Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Blog Public Beta for Starlink Internet Service Coming Soon

This site may earn commission on affiliate links.


SpaceX Chief Executive Elon Musk said the company plans to roll out a public beta of its Starlink broadband service soon.

Musk said on Twitter that a “fairly wide” beta version of the service will be ready for action as soon as the latest satellites “reach their target position.”Musk said the service will provide internet to the northern United States and “hopefully” southern Canada.

SpaceX is building a constellation of satellites to beam internet service around the world. Musk has said Starlink will require about 12,000 satellites to reach that scale. SpaceX tweeted earlier this month that initial tests of Starlink show speeds capable of streaming “multiple HD movies at once and still have bandwidth to spare.”

SpaceX has grown the Starlink constellation to around 800 satellites to date, including 60 that were delivered Tuesday in the 13th Starlink mission to date.






A tweet last week from Washington state’s Military Department and Emergency Management Division revealed the first known application of the service. Starlink was employed to provide service to emergency responders battling wildfires in Washington state. SpaceX Chief Executive Musk replied, saying the company is “prioritizing emergency responders & locations with no Internet connectivity at all.”

Photo: SpaceX

 
Last edited by a moderator:
  • Helpful
Reactions: Snerruc
This calc
What's the coverage area in terms of degrees latitude of a single Starlink satellite at any given moment? : Starlink

says (if I am reading correctly) that a sat at 550 km above the Earth projects a circle of coverage equal to 2.78E6 square km.

Is that correct ?
For reference, the surface area of the earth is 510E6 km*km

I infer then that ~ 150 sats covers the Earth between +/- 53 degrees lat, and every 150 sats more adds a multiple of bandwidth. Ballpark ?

we know there are almost 800 in service now and we don't have 100% coverage between +/- 53 degrees lat yet.

It will take more like 1440 sats to cover the earth 24/7 between +/- 53 degrees lat (with multiples of bandwidth at 53 N/S as compared to equator).

Then there will be an additional multiplier in bandwidth (2x) when the next 1440 are added to that.
 
I'm happy to read a lecture for dummies but I should say upfront that I have been presuming that an homogenous distribution of sats is neither desirable or technically required. So while I get the notion that 45k sats spread evenly over the earth surface results in about a 100 x 100 km cell area and it has a ~ 16 Gbps bandwidth, I accept that sat coverage overlap can change the bandwidth areal density drastically.

Few things:
--Homogeneous distribution of satellites is impossible
--The limit on data for a single satellite is primarily determined by its beamwidth (there are hundreds if not thousands of "hopping" beams on a starlink), or more precisely the area that beam covers on the ground, and the PFD (power flux density) of that beam. The narrower the beam the better, because you can more or less transmit the same amount of data to fewer users, but the narrower the beam the more complex everything gets (you need more beams which means you need more processing which means you need more power which means your sat is bigger, and so on). PFD is basically a hard limit regulated by the ITU and FCC. You can't go above whatever level you get approved for. The combination of the two makes for an asymptotically more difficult effort to add user density. As @doghousePVD has indicated, ~hundreds of users per square mile is about the limit of 45k worth of satellites (which, again, will not happen for a decade if ever), and that's basically an order magnitude more rural density than the burbs. So, satellite internet WILL NEVER BE the primary connectivity solution for members of the human race. From a residential perspective anyway, it will be for infrastructurally underserved people that actually have enough money to afford internet in the first place, which, whether we like it or not, is the US and Canada for the foreseeable future. Of course there will be global customer base, but the majority of paying customers will be in North America.

Can you explain why the first ~ 600 or sats provided coverage to Northern USA and Southern Canada ? How wide a west-east swath did it include ?

To add color to what's already been stated, the way orbits work, satellites converge near the top and bottom of their orbits (from a latitude perspective) and are generally farthest apart from their neighbors at the equator, which means any constellation will always have higher satellite density near the high latitudes (plus and minus) and low density near the equator. So even if you don't have enough satellite density to provide meaningful service at the equator you likely have enough sat density at high latitudes to provide, in this case, beta service. These high density bands circle the globe in the high 40's and low 50's equally in the north and south, but Starlink sats also [currently] need to have a ground station in view to provide service. SpaceX wisely decided to put their first ground stations in North America, which is why the beta is in NA. Users in Europe have the same satellite density as an equivalent latitude in NA, its just that there are no (or at least not a useable density) of Starlink ground stations there yet.

Here's a decent animation, but there are plenty out there to study:

I'm not understanding why a ring of reception at ~ 50 degrees lat developed from an orbit inclination of 52 degrees relative to the equator

Those are the latitudes where there is the most satellite density. Users above 53 degrees can see satellites as well, but someone at 55° will see many fewer sats than someone at 51° because of the way the trigonometry works.

I also do not understand the calculation of how much surface area on earth is covered by a single sat (ignoring the fact for a moment that the sat is navigating the globe, just imaging it standing still for a moment.)

The satellite's field of view is a cone, which projects on the ground as a circle (just like a flashlight). The radius of that circle is ~1000km, so each satellite can see ~3M km^2.

The issue is that the satellite cannot provide full service to everyone in that area, so the constellation has a gazillion satellites such that there many many overlapping coverage circles, allowing the constellation to intelligently manage traffic by balancing otherwise over/underused satellites as well as onboard resources like processing power, DC power, thermal, etc.

As to why .8 degrees difference that might be to keep them from having the same clumping all the time?

Depends on how much higher the orbit is, but if they're very close in altitude the inclination offset prevents upstairs neighbors from flying over downstairs neighbors for an extended period of time, instead resulting in more crossing orbits.
 
Why do we see criss-crossing orbits in the animation ? I thought all the orbits are ~ 53 degrees inclination ?

They are all at 53°, but the orbital planes all rotated ever so slightly around the spin axis of the earth (Eventually I think there's suppose to be 72 planes?, so in radial threespace they explicitly all cross every other plane. The angles between planes are largest at the equator where the planes are farthest apart and closest at the top and bottom where the planes are almost on top of each other.
 
Does that mean they start off at 53 degrees, and then rotate ? And that they are out of phase with each other ? Is the rotation a wobble, or do they complete a 360 degree rotation eventually

So technically yes the planes all rotate around the earth (which is rotating under the planes at a different rate) and technically yes there is some small degree of "wobble" that each satellite experiences that needs to be corrected from time to time...but...in context of the question, no to both.

Each plane is at a 53 degree angle to the equator. Each plane remains at a 53 degree angle to the equator forever. The key is that each plane is rotated a bit around the earth's spin axis from its neighbor plane. In Starlink's case, for 72 planes (we're not there yet) each plane is rotated 5 more degrees around the earth's spin axis (360° divided by 72 planes = 5°) then the last plane. There's an orbital parameter for this angle, usually called RAAN (right ascension of the ascending node), and if you're interested in learning more check out 'orbital elements' or 'Keplerian elements'. ...But the main point for visualization is that satellites from one plane are crossing the equator exactly 5° east/west from the satellites in their neighboring planes.
 
  • Helpful
Reactions: SageBrush
I'm starting to follow -- Thanks !

Why do we see criss-crossing orbits in the animation ? I thought all the orbits are ~ 53 degrees inclination ?

Let me share my links, take a deep dive and view a few of these and see if it doesn't make more sense


* for signing up for service availability notification
Starlink

* for seeing upcoming launches
SpaceX Launch Manifest

* for seeing Starlink with your own eyes standing on the ground
️ See A Satellite Tonight ️

* for seeing how good coverage is in your area
Starlink daily coverage estimates

* for seeing where ground stations are being built
Starlink US gateways - Google My Maps

* for seeing how long it takes sats to raise to proper orbit after launch and to visualize where those sats are in relation to you
Starlink - moesalih (I call this one "candy color" because it color codes sats per launch)

* for a nice 3d render
Stuff in Space

* simple range circles
Starlink satellite tracker

* for the deep dive
CelesTrak Orbit Visualization


on this one I prefer to

* go to the menu at top left - viewer options
* switch reference frame to FIXED.
* exit the viewer options with the X at the top left
* drag the earth around to look at your part of the world
* zoom in if you want, but not far. You need to be able to see thousands of miles around your area to have an idea if sats are nearby or not.
* drag the dark ball near the play button down and to the right until the numbers inside the arc are similar to 200 x 59
* click the play button


when you see a train of 40+ sats go around you can easily see what the movement is like and when you see individual sats from separate orbits curving around near 53 degrees you'll see the interaction and you can speed it up or slow it down until you understand it.

once you've internalized the planes and motions that are happening you'll be able to go back and read articles, questions, answers and see it all differently than you do now.
 
Last edited:
What are the ground stations, for? The end user devices connect to a dish in the user premises, which directly talk to the satellite, correct?

Ground stations are also called ‘gateways’ and they’re the constellation’s interface (‘gateway’) to “the internet”. Only in very rare situations will there be 100% Starlink user to Starlink user traffic, everything else has to drop into (or out of) terrestrial infrastructure in order to actually have any useful connectivity to the rest of the world.

Right now, every sat needs to see a gateway and so the link goes something like:

User Downloads:
—comes from “the internet”
—goes up the forward gateway/feeder link
—turns around in the satellite and gets sent down the user forward link to wherever the user is

Uploads
—goes up the user return link
—turns around in the sat and goes down the feeder/gateway link to one of a handful of ground stations.
—drops into “the internet”

Once SpaceX turns on the optical inter-satellite link network there will be sat-to-sat communication than can hop through space to a ground station near the ends of the data stream, be it the user out in the woods or the server/data center/whatever. That doesn’t change the paths from above but for long distances (like, trans oceanic) it means the traffic can follow a more efficient path through the satellite network rather than bouncing from fiber hut to fiber hut on the ground.
 
What are the ground stations, for? The end user devices connect to a dish in the user premises, which directly talk to the satellite, correct?

They have 2 purposes / 2 ways they are used not counting the incoming vs outgoing distinction.

1. They allow traffic to go from satellite to satellite. See the video below for this concept in depth, very nice visuals and voice over walking you through some of the important concepts. Just beware that the video is 10 months old and the network structure has changed slightly since then so don't get hung up on exact numbers, just use that video for the general concepts.

2. They allow traffic to go between the internet and starlink customers (to and from) as discussed by bxr140 above.



@SageBrush this video will help you as well, quite a few visualizations here that would highlight some of the scenarios you were asking about.
 
Last edited:
The key is that each plane is rotated a bit around the earth's spin axis from its neighbor plane. In Starlink's case, for 72 planes (we're not there yet) each plane is rotated 5 more degrees around the earth's spin axis (360° divided by 72 planes = 5°) then the last plane
Ahh ... I'm pretty sure I get it now. Thanks for your patience.
Fwiw, my error was in imagining a series of orbital planes all in parallel due to the 53 degree inclination requirement

I am still light years away from being able to imagine sat density
 
Last edited:
Funny nobody talking about the subscriber terminal. Needs decent antenna gain and broad aperture to communicate with LEO sats. Likely won’t be cheap unless heavily subsidized. How much of the sky can you see? In the woods and peeking through the trees?
Terrestrial WISP is likely more practical.
 
  • Disagree
Reactions: doghousePVD
Funny nobody talking about the subscriber terminal. Needs decent antenna gain and broad aperture to communicate with LEO sats. Likely won’t be cheap unless heavily subsidized. How much of the sky can you see? In the woods and peeking through the trees?
Terrestrial WISP is likely more practical.

I expect that we will see some WISPs using StarLink for backhaul, as long as StartLink has an offering that permits it in the TOS (and probably a few that don't care to violate the TOS).

We will have to see what the final terminal equipment costs are, and the overall growth of the network, but it still stands that StarLink in a lot of ways will turn the "rural has sucky bandwidth" on it's head.

Once the inter satellite links are deployed, you could actually have better bandwidth in the middle of the Ocean vs a large city.

-Harry
 
Fwiw, my error was in imagining a series of orbital planes all in parallel due to the 53 degree inclination requirement

It’s a spatial visualization that makes sense once you get it, but it does take some staring for it to click the first time. Everyone learns different ways of course, but for me its easiest to actually look at an rendering vs reading text. Check out the various starlink animations and try to isolate one orbital plane, then try to isolate its neighbor, etc.

Another good exercise is to check out animations for smaller constellations--GPS is a really good one because the altitude is much higher than LEO and there's only a handful of planes (I think 4?) so its not quite so busy to stare at while you wait for the pirate ship to come into focus...

Funny nobody talking about the subscriber terminal. Needs decent antenna gain and broad aperture to communicate with LEO sats. Likely won’t be cheap unless heavily subsidized. How much of the sky can you see? In the woods and peeking through the trees? Terrestrial WISP is likely more practical.

We've talked about it a bit in the main starlink thread. (This thread really is a bit superfluous, or at least not staying on the intended topic). Reports are that cost is high (~$1k or so) and difficult to bring down; Who knows what the price will be. But...we also know there's functional production models out there supplying good rates, so there's no reason to believe there are any major technical issues. Just pretty high dollar phased arrays (and apparently a motor or two).

Its unclear how much of the sky the terminal needs to see--we know the look angle is 25° and if I had to speculate wildly I'd guess that with a full phase 1 constellation of ~4k satellites any single UT needs a clear view to probably 20-40% of the sky above 25°, generally more north (at least in the northern hemisphere). A full view or near full view is an unreasonable ask for a lot of potential users and I'm sure SpaceX has thought of that. I've speculated in the past that the UTs will basically map their FOV and so will know where and where not to look. Even with a relatively small % of useable view there will be enough satellites passing through to maintain a link, and its reasonable to assume the Starlink network will input all of those partial user FOVs for the purpose of efficient (and continuous, obviously) traffic routing to all of those users.

Yes, WISP, or at least the future of terrestrial wireless (traditionally, WISPs kind of suck) is likely going to be the major competitor to Starlink. As I intimated upthread, we're in the calm before the storm relative to the auction coming up in December. Certainly the service types won't completely overlap, but you can guarantee the terrestrial giants we all love to hate (suck it Comcast...for no particular reason this time other than you probably deserve it and you definitely will deserve it at some point) will push deeper into the less densely populated regions of the first world with wireless where its currently unprofitable for them to lay cable to a bunch of users...or in other words, Starlink Country.