Good for them. Bringing production costs from $3000 per terminal in 2020(?) to $1300 in early 2021 to less than $600 now is quite the achievement, and not subsidizing terminals will make rapid expansion a lot cheaper.
Honest question: rapid expansion to where exactly? Are you thinking poorer, rural countries? E.g., with cheaper terminals Starlink will more aggressively sell to rural Brazil, Mexico, India, etc? If they can sell the terminal at cost, how low can they price the monthly service for these poorer countries to fill up unused capacity?
I live in the LA-metro area, 15min away from downtown LA This is a pretty dense urban area, and we can only get Spectrum, which goes down at least once a month for hours at a time, in the middle of the day
The only other alternative is getting a microwave antenna for $300/mo with a 4 year contract from a company called Geolinks
Had a similar issue when living in SF
Some buildings, and sometimes entire neighborhoods or small cities, only have one (crappy) vendor
I’m no fan of Starlink or Musk, but there’s definitely a market for them in well established, densely populated areas in rich countries
The reason they specified sparsely populated areas is that Starlink is designed for sparsely populated areas. In a denser area the performance would drop drastically because you'd need far higher satellite density to sufficiently distribute the load.
That may be true with current space vehicles, but there’s no fundamental reason that two or more SVs can’t form a distributed phased array (for MIMO a la WiFi). At least from a physics / spectrum perspective. The engineering might be tricky.
With SVs separated by hundreds of km, the beam width would be absolutely tiny, and channel reuse would be effectively infinite.
I imagine the total cost of installation and ongoing maintenance of all the infrastructure required for this proposed setup for a substantial amount of urban users dwarfs the equivalent total cost of just installing new cabling and the lobbying required to get that done, especially over a 20 year period.
The distributed array will however share one link and satellites don't have unlimited capacity either.
Wonder why you'd rather fill LEO to the brim, emitting vast amounts of CO2 in the process and filling stratosphere with metal oxides, than figure out how to improve infrastructure in your neighborhood.
I've never heard anyone complain about filling the stratosphere with metal oxides, do you have somewhere I can read more about this risk? Naively, I would have expected that even thousands of 250kg satellites would be insignificant on the scale of the whole stratosphere (we're not talking about surface-level CO2, where we're pumping out gigatons per year), but I have no idea what that ecosystem is like. Is the metal oxide concentration measurable?
I do think this is a temporary issue; once the cable ISP monopolies are no longer valuable because anyone can just get Starlink, that competitive pressure will push down hard on those cable monopolies and infrastructure laws will have to return to sanity. Terrestrial ISPs with fiber-sharing requirements and appropriate infrastructure build-out could absolutely out-compete Starlink, but right now they can just collude together and lobby against changing the status quo.
I would much prefer to have a functional bureaucracy and intelligent, efficient infrastructure! But it's probably easier to fix that by launching the largest satellite constellation ever than by playing politics against some of the wealthiest media companies on the planet.
"Space junk," was a boogie man the news threw at the wall and didn't quite stick. Many of the pictures/diagrams of the satellites made them appear about 1000x bigger than they actually are.
I agree with the sentiment about media sensationalism but I don't think "space junk" is an issue we should be so quick to dismiss.
Just because the problem hasn't gotten too bad yet doesn't mean it isn't real (especially because the problem involves a positive-feedback "domino effect" and junk might be infeasible to clean up).
With regard to Starlink: I was a little bit concerned because of the large number of satellites they're putting up, but as far as I understand, Starlink isn't actually a big risk right now as far as space junk goes because the satellites orbit at low altitudes (340 miles) and are expected to naturally de-orbit relatively quickly due to atmospheric drag. (The decision to put them at low altitudes had to do with network latency.) If Elon were to say tomorrow, "hey we're moving Starlink out to 500 miles" I would hope some kind of regulation would be able to prevent that.
Everything we launch has to pass through these low altitudes. If these altitudes become full of debris we might have to stop all launches for several years [1] and no human presence at space for a bit longer. Quite scary, better to handwave it away "aint happen space is big".
If they can get the interconnect working that issue goes away (its mostly groundstation bandwidth limiting things).
The 'end goal' is to have it use the laser link to send the traffic across to a satellite close to the final destination, that should in theory help massively redistribute the ground station traffic.
It's obviously not foolproof, there's still bandwidth to be shared there but using a combination of some clever routing you could theoretically spread traffic across multiple satellites and groundstations with minimal latency impact.
It’s actually not the satellites as the bottleneck typically from what I understand, it’s the ground stations that serve those areas getting too congested.
The satellites may not currently be the bottleneck for certain areas, but they definitely don't support anything resembling a densely or even sparsely populated area.
Typical calculations (using only publicly available info as a caveat) generally point towards the current starlink constellation supporting a really low density of antennas.
If you want the details, the person who created starlink.sx has a great blog with some incredible calculations[0]. The section that is most relevant is under the heading "Simulation with TDM and beam spread combined".
Basically, the result is that starlink could support around 15 antennas/user terminals per "cell" at 75mbps, where a cell is about 252 square km. This is roughly 1 terminal per 17 square km.
If you allow for 10x oversubscription, fairly common for ISPs, that gets you to 1 terminal per 1.7 square km. Not really that impressive when you compare it to the density and services where 99.9% of people live, but it clearly helps out the small number of people way out there.
He's the driving force behind some of the greatest innovations of our time. He's also just a person. I doubt many people can say they've had greater impact or could withstand the same scrutiny.
People would have a much better opinion of him if he stuck to realising crazy projects (an electric car thats faster than a supercar, reusable rockets, solar panels that look like normal roof tiles...) and kept his questionable political opinions to himself.
Everyone is free to have their political opinions.
The issue with Musk is that those opinions are being manifested in the product decisions at Twitter e.g. taking a laissez-faire attitude towards hate speech and deliberating promoting controversial accounts in order to drive engagement.
i'm not sure what is the source of that but I've noticed that a lot of people has the need to take everything personally, and can't separate things.
for example I really like tom cruise as an actor, one of my personal top 10, but I don't like what he is doing in his private life (the cult thing). so what i'm doing is separating things and take what I like, and ignore (to some extent) what I don't like.
I feel like you guys take things way to serious, and expect everything to be perfect, which is not possible.
Not going to argue with the "greater impact" part, but mean pretty much every C-level executive in America with a twitter account manages to refrain from using it to repeatedly commit securities fraud.
Adding "I'm not fan of Musk" is odd here. Is this some kind of tribal signal? It's like writing about the market for new AWS services, preceded by "I'm no fan of Bezos but..."
I could be wrong, but I would bet you'll start to see alternatives available to you within the next year / 24 months. There's so much federal money flowing into fiber right now...
Civics and industry co-exist in a healthy society. Musk has consistently demonstrated that he has no civic sense and, when injured, will randomly lash out at whoever the hecklers around him tell him to lash out at.
That's not a quality that's worth celebrating, regardless of how much innovation he might enable. We can and should expect better of people, especially extraordinarily wealthy and powerful ones.
What about the civic sense of proving that great electric cars can be made at a profit, space launch costs can be cut 90%, and Starlink can enable communications by people who live under oppression?
Doesn't actually doing things that matter to the human race matter far more than saying comforting words and avoiding giving offense?
I don't think any of these things imply civic sense. At best, it shows a man with decent business acumen. At worst, he's an opportunist and a reactionary who has substantially degraded the level of political and social discourse of the country he lives in.
Of the things you've highlighted, Starlink has perhaps the strongest case for being evidence of civic sense. But that case is a potential one; in practice, Starlink (and Musk's vision) has yet to impress[1][2].
> I don't think any of these things imply civic sense.
I recommend a biography of Musk.
He makes his electric car patents freely available, to advance the transition to electric cars. He started SpaceX with the purpose of making humans a multi-planetary species. He's repeatedly bet his entire fortune on these goals.
He's civic minded where it counts. And he's delivered results.
As for business acumen, he takes financial risks any other businessman would consider to be insane.
His accomplishments are enormous - showing how to make electric cars at a profit, and cutting the cost of space launches by 90%. And now Starlink, too.
Standing up for free speech includes standing up for the right to express ideas you don't like. I'm sure he'll allow you to post your opinions, too.
Suppressing free speech is the first step towards tyranny.
Just be mindful that the US is the only country that has this attitude towards free speech.
Everyone else including Western countries like UK, Australia, Canada etc have a different take. Which is that hate speech is not conducive to a healthy society and that it doesn't lead to tyranny. Whatever that even means in modern discourse.
They’re rapidly launching more and more capacity. The new v2 mini satellites are more than twice as big as the v1.5 satellites, which themselves are not yet totally in service.
And as they improve the cost of terminals, there’s no fundamental limit in the amount of capacity for a given frequency allocation. Geometric multiplexing is only limited by the number and cost of the phased array elements.
So rapid expansion everywhere.
And then launching the full sized v2 Starlink satellites on Starship will account for another ~2-2.5x increase in size compared to the v2 minis. Then comes FCC approval for the full sized constellation which is about 5 times as numerous as their current constellation, which itself only is about half filled.
So they have about a factor of 100 capacity expansion in the pipeline, and they can increase the satellite size even further due to Starship’s low costs. They can expand a lot, even in existing countries. Lots of people are on waiting lists due to capacity being filled up in many regions.
> Then comes FCC approval for the full sized constellation which is about 5 times as numerous as their current constellation, which itself only is about half filled.
SpaceX has launched just under 5,100 Starlink satellites to orbit, of which a single digit percentage have either been deorbited for testing, failure, solar storm knocking them out, etc. Their current constellation total size is approved to be around 12,000 satellites. They've been working to get additional approvals up to 42,000 satellites.
Phased arrays aren't magic. The sharpness of the aperture depends on the physical extent of the antenna and very sharp mainlobes come with additional sidelobes.
There's an additional consideration regarding the price, that poor places can probably share one terminal among one or more villages, thus distributing the cost.
Another possibility is distributing costs by using Starlink as backhaul. India for example already has very cheap high speed mobile data, so a telecom company could spread out the costs of terminals and service (maybe even get a better deal for bulk usage) while simplifying and strengthening their own infrastructure for remote or disaster-prone areas.
Plus, for SpaceX, any point where a satellite is inactive is money left on the table, so they should have significant price flexibility for places which can't afford the current service fees but can afford something that is more than the cost of bandwidth at the downlink station.
Poorer, rural countries like 20% of the US. In rural NC, my brother was paying over $100/mo for 2mbps down cable Internet as 2021.
Starlink's customer acquisition cost has dropped $600 (or whatever they were losing on each antenna), so they can afford to advertise way more than before. If they can pick up a decent portion of the 60,000,000 rural Americans, that's a huge revenue stream (and huge market power, for whatever that's worth).
My dad's (now brother's) house a few minutes outside a decently-sized town by Downeast Maine standards had a barely 1Mbps down "broadband" link and it was the last house on the road that could even get that. With Starlink, it's actually viable to work from there in addition to streaming video etc. Of course, he wouldn't pay literally any amount but he'd probably pay a lot; I would.
Yeah, I suspect they could have gotten my brother to pay the old full cost plus extra. Still, many people in rural areas aren't willing/able to drop $1k on Internet setup, so keeping costs down is worth more than gouging a few remote software engineers.
Certainly. I'm sure there's a correlation between poor wired broadband and lower income even if it's not perfect. But Starlink is a game changer for a lot of people who otherwise can't get decent wired Internet--and, especially if you can't get wired broadband, the options are mostly pretty bad. Probably changing some with 5G but still not ideal for many people.
Sattelite communication is generally more expensive than cable or cell towers. Though about 70% of the earth is unsuitable for these solutions. Alot of commercial boats are still on very low bandwith connections but could potentially benefit from higher bandwidth if only to make video calls or netflix more available for a similar or higher price. Then there are alot of smaller, private, boats that would benefit from sattelite communication if costs werent prohibitive.
Large swaths of the US have bunk internet like Earthlink. My rural family in Appalachia already like Musk. I think they'd love to have Starlink if it was a little more affordable.
We live in a very rural area. The one time the federal government really came through for me was fiber development grants a few years ago.
We went from Wild Blue internet, which was garbage all around, for $120/mo to a full gig fiber connection locked in at $80/mo for life. It cost us $200 to hook it up.
More and more rural isp's are getting into fiber with grants.
Wow. When I was a kid, we would steal "ELN" accounts for dial-up and then hop on to AOL via TCP/IP.
Edit: we did that so when we got punted we didn't have to go throigh the whole dial-up sequence. Which happened a lot in those days if you were twelve years old.
I'm always kind of puzzled by this question. There are like 7-8 billion humans. Millions of ships, planes, farm machines and other equipment.
Expansion will happen in all direction at the same time, in all market. In very poor country a single village might share one antenna. All the way to very expensive antennas for military aircraft.
I'd buy service to put it on my rooftop to both put economic pressure on the local ISP duopoly in the form of an unambiguous publicly visible third alternative and, also, as a backup.
This is part of why I bought a Starlink dish and put it on the roof of the house I'm doing a full rehab on before the siding was installed. It will give me dual homed internet for when the crap local ISP has issues and a hedge against a single ISP with spotty service.
Ironically, in doing this, I have managed to meet and convince a project manager of a new ISP to add our neighborhood to the list to places to pull fiber to, but it will likely be about a year out.
I know of a couple places that are using starlink as failover because they're only serviced by a single ISP. Makes sense really, and cheaper than 4/5g failover.
I'm also not sure the financial aspect is going to work. There's not many developed countries without proper fiber rollout and very poor countries likely can't support the high infrastructure cost. You're left after that with marginal use cases (boat, remote places...) which won't be enough either.
There's a reason why the comments are mentioning rural US, because it's a clear outlier among the developped world.
The developing world is typically unequal, not universally poor. There are a lot of people in Africa and South America who would get drastically improved internet with Starlink, and can actually afford it.
I'm sure you can find such places, is there enough of them to fund the service which is very expensive? That I'm not so sure.
When people are richer, they also tend to live in countries where the internet access is very important to the state and where installing good internet is an important public policy.
Of course you have exceptions to that general case but I'm not convinced those exceptions are enough by themselves.
They are in a very different market than a traditional product, they are basically servicing the people that nobody else want to service and the distribution of revenue in that group is going to be drastically different from the norm.
The sattelites give coverage everywhere except polar regions at no extra cost. Africa has 1.4 billion people. South America another 400 million. Even if only the top 2% are your target market that's still roughly the same as the entire rural US.
For my parents it’s replaced their cable internet which was slower and less reliable. I wish I could get it where I live in a city apartment. The internet is usually fast, but not reliable.
The only way to have satellite in an tall apartment building is to put it on the roof, which is complicated.
Apple is offering satellite phone service. Starlink shown it's important in the Ukraine-Russia war. There is Starshield for the government uses. Whatever shortcoming of current telecommunication, Starlink can supplement it.
StarShield came out of that though, which might be a legitimate compromise. e.g. governments purchase StarShield and then just do whatever they want with the constellation.
I don't disagree with your sentiments here. But if you've got a service and it's being used for war (regardless of whose side you're on or how patriotic you are), it's not abusive (from my perspective) to just opt out providing service. If Musk is completely a pacifist, which I doubt but let's go with it, then StarShield might end up being a reasonable compromise.
That turned out to be yet another instance of CNN lying about/acting on insufficient/poor information and has since been walked back by the very person that made the original allegation. Confusingly in direct contradiction with supposed excerpts from his book.
Isn't even the first time they falsely alleged that SpaceX shut down Starlink to cause Ukraine trouble and failed to follow up with a correction as loudly as they made the original claims.
Near term naval and aircraft.
Regarding rural third world: the thing about a satellite is it is basically going around in a circle. If the business plan was built around being profitable over the first world then everywhere not over the first world is basically free because you have to build whatever bandwidth you want to provide the first world around the entire orbital ring.
Starlink is not yet ubiquitous even in western countries (EU, UK, US, Can, ...). I believe the rapid expansion meant becoming more ubiquitous in western countries and commercial operations (shipnav, etc).
They have a long, long way to go before even talking about developing countries which already are densely populated and mostly have fast, cheap internet already.
If they get 10 million subscribers at $120 per month, that's $14.4 billion per year which is way more than the cost of the network. And they can easily get 10 million in just the US and Canada. So they can price new subscribers in poorer countries at whatever price they need because the marginal cost to add to the network is near zero.
"A 2015 presentation that "SpaceX used to raise money from investors" reportedly projected that in 2022, Starlink would hit 20 million subscribers and generate nearly $12 billion in revenue and $7 billion in operating profit. The WSJ said it obtained the 2015 presentation and recent documents with numbers on Starlink's actual performance in 2022.
Actual Starlink revenue for 2022 was $1.4 billion, up from $222 million in 2021, according to the report. The documents apparently didn't specify whether Starlink is profitable."
Additionally, holy crap has a lot changed in that time. Falcon 9 didn't complete its first landing until the very end of 2015 with Flight 20. What's now routine and key to their costs is a feat less than a decade old. SpaceX was still talking about the Mars transport/BFR at that, which were very different in a variety of ways from the Starship SS/SH system we see now. What they've got now feels a lot more practical and sustainable long term, moving to steel for example was a great change.
So as far as long term forward looking plans go I'd agree that execution here was pretty darn good, and they've got a very clear roadmap for the next 5 years as well with full size v2 sats waiting in the wings and Starship proceeding along at a solid clip. It was a curiously negative feeling piece particularly for the WSJ. How many companies execute ambitious highly capital intensive global scale infrastructure rollouts using never been done combos of tech at never been done price points and pull it off like that?
60 million people live in rural areas in the US, over 40 million of those already have broadband at home[1], and less than 2 million people subscribed to satellite internet before Starlink launched[2]. His projection of 20 million subscribers would have meant that literally every single rural individual who didn't already have broadband would sign up for Starlink within a couple years of being available. Okay, that's not entirely fair because Starlink does business in more than just the US, but still, that was an unrealistically aggressive projection even 7 years ago.
I think Starlink is doing great, but Musk's projections are notoriously bad, and there is no reason to defend them.
This is misleading for several reasons. The FCC defines broadband as 25/3 mbit up/down, which is not fast. Also it doesn't take into account latency which is a major problem with geostationary satellites. So plenty of those 40 million with "broadband" are going to want to switch to better service anyway.
Your first article concedes as much:
> Even though rural areas are more wired today than in the past, current infrastructure does not support consistently dependable broadband access in many rural areas. This lack of reliable high-speed internet access has come to the forefront of discussions about navigating remote work and school during the coronavirus pandemic.
Your second link states that there are 1.7 million satellite internet subscribers in the US. But Starlink already has that many subscribers, so it is clear that the market size is much more than just existing satellite users.
Even if 100% of rural customers were dissatisfied with their existing service, projecting that you will capture a third of the potential market within a few years of providing service is unrealistically aggressive.
> until they stop selling a dollar for 90c, then they don't have real business model
Starlink is glued to a transport monopoly. It doesn't take much imagination to see how economies of scale on production and marketing make those numbers work.
Amazon sold a dollar for 90 cents for 20 years, and I don't see anyone laughing now. It's not a bad business model for a massive fixed cost but minimal marginal cost business like Starlink.
It does have minimal marginal cost, as long as you live in the right place on earth, without many other customers next to you.
Once you exceed, fairly limited amount of people that can use it one cell, you need to grow massively your fixed cost, as you need to grow the whole constellation or just won't be able to grow.
Time will tell if that will add up. I don't think it's obvious it will.
The criticism chain for Elon Musk's ventures is usually "It's impossible! This is a scam!" -> "He promised he would have X five years ago!" -> "He has too much money / power / a monopoly! Something has to be done!"
Starlink was good but not for me because I was surrounded by trees. I had an interruption every 3 mins. Sometimes for a couple of seconds and sometimes for a few minutes. There’s no disruption when you’re streaming things but it makes video calls almost impossible which is what I needed the most for my job. Ended up keeping starlink and reenabling my Viasat system just for the video calls. I have a wired fiber connection now and it’s heaven.
Starlink has historically replaced them for free if lightning strikes them (this was early in beta when I paid attention, so the policy may have changed).
As far as I know, the terminals were mainly so expensive due to the large phased array antenna. I wonder if these cost reductions are mainly from managing to produce them cheaply and if that might be valuable as a commoditized phased array for other applications.
Thanks for the clue, apparently it's the GaAs MMIC technology according to the ChatGPT-4 [1]. I've should have figured it out in the first place due to my previous research work of Gallium Nitride (GaN) technology that now become the successor to GaAs technology due to high power Radar requirements. I suppose this is where ChatGPT become really handy for research, as Google is extremely useful for search.
Similarly with LIDAR, the autonomous vehicles demand has made the LIDAR sensor now become very affordable for the masses.
[1] ChatGPT-4 response:
"The key component of the F-22's APG-77 phased array radar that was made cheaper by Wi-Fi technologies, according to Bill Sweetman's reports, is the Gallium Arsenide (GaAs) Monolithic Microwave Integrated Circuit (MMIC) technology. This technology was significantly reduced in cost due to its widespread use in commercial Wi-Fi systems, allowing the defense industry to leverage economies of scale and reduce the overall cost of producing the radar system for the F-22 Raptor."
I would be surprised if the older, higher costs didn't include broker fees for scarce ICs in the supply shortage. I'm sure they've gone through a few revisions but part of that cost drop is simply price of components dropping.
There's always a guy in these threads and this time it's me.
Sure wish they had an a-la-cart option. We're in the Trailer 6 weeks a year, over 5 trips, and starlink internet access is $150 a month, for the full month.
This isn't unique to starlink, if I need any kind of bandwidth, it's $80 for a 100gb month of Verizon.
I'm already paying for tethering on the AT&T phones, and Comcast at home. This stuff really adds up if you're doing any kind of travel.
Sounds like a secondary market opportunity. Keep 20 Starlink subscriptions, rent them out for a few days at a time at $8/day. $600 deposit on the antenna.
Do you mean an option to pay per Gb of usage, rather than monthly?
At least its a possible offering now that they won't be losing money on the terminal (thus necessitating predictable time-to-recover via monthly subscription).
That said, I wouldn't hold my breath. There are good reasons very few businesses offer that option alongside monthly plans.
I'd be good for that, but those situations generally end up being more expensive than just buying a month.
Not only that, it's trivially easy to have a single poorly choked device blow through a cap. (e.g. one laptop decides it wants to download paath updates...the Tier you have is $60 for 25gb or $80 for 100gb...or, you know, $150 for uncapped starlink.)
The amount of brainspace required to have reliable internet access in an RV is crazy....and it won't be cheap.
Does Starlink offer a plan for multi-tenant buildings? I'm thinking condo with 30 units. It seems like it should be more efficient to run one high bandwidth connection than multiple regular customers. Could we share cost among all tenants and get a really good deal?
>Does Starlink offer a plan for multi-tenant buildings? I'm thinking condo with 30 units. It seems like it should be more efficient to run one high bandwidth connection than multiple regular customers. Could we share cost among all tenants and get a really good deal?
I'm not quite sure what you're envisioning here? Starlink offers a business class service with a more powerful terminal, higher QoS, etc. One just gets that and then have it serve as the uplink to a LAN which you can subdivide in all the normal ways however one wants, that's what I've done with Starlink for multiple clients now including one that then feeds to a dozen or so different units over a half mile or so. There isn't any need to SpaceX to be involved in what happens after their terminal, it's no different than a fiber connection WISP or ADSL or cable. Everyone can agree to divide up paying for service and kit as they wish. Same as if it was going to a business that has 2 employees or 20 or 200 or 2000, SpaceX only cares (and should only need to care) about the bottom line of the monthly bill getting paid and any acceptable use/data TOS respected. They don't need any visibility into the network beyond their terminal. And Starlink can be part of a failover or more complex load balanced WAN too.
Starlink's ToS prohibit reselling of service, and dividing cost of service among several tenants could be considered to violate that. Most ISPs would require a specific agreement for those sort of things.
>and dividing cost of service among several tenants could be considered to violate that
Show me the caselaw where multiple people splitting a bill would be "reselling", not your head canon make believe if you're going to make such an assertion. ISPs grab enough power already without us needing to invent new things they can do. Reselling as far as I understand means taking a good or service and then selling it again for a profit. I'd really like to see any examples of people at one address being sued for privately deciding to share a bill. Also FWIW I don't think it should be legal for ISPs to prohibit reselling anyway, but I don't dispute they probably can in theory. Splitting a bill though?
>Most ISPs would require a specific agreement for those sort of things.
>He's right. Usually you need an agreement to resell.
That's not the argument though. The argument is if a few people living together in a building decide to just divide up a bill counts as "reselling". Or one person pays the internet bill and another then pays some other utilities or something, which is legally the same outcome. Ie, $90 month, each pays $30. It's not getting resold as a whole new connection. I really don't agree that, even if it was possible for them to find out, ISPs would go after anyone for that. I've never heard even a hint of it in 30 years, but I don't have a global perspective either. Is this actually a thing in some places? Like 3 friends rooming together somewhere and they're all supposed to buy their own separate WAN not divide it up like other utilities?
If you're rooming that's one thing. If you have different addresses for mail delivery and utilities that's another matter. I've seen someone have an ISP refuse. They had to cancel then buy the same exact service over the same exact Fiber from a reseller. Imagine Verizon/MCI refusing to allow you, then you buy service from Zayo over the same Verizon fiber that never got physically removed.
It still had limitations but it allowed them to deliver service to an entire neighborhood.
>> He's right. Usually you need an agreement to resell.
I wasn't thinking of reselling, more like cooperative buying. We'd have co-owners pay for it with their HOA fees and just let everyone in the building use it. Obviously we'd have bandwidth limits per connection, but I'm thinking several wifi APs around the building would do the job and nobody would have to pay an ISP directly anymore.
That's doable. The HOA doesn't sell internet. Just pays for a Wi-Fi network that happens to reach you.
I'd advice two high performance dishes (dishes are known to fail and support is an issue so one on standby while waiting) and a business connection. You'll need a third party router with fair queuing, protocol and service speed shaping etc etc. I'm sure both openwrt and opnsense will do. But check this out https://libreqos.io/
I'd assume the likes of airlines and cruise companies would all be cutting custom special deals with SpaceX, just like government, large enterprises etc? For SLAs, prioritizatoin and support if nothing else, if you're going to be dropping 6-8 figures with them the public deals are probably no longer the only thing on the menu. We know the DOD at least has custom projects going with SpaceX wrt Starlink. No idea how independent hotels, B&Bs and other small businesses might work with it though. Maybe that just comes down to selective enforcement, SpaceX has it there as one of a few possible clauses to respond with if they feel some entity is being abusive or making too much off of it. Still relatively early days too.
There's gotta be a point where the cost of running enough fiber drops below a sufficient amount of starlink capacity though. Hard to say where that number is though.
You'd be surprised. In dense cities like NYC/Brooklyn fiber availability varies from block to block, I assume because adding new infra is so expensive. We recently moved about five blocks away and lost access to fiber :(
(Another oddity; the only way to drive into that down is through a 2.5 mile tunnel with a single lane shared by trains and cars. You drive on the train tracks.)
I want to like Starlink, but there is no way the product can both continue to grow and also service customers at advertised speeds. It will literally only get slower over time. And it's already spotty.
Ultimately bandwidth per customer is just a revenue vs cost of each satellite question. They or a competitor can scale the number of LEO satellites and their mass basically indefinitely.
Also inline traditional communication satellites these things have a short lifespan which means the network can see significant improvements over time as technology improves.
From what I heard there is literally a cap on the number of signals that can be serviced per cell, extra satellites will not expand potential signals per cell, so you literally run out of how many people can be connected in a given area. Bandwidth is already limited and becomes moreso as the cell gets crowded.
Cell sizes can be shrunk it’s really dependent of the technology involved. Starlink is currently tied to a specific technology stack but different providers can be using different frequencies etc.
If you’re looking for the ultimate limit you need to consider stuff like this 100Gbps optical link. https://news.mit.edu/2022/communications-system-achieves-fas.... That’s not directly viable for home internet users, but an airline or cruse ship might be interested in such technologies.
In that case, it is possible and feasible to create a third party business making and selling cheaper antenna systems for the Starlink customers.
As I've mentioned in my other comments, I suspect the decreased in manufacturing for the antenna price is not so much due to the SpaceX better manufacturing iterations, increase yield, etc but mainly due to the recent availability of the high speed Direct RF sampling transceivers that now encroaching Ku-band territory.
Amazon claim they'll be able to hit $400. In Ka band
SpaceX spent $2.4 billion on dish R&D and production of 1 million antennas upfront with ST Micro. That kind of commitment goes a long way.
It's probably not. SpaceX doesn't have a separate modem. It's in the dish. These two companies got in via the military. OneWeb is in Ku band like starlink, they may be open to talk
Ka-band is still very high for the Direct RF sampling transceivers, if Amazon's claim is true then it's mighty impressive.
$2.4 billion is a lot of money to spent on antenna design and development, and if they don't have separate modem that explains the astronomical cost due to down/up conversion and perhaps most of the beam steering signal processing are done inside the antenna systems.
If the Starlink antenna has any patent applications, it will be very interesting to see their architecture. I think Direct RF sampling transceivers can make for much cheaper Starlink at least in the Ku-band antenna if their price are becoming cheaper themselves.
Based on this thesis (2021) that proposed a lower cost antenna systems for mobile terminals, Starlink has separate TX-Rx architecture that make their antenna system rather expensive [1].
"The limited operational bandwidth of the radiating elements integrated in their low-profile alternatives (i.e. slotted waveguide antenna arrays) actually obliges to split transmission (Tx) and reception (Rx) in two different panels, keeping a relatively high cost for the end-users."
There's a Reddit discussion on the ST developed antenna systems from 3 years ago [2].
There's also a teardown video of the Starlink antenna systems by Kenneth Keiter [3].
[1] Innovative 3-D printed Ku- and Ka- band antenna solutions targeting cost effective satellite on the move and 5G phased arrays
The $2500 "high performance" (square, not rectangle) dish is a direct ethernet connection.
They still use a proprietary connector (mainly for weatherproofing and durability reasons) and supply the power on different pairs than standard PoE (probably so people don't try to use their own PoE injector). But it's:
dish -> PoE injector -> rj45 ethernet
No wifi anywhere. Also no router, unless you count the dish itself. Each "->" is a cable.
The dish->injector cable has proprietary connectors on both ends; the injector->rj45 cable has a proprietary connector on one end and an ordinary rj45 plug on the other end. Apparently this is done because the proprietary connectors are weatherproof and rj45 isn't, and people expect to be able to put the PoE injector outdoors.
The proprietary connectors on the high performance dish are slightly different from the ones on the cheap rectangular dish, so you can't just use the proprietary->rj45 cable from a high performance dish with the cheap dish.
Honestly I with Starlink would've just coughed up the few extra bucks it costs to use standard rj45 with a weatherproof shroud.
I was very surprised when our starlink showed up and I had to buy an Ethernet adapter! Extremely weird that they don’t just have the port on the router.
Is it really that surprising? there's two hardware pieces, both are rated for outdoor (permanent) use. I don't see a way to guarantee that sort of thing with a user serviceable ethernet port except the way they did it.
I actually debated buying the longer dish wire vs leaving the thing outside, and made the decision to buy the longer cable and the ethernet port adapter, because i live in an area where sustained power outages longer than 24 hours will allow moisture ingress into any device left outside, and the higher the IP rating, the more likely this is to damage said device.
User serviceable ethernet port on outdoor antennas is standard on wifi devices. Just put it behind movable cover and direct it down, partially recessed in plastic case (so it is protected from flowing water, rain and snow). See e.g. https://www.wifihw.cz/img.asp?attid=4754
that sort of thing works great if you can guarantee power to the device, 24/7. At least where i am. the moisture in the air at night will be pulled into the device, through whatever seals exist, if the device is unpowered. Then, during the day when the pressure inside the device increases (due to the seals), the extra air is released, but the condensation remains.
here, it takes literally 2 nights without power to kill stuff like Wyze cameras, Solar powered spotlights, etc. and those don't have a half inch rectangular hole allowing ingress.
I'm not saying that stuff like what you're talking about doesn't exist. I've installed relays that had "downward facing rj-45 behind a panel", on top of buildings... but the building had generator backup that we were shareholders in, as well as our own UPS systems that kept the POE enabled in the event the generator failed, but shut off all "data" to the radios, leaving them using idle power. If there were no generator, we'd have used a different type of radio on the roofs.
Well the entire rest of my system is a Unifi mesh Wi-Fi system with two Ethernet ports on every outdoor rated device I am using so yes I found it very surprising.
The extra Ethernet dongle is a silly extra expense in my view - I don't want their WiFi, I use the base station as a bridge between physical interfaces the same as a cable modem.
I'm looking forward to the return of the port, too.
Well, it's a few more cents on the bill of materials and some PCB real estate, and constrains the enclosure form factor.
Ethernet ports have a transformer (or some other kind of electrical isolation) inside as well, from memory. Making it an add-on means you can spend more on higher quality electrical protection so it actually keeps the rest of the terminal safe, for those who need wires. And you can replace the add-on when mechanical overstress (aka tripping over the ethernet cable) is applied.
their router can be submitted to the elements, outdoors. good luck doing that with an rj-45 on the device. Now, CPE wireless stuff (like at&t fixed wