely functional areas in Africa (outback Namibia/Zimbabwe/Tanzania etc) and possibly also the Americas and the Middle East).
The long term plan is to sell service to anyone who will buy it, to fund Starship for Mars. The ship and plane market can't be served by the current constellation because it doesn't have links between satellites. This also excludes the high frequency trading market (proving that SpaceX is not relying on traders to make their revenue goals). The initial market is rural areas of the US, and then other countries as they expand their ground station footprint.
Oh, sure, the latency will be lower - it's just that without these satellites it was simply not possible to get continuous coverage on long over ocean flights, so something is better than nothing.
While for airplanes I think the latency might not be that problematic (I think VoIP in plane package full of people would not be very welcome) you indeed want low latency connection on a cruise ship.
NYC-LON are already connected by shortwave systems built using huge-ass yagi-uda antennas and special bespoke modem + amplifier systems. Shortwave based systems have even obsoleted well known HFT microwave paths such as Chicago-NYC.
source: I know some of the people who built several of the Chicago-NYC links.
Be careful, everytime I make this point I get berated by the Spacex fan boys. The reality is, making high speed trading faster adds no true value to society. And therefore I find this entire project sad.
SpaceX is launching the initial constellation without the laser links required for the hypothetical low latency service for traders. This would never happen if traders were the only or primary market. SpaceX has stated that their target is the consumer market and their actions are in line with that.
You don't need inter-satellite links for low latency, as Mark Handley has elegantly pointed out in his incredible video. I don't think internet access will amount to even 50% of the revenue of Starlink.
You don't need satellites at all to beat fiber. Terrestrial microwave does it too. Starlink can't beat microwave except for places where it's impossible to install microwave, like over the ocean. But without cross links Starlink won't work over the ocean either.
Source? There is no indication whatsoever that the service will be cheap enough to be for the consumer market. In fact, all serious cost analyses show it's not.
The only problem there is cost, upfront and ongoing. JSW is super expensive. We were very lucky Hubble wasn't a complete write off and even then the fix was very expensive.
All that vs. laying down some concrete and a telescope in a remote location.
The cost of JWST is way out of control. You could build 10 space telescopes for less if you actually cared about efficiency and didn't need everything to be perfect for the first and only launch. Especially since the crazy folding mechanism would be completely unnecessary if you just launched them on Starship, which is big enough to hold JWST's mirror unfolded and possibly even the sunshade.
If you built 10 space telescopes for the price of JWST, I believe that EACH one would still be more expensive than the most expensive terrestrial telescope ever
I was only thinking of telescopes that currently exist, since it's hard to determine the cost of something before it's built. Inflation and such tends to make a direct comparison a bit trickier— it's not fair to compare JWST dollars spent in 2000 with dollars that won't be spent until 2025. Plus, even though these sorts of things typically end up being more expensive than the original budget, I'm being optimistic :)
I'm not sure switching launch vehicles is a realistic option for a project being in development for 24 years by now. Especially not to a not-yet-existing rocket.
If only there were launch platforms that could put up smaller telescopes (optical and radio).
I mean you can do a lot with telescopes that are much smaller than Hubble Class or JSW. A lot of people are doing work with telescopes that are much smaller. Also, as long as frequencies and locations are published, much of this noise (not all!) can be removed. Sure, the VLA is going to have a major headache, but LIGO wouldn't exist if we didn't know how to remove noise (LIGO had to deal with noise from cars many miles away and all kinds of vibrations because it is ridiculously sensitive).
LIGO can isolate itself from seismic backgrounds, but it has no artificial gravitational-wave foregrounds. (Yes, Newtonian Noise is a concern for future generations, but it is expected to be modeled and subtracted well-enough.)
Arrays like Starlink are a persistent foreground for the big radio telescopes. SKA and the like are in the Outback for a reason, but for this there is no escape.
Typically for high resolution images you actually take multiple images. So it would be more work, but you can definitely remove foreground objects. This is like how you can take a bunch of photos of a landscape with people moving around in it and remove all the people by just looking at what isn't changing in the images.
I'm not saying it is easy, but it is definitely possible.
Having worked at NASA, let me tell you that number is 100% BS on all levels. You have to go out of your way to make a satellite cost that much, which they did.
Look at how, say, Planetary Resources was putting together their much smaller telescopes and you'd get an idea of how it could be done on the cheap. Then to scale up you'd focus on cost effectiveness instead of maximizing capability at all cost. Do multiple cheap launches ($100m each) instead of spending billions on a complicated automated unfurling mechanism, etc.
I bet SpaceX could assemble a huge telescope in space from smaller telescopes, and they could do it for much less than NASA ever could. NASA is a jobs program.
NASA is first and foremost our nation's science program. Everything from the network card driver your AWS instance is using to the planes flying into hurricanes to determine how strong they will hit was built with major contributions from NASA engineers.
LASIK eye surgery, the nutritional supplements in baby formula, solar panels affordable for consumers, OpenStack, even the BowFlex were all developed by NASA or under NASA funding programs.
You could at best say NASA is a jobs program for HN readers. :)
It is correct to call NASA’s human space flight a jobs program. It is specifically engineered to maximize the number of jobs spread across key congressional districts and states. And unfortunately this makes it extremely wasteful.
It only costs that much because launches are expensive. If you live in a world where your satellite costs $300m+ to launch your going to spend a lot on the satellite to make sure that it is reliable and has a long life.
If suddenly launches only cost $50m then you might build a cheaper and slightly less reliable satellite as you know you can always launch another for only $50m
That's just one though, meanwhile SpaceX is complicating all ground astronomy, you can't just replace all observatories on Earth with one big dish and expect things to be better than before.
According to a broad study of astronomical surveys[1], the impact of Starlink on the vast majority of observatories is quite minor. There were exceptions, like the upcoming Vera Rubin Observatory in Chile which uses a wide angle at twilight so is particularly impacted.
Starlink used that study to come up with the night shade idea to mitigate those impacts.
In that survey, radio telescopes were in the "minorly affected" list.
You must have pasted the wrong link, because the linked article mentioned "survey" 4 times and 3 of those was in a short sentence saying searches for near-Earth objects will loose all twilight...
The severity of the noise for radio was not graded in the slightest, and the word "minorly" is never used.
Only if they are operating when they are in range of the telescope. The satellites know where they are in space, they could just shut off when passing over. Usually not many customers where these things are installed anyway.
Starlink will has promised better latency, bandwidth and prices compared to existing satellite internet. Perhaps that's where the confusion on the first point came from.
It's really baffling how someone motivated and talented enough to make such a tracking site would have such an odd opinion section.
Obviously, Starlink is not the best choice in all situations, but it is so compelling on some dimensions that such a flatly negative opinion is bizarre for anyone with even a cursory understanding of what's going on.
What dimension? It's not competitive for anyone who can get high-speed internet right now, and it's not clear if it will complete with traditional satellite in cost.
This is demonstrably false. Look at the YouTube video posted this week that simulated delayed. The best delay, with only hops between satellites, was 40ms or so. That didn't account for any of the higher later delays or transport network.
You're comparing apples to oranges. The 40ms in that video is for Seattle to NY via multiple satellite hops, which is a hypothetical configuration. Elon's tweet is talking about one hop only, from satellite to ground station, which is the intended configuration for v1. There will be dozens or hundreds of ground stations in the US to support this.
For existing GEO satellites, the equivalent measurement is 600ms. That's the relevant comparison. With just 20ms added on top of regular internet latency it would still be usable for competitive gaming at all but the highest levels, as the tweet says.
I'm referring to the satellite to gateway hops. There absolutely are many of those just to cross the continental US. The satellite altitude is such that it only covers a small part of CONUS before it has to have off to another satellite.
Why would they use satellite hops for those? They'll just put your packet onto an internet backbone fiber. Starlink is last mile service, not backbone.
I don't see why it couldn't. Maybe people here are used to the top of top internet, but I've lived in many places where the best I could get was ~5mbps, with 60ms latency and 60$+.
Those are very easy numbers to beat. Yes it might not beat prices and latencies for a super competitive city with true competition, but those are honestly rare and few between. The majority of north american cities have shitty internet at high prices and high latency.
Also, slap this router on a Tesla car, allowing it to have connection anywhere it drives, and you're golden.
What's funny is if you go sailing on the ocean for extended periods (months) you realize that sat data / phone costs are RIDICULOUS in this age. $3 - $5/MB. So your 4 GB movie is $20,000. You'd think, why would anyone pay this, but there can be a lot of pressure these days to be able to send videos etc so you sometimes get jammed up with these types of costs (ie, if you do a telemedicine call the DR may want a video of joint movement etc etc, recreational trips people want to post videos and don't realize their phone records at pretty large file sizes etc etc).
So if they can do better just on COST then this it would be wonderful.
At least V1 of Starlink, there needs to be a ground station visible to the satellite, so it's not going to work in the middle of the ocean.
It might work a few hundred miles offshore, but not further than that. So it'll be a while before you can have video calls from the middle of an Atlantic crossing.
> At least V1 of Starlink, there needs to be a ground station visible to the satellite
At an altitude of 340 miles a starlink satellite has line of site up to 1800 miles so conceivably a ship at sea could be 3600 miles from the nearest land station and still be able to bounce off Starlink. Of course that's ideal conditions, but it wouldn't take too many ground stations to ensure 99% of oceangoing traffic is covered.
Of course even with line of site, the antenna may not be oriented in such a way as to cover more than a couple hundred miles of surface even if there is line of site.
Even though the theoretical line of sight is 1800 miles, I expect that you'll get much less range. The satellites move very quickly— the satellite that's halfway between you and the ground station is very quickly going to be more than 1800 miles away from one of you. Transmitting signals at such oblique angles is also a lot harder. Part of it is just typical signal strength los due to r^2 law. But if you're transmitting thru that much atmosphere, you're going to experience a lot more interference from things like rain fade.
It just has to be above the horizon, because this is the middle of the ocean.
When I put 550km into a horizon calculator I get a distance of 2700km.
Edit: Okay, I want an explanation for the downvotes. If you're upset that I didn't say "the user terminal is horizontal at max range", I thought it was clear from context.
I didn't downvote you, but the size of the beam is heavily dictated by the elevation they're willing to work with, and the wider the elevation, the more expensive the antenna. So give that they need to decrease the size of the beam to keep antenna cost down, it's highly likely that the beam size is much, much smaller than what you cited there. Just take a look to see how many handoffs there are over CONUS alone: https://www.youtube.com/watch?v=m05abdGSOxY
We're talking about the user terminal, right? A ship can easily have a more expensive antenna array. Maybe even with a motor, or with three tilted in different directions.
> Just take a look to see how many handoffs there are over CONUS alone
That's just a simulation picking angles he thought looked good, in an area where there are plenty of options. It doesn't tell you anything about the limits of beam size, especially when there is no other sat in range.
Sure, ships can have more expensive antennas. But you need a lot of those ships out there and the antennas are extremely expensive. I'm not sure what you are arguing about the simulation. Are you saying it's inaccurate and there's a better one?
> Sure, ships can have more expensive antennas. But you need a lot of those ships out there and the antennas are extremely expensive.
We're talking about receiving internet on ships, right? You need one per ship that wants internet. And it doesn't have to be 'extremely' expensive, it could be a moderate multiple of the normal user terminal.
In case you forgot the context, this is not a conversation about using relays. This is a conversation about how far offshore you can go in a single bounce.
> I'm not sure what you are arguing about the simulation. Are you saying it's inaccurate and there's a better one?
It's perfectly fine, but the goal of that simulation is to show a way of doing long distance high speed relays. The goal is not to show you maximum range if you don't relay.
I'm talking about ships used for relay. Ships used for just internet over the water won't be possible if there are no gateways (other ships with relays) over the water. Those terminals do have to be expensive, just like the gateway ones, because the antenna performance is much higher than the user terminal's, and the radome is very large (see other pictures from reddit on here).
For the non-relay case, we've already established that due to beam size you can't get very far off shore before it won't work. The goal of the simulation was also to show beam size. If the beam sizes are arbitrarily large then obviously the simulation would have shown a single hop.
> I'm talking about ships used for relay. Ships used for just internet over the water won't be possible if there are no gateways (other ships with relays) over the water.
I figured that out, but that's not really what the conversation was about. If you add those gateways then you don't have the range problems, and this was a conversation about range problems.
> Those terminals do have to be expensive, just like the gateway ones, because the antenna performance is much higher than the user terminal's, and the radome is very large (see other pictures from reddit on here).
The video you linked talks about using normal user terminals as relays. A couple of those won't be full bandwidth, but your mid-ocean service wouldn't need to be full bandwidth.
If you're actually making a dedicated relay ship you'd probably want a full radome link, but only if it's cheaper than the ship you're putting it on. The limiting factor is the cost of the ships.
> The goal of the simulation was also to show beam size.
I strongly disagree. Beam size is modeled in a very simple way, because it's not the goal of the simulation. The goal is to show how links work in different ways, using normal-scope full-bandwidth service beams.
I strongly doubt that an almost-idle sat is unable to aim even a few degrees wider if they wanted to.
But the person making that sim would have no way to model that, and it's not really worth it to get an extra fraction of a millisecond. Better to make the simulation based on confirmed capability.
But for ships at sea it's not a fraction of a millisecond, it's the difference between connectivity and no connectivity. If ships at sea were starlink's primary customers then I bet they could and would get better range even with zero hardware changes.
> The video you linked talks about using normal user terminals as relays. A couple of those won't be full bandwidth, but your mid-ocean service wouldn't need to be full bandwidth.
If you see my comment and others, using user terminals as relays is going to be very difficult, if not impossible. The EIRP/GT of those terminals will likely be very low, and that single relay will cut the bandwidth significantly. They also have much lower availability than a gateway, which makes the routing decisions harder.
> I strongly disagree.
What beam size do you think they're using? It's not really a question of IF they want to. Doing so on a cheap terminal can easily break FCC interference guidelines due to sidelobes, thus making it not their decision. That's the reason why these systems design for a very specific elevation limit from the user side, because anything else would either have too poor performance at the detriment of the entire satellite, or it's illegal if you're transmitting where you shouldn't. The performance degradation is very serious as well. It can be as bad as a single user consuming 10x more satellite resources than a nominal user, just because they're outside of the coverage of the beam.
So I realize where you're coming from in that they can do it, but I guarantee given the link budgets and cost-benefit tradeoff, it's simply not worth it to cover that far off boresight.
This is clearly not representative of starlink, but rather a GEO constellation. But the same concept applies: they will have tiny 0dB contours, and it rapidly falls after that. Serving users outside of the main contours, while it might be feasible given their antennas, is a massive hit to capacity on the entire constellation.
By the way, I think this dialogue is good and neither of us are going to convince each other. I think we will have to wait and see a year from now and revisit these comments.
Are the FCC interference guidelines exactly the same in the middle of the ocean?
And for what it's worth I looked up one of the licensing sheets earlier and it talked about the allowed signal strength below 25 degrees tapering off by 15dB. That's not enough to stop you from having a signal.
> too poor performance at the detriment of the entire satellite
> Serving users outside of the main contours, while it might be feasible given their antennas, is a massive hit to capacity on the entire constellation.
I think you're agreeing with me here.
It might cost a lot of the satellite's performance, which is why you wouldn't do it over land, and why it would be even less reasonable to include it in the simple simulation you linked.
But satellites over the ocean have nothing better to do with most of their capacity.
> By the way, I think this dialogue is good and neither of us are going to convince each other. I think we will have to wait and see a year from now and revisit these comments.
I'm fine stopping here, but revisiting in a year probably wouldn't help. There is a huge difference between what they can do, and what they care enough to do. Servicing ships that are more than 300 miles offshore, but not too much more, is definitely not a priority.
But you can bounce it off "ground" stations sitting in the ocean. For instance on barges and islands, or just on other customers ships. It doesn't take that many ground stations to be able to be routed to land. Or from land on one side of the ocean to the other.
Is there a rough v1/v2 timeline? The thing they wanted to release by the end of the year is v1 I assume?
Is v2 just a software update? Does it require sending a whole new set of satellite or it something they are planning to patch into existing satellites someday and have inter-sat support?
The original plan required a laser link, but those were removed from the v1 satellite because the lasers would survive reentry and SpaceX didn't want their satellites falling on anybody's head. So this is definitely a hardware upgrade requiring new satellites.
But that's not a big stumbling block -- SpaceX plans on replacing 1/5th of their constellation every year.
That’s kind of a moot point as SpaceX is not going to put satellites over the Atlantic without that capability or even build out their network. Everything is still very much in testing phases right now.
None of these satellites are anywhere near geosynchronous orbit so every satellite circles the globe. They don't really get much of a choice as to where to put them other than altitude and inclination. The satellites are in different orbital planes but that plane rotates with respect to the earth so there's no getting around the majority of satellites sitting out over the oceans.
To be clear, I am saying their plan is so have the links up before the consolation is finished. These early launches are going to be in a higher orbit which will allow them to offer the service on land before the full constellation is finished as well as provide redundancy.
We used a sat link for getting the latest weather information, though you can also get that frequently on V(HF), so .. not terribly useful. But, terribly slow and unreliable (Iridium).
That's what I probably misunderstand about the technology - I don't understand how you can have lower latency than current WISP-based infrastructure (are they making these claims against ground-based infrastructure?) when there's satellite internet connections that fade out when a cloud goes by... Also, I would assume they'd have massive capacity issues if they're planning on over-subscribing (they'd have to if they want to sustain a viable profit model, unless they're surviving on government funding?)
They're claiming lower latency than existing satellites due to lower altitude orbits and are using more satellites to compensate for the decreased coverage from each satellite.
Presumably this setup would also help with the cloud coverage issue due to the shorter travel distance and higher likelihood of redundant satellites to connect to.
To be more precise, some of the lowest flying Starlink satellites will be at 340km, which is very low — lower than the ISS, even. By contrast, traditional internet satellites orbit at ~35,700km.
The speed of light in a vacuum is about 50% faster than the speed of light in glass.
But that requires the inter-satellite links be viable, which they aren't yet: the required lasers would survive reentry and they want the satellites to burn up completely on de-orbit so current versions of the satellites don't have them.
And yes, there are significant capacity limitations. Only the truly rural will get good data rates.
Their competition are geostationary satellites that have high latency due to speed of light delays.
In theory a LEO satellite network could also beat transoceanic fiber connections since the high refractive index causes the speed of light in fiber to be only ~2/3rds the vacuum speed of light. But that would require satellite-to-satellite optical links which the first generation of starlink doesn't have.
> Their competition are geostationary satellites that have high latency due to speed of light delays.
Actually it is due to their distances.
Speed of light delays is something that exists everywhere, like fiber internet. In fact, speed of light is faster in space than it is in optical fiber. The problem is distance.
'Actually it is due to speed of light delays. Distance is something that exists everywhere.'
Your weird pedantic distinction doesn't make a difference. They're synonyms in this context. Obviously the difference is how much you have, everyone knows that.
I used to work on rockets and I'm very confused by the author's opinions.
> No evidence a working starlink network can offer lower latency
This is weird. Even if we pretend that these satellites are directly competing with terrestrial based systems, their orbits put them in range for a latency that is at least competitive with terrestrial systems. It's unproven that they get those latencies, but there's clear evidence.
Note: we're talking about like 50ms (IIRC 20ms - 100ms depending on which constellation). Geostationary satellites (which is what they are competing with) are like 250ms (these numbers are off the top of my head so correct me if I'm off). Starlink is competing with geostationary internet satellites AND places that have no internet at all.
> lower prices
Seems to me they can service a lot more people. Isn't this "scale?" Aren't they targeting something like a billion+ users? I know plenty of people with boats and RVs that are excited about this because the terrestrial solutions just suck (and generally aren't terrestrial based solutions).
> Impact on earth based telescopes has not been discussed
What? You're literally discussing it. It is pretty easy to find articles from astronomers getting upset [0][1]. I can tell you that everyone in my community (which includes astronomers and astrophysicists) is aware of this issue. The fact is that most people just don't care. They see greater utility so it is worth the cost. Of course, there are also plenty of people that disagree.
> Starlink is a play for pentagon funding to keep SpaceX launch schedule busy?
Little bit of this, little bit of that. Having worked at a rocket company I can definitely tell you that, like every other manufactured product in the existence of products, flights are cheaper the more you do (especially true for reusable rockets!). Sometimes this is called the "launch chasm." I'm not sure why you're mad at SpaceX here. If you have an issue with this, it is with Pentagon funding.
Look, I'm not the biggest fan of Musk or SpaceX either, but these aren't great arguments. The things to dislike are a little more nuanced whereas this just comes off as hating SpaceX to hate on SpaceX.
> > Impact on earth based telescopes has not been discussed
> What? You're literally discussing it. It is pretty easy to find articles from astronomers getting upset. I can tell you that everyone in my community (which includes astronomers and astrophysicists) is aware of this issue. The fact is that most people just don't care.
Their latencies are not at least as competitive as anything if they're not proven. With low orbits, they need multiple hops to get to most destinations, along with queuing and network delays. Propagation delay may not even be the dominant factor.
Your link does calculations on $50 a month. That's crazy cheap, not expensive.
Your link's math says that SpaceX is marginally profitable selling 220GB of internet at $50 per month using 2016 bandwidth levels, which is pretty impressive.
Those numbers are from the original 2016 FCC filing, they've got a lot more bandwidth out of the satellites now because 4 years is a long time in technology. Musk says 3Tbps per set of 60 satellites. That's 50gigabits per satellite, not 20. Shotwell has indicated a price of $80 a month, not $50. And you know that's going to be the low price, not the high one. So let's assume that includes a 100 gigabyte cap, which is still about 10X as much as you get with the base plans from the geostationary satellite providers.
Those new numbers increase the subscribers per satellite from 7000 to 38000. Revenue per satellite per month jumps from $35K to $300K. At 60 satellites per launch over a lifetime of 5 years, that's gross revenue of $1.1B per launch.
Re: your other point about latency. They've already got 28 base stations in the US. That's enough to cover the continental US without any hops.
I think you're going to need to explain how a satellite at 550km in the sky with a low elevation covers the US without any hops. That's contradictory to what every technical piece of information about Starlink says. You have to hop between base stations. That's a hop, regardless of whether it's through networking equipment or a gateway.
They'll use the internet backbones for communications between their base stations rather than satellites.
To be more explicit, before they sell you service at a location, they will ensure that they have enough base stations close to you that both your terminal and a base station are both always visible to a single satellite. The base station is connected to an internet backbone. So the signal goes from you to the satellite to the base station to the internet backbone. The reply goes in reverse.
So yes, this means that they need to have a similar number of base stations as satellites in a minimal constellation. They've said that a minimal constellation for full US coverage is 1000 - 2000. They have 28 base stations in the continental US, 2% of the globe's area. Yup, 28 is about 2% of 1000 - 2000.
And therein lies the problem. Fiber latency from coast to coast in CONUS is 20ms in the very best case. But that will never happen, and you instead of many hops, where the latency is typically upwards of 40ms just for the fiber. Elon is saying the latency will be around 20ms. How? Because he's quoting a single path up and down in the same beam in the best case. Nominally, it will be much, much higher.
When people quote you their ping times, they quote it to the nearest AWS or Google CDN, not to servers on the other side of the country. They type "ping google.com". If the Starlink base station is co-located with a Google CDN, best case is an 8ms ping time. (There might not be a google CDN but there definitely will be a Netflix cache in each base station) 550km one way at the speed of light is 2ms. Up & down for the ping and then up & down for the pong gives 8. Normal case is probably about triple that because the satellite won't be straight up and the CDN won't be co-located. But lots of base stations will be in the same metro as a CDN.
I've never heard of people quoting their ping time based on the closed DC to them. Let's say you're playing a game (fortnite, for example). You have a ping time based on where that server is. It may be across the country from you, and there's nothing you can do about that to make it better. Their ping time will not be anywhere near 20ms as that tweet alluded to. I'm pinging google.com on a fiber connection right now, which would go to one of their load-balanced DNS entries. I get 30ms. When people say you have 30ms of latency, they expect that latency for the normal things they do, not in some artificial test of path delay. Otherwise traditional satellite is 480ms and not 600ms, and fiber is 15ms and not 30-40ms.
The ranges are big. It doesn't take many base stations with fiber backhauls to cover the entire country. No need for multiple hops if that would be slow for some reason.
That would explain why there are dense clusters of satellites over middle-of-nowhere oceans, but they're sparse over continents, particularly Asia and Africa.
The satellites orbit the earth every 1.6 hours. They're not in geo-stationary orbit, they're in very low earth orbit to reduce latency. And any orbit will spend a significant portion of it's time over water.
Also, only the green circled satellites are in their final orbit, the yellow circled ones were recently launched and are not yet at their final altitude and position. It takes a couple of months for orbit raising.
Yeah, as I understand it the current generation of Starlink satellites are likely to be pretty much unusable when they're over the ocean since they need to have line of sight to a ground station.
A ground station doesn't have to actually be on the ground. They could park a boat in the ocean between London and NYC with equipment to act as a repeater. This YouTube video (along with others on this channel) does a good job of explaining how this could work. [https://youtu.be/m05abdGSOxY]
> They could park a boat in the ocean between London and NYC with equipment to act as a repeater.
They could, but I bet they won't. Speculation is fun but permanent floating platforms are expensive, it's easier to just wait until inter-satellite links.
without interlinks an airplane, ship, oil platform, or island could talk to the starlink sat, but the sat couldn't talk to its ground base station to forward their messages to the rest of the internet. Coverage for deep ocean will require a second generation that can talk sat -> sat.
This map seems to have some inaccuracies. I'm not sure what the circles are supposed to be, but they're wrong for the satellite's coverage limits - looks like they're shown at about half of the actual 940km coverage radius. It also looks like their altitude is incorrect - the ones in operational orbits are at 550km. Eyeballing this map, it looks like they're shown at something like 1/3 to 1/2 of that - certainly much lower than they should be.
In the /about it's explained that the circle represents the locations where the satellite is at least 60 degrees above the horizon.
550 km is the altitude for the operational orbits, the satellites are deployed lower and need to raise their orbit. (But I see most are labeled around 550 km, so I'm not sure what you're referring to.)
Those launches are shown way too low in this map. From measuring pixels, I make them to be roughly 200km.
Edit: maybe I'm being pedantic here, but the combination of showing artifically small coverage zones, plus showing satellites much lower than they are so giving a misleading impression of how much area each satellite can see, taken together give the misleading impression that Starlink coverage will be worse than it will actually be.
I think you, and others, might just be misinterpreting the point of the map. My take is that it is to help people look for the satellites. A 60 degree above the horizon cutoff isn't unreasonable, and making the circles much larger would clutter the map.
I find it helpful, anyway. (And to me, at least, it is more helpful than a hypothetical coverage map, since I can't get Starlink internet now.)
No it's not being misinterpreted, the parent comment is right - the website author has a negative biased agenda against Starlink and Elon Musk, as evidenced from the website's own About page:
* No evidence a working starlink network can offer lower latency, or lower prices, than terrestial
* Impact on earth based telescopes has not been discussed
* Starlink is a play for pentagon funding to keep SpaceX launch schedule busy?
"No evidence a working starlink network can offer lower latency, or lower prices, than terrestial"?
The FCC themselves don't believe the 100ms, and the price nobody in the industry thinks will be affordable since they haven't released any info about their user antenna.
60deg is more likely to be due to the limitation of the smaller "pizza box" user antenna, as opposed to the (likely much larger) spaceX operated gateways, which would feasibly reach down to 25deg.
If I recall the FCC filing correctly, the constellation was intended to always have a sat >50deg above horizon.
The original filing was for reachability down to 40 degrees above the horizon. When they lowered the altitude from 1100 km to 550km, they reduced the angle to 25 degrees above the horizon for the first phase.
Code is included in screenshots, and you can do this yourself in chrome, ctrl-shift-c to open dev tools, navigate to the area in the code I have open, and modify it. Please check my work.
Doesn't include coverage yet but I do plan to add that with some kind of heatmap. I think these static circles in the OP are a bit deceptive in indicating 'coverage' as these satellites are constantly moving.
The (or some of the) Starlink satellites stand out as a very distinctive line on http://stuffin.space/ currently over the Pacific and West Coast North America.
Is there any decent resource explaining how the whole mesh of Starlink sattelites communicate to each others and to the ground ? Do they use RF radio or modulated laser ? What equipment should be used by end users and what frequency range ?
Those are the trunk link earth stations. Not the CPEs.
To be pedantic the radomes are only a couple of thousand dollars each, it's what's inside them that's expensive. Spherical radomes up to a couple of meters in size are kind of a commodity item. The Cobham-manufactured, agile, ku-band tracking dish antennas with Rx+Tx RF chains are expensive.
The small white things not in radomes shown in the picture are possibly beta test CPEs.
Right, sorry. I was referring to the large white ball. Ka/Ku radomes are not cheap. The only information on antenna size is from Elon saying it's the size of a pizza box. Those look too small to be the CPE (imo).
I was referring to what appears to be the gateways. If the smaller one is the CPE I'd be curious to see where the motor is, and how it will be prevented from weathering/failing.
TL;DR User terminals are RF phased array antennas. Ka-band and Ku-band, plus E-band for ground stations. v1 has no cross satellite links. Plans for v2 include laser cross links.
From my point reducing electronics while adding more mechanical parts will most likely lead to both cost and size increase which will make product less attractive. Fast FPGAs are getting cheaper daily.
Will they cover the North and South poles as well? I know there isn't much of a customer base, but it would be interesting to know that every part of the planet is covered and connected.
No. If you look at the map you'll see that there's an equally sized coverage gap at each pole. I don't know if any reliable source has stated whether they will put any satellites in orbits that can cover the poles, but I imagine it's a pretty low priority commercially.
There is already some polar coverage from the Iridium network, which uses orbits with higher inclination.
In internet coverage, like with most infrastructure, there are power laws. I'm making up numbers but it's like connecting the first 50% costs as much as connecting the last 5%. Starlink is for those 5% who live in extremely remote areas.
Fewer potential users in the near-polar regions,
so the orbits are angled & aligned like wrapping
a ball of twine, while polar coverage would require
multiple tracks in north-south orbits.
The orbital planes top parts are all up there at high latitudes. The circle you are seeing is actually made up of several different orbital planes that swoop down toward/past the equator and then up/down to 60 degrees N/S.
Seeing them move will make it clearer.
Here’s an awesome visual explanation with a ton of other cool info:
There will be in v2. In v1 there's no point because the satellites don't have cross links, so they can only provide service while in range of a ground station, and there won't be any ground stations at the poles.
A satellite orbit has an inclination. If you launch it lined up with the equator then it will always stay over the equator. So you choose how far the satellite will wander north and south by adjusting the inclination at launch. The less the inclination the fewer satellites you need for coverage. So it is a tradeoff. Obviously the decision was made to only cover up to the place that population density falls off in the northern direction.
Added: Because the satellites spend more time at the northern and southern extremes they get denser coverage there. That is likely why they are initially offering coverage near the US/Canada border.
All the satellites in the first phase of Starlink are in 53 degree inclination orbits. This will give them coverage from about 60 degrees North down to 60 degrees South. In later phases they plan to have satellites in higher inclination orbits, but in the early stages, there just aren't enough people living that far North (sorry Scandinavia, you'll have to wait) for it to be worth spending 30% of the satellites' time over the poles when they can be over more populated parts of the planet instead.
Electric hall effect thrusters that use krypton as propelant. These give the individual satellites a really efficient thruster with a lot of delta v. They also use krypton instead of xenon, which everyone else uses for electric thrusters due to it being cheaper.
This is specifically not a concern for Starlink, which is in an extremely low orbit. Dead satellites (and any theoretical debris) would deorbit and burn up in less than 5 years. End-of-life satellites are intentionally deorbited immediately.
Additionally, the starlink satellites perform active collision avoidance.
>Additionally, the starlink satellites perform active collision avoidance.
Is that currently true, or a planned feature in the future? IIRC the whole Aeolus kerfluffle happened because someone wasn't paying attention to their pager.
Here is my attempt at estimating actual coverage, based on a 25 degree angle from the horizon https://i.imgur.com/poj8cae.png
I've included a screenshot of the code I modified so people can play with it themselves. I made a slight simplification in my math (explained in the comments in the screenshot), but I think it should be close enough. Edit: Pictorial explanation of simplification https://i.imgur.com/XtADkmh.png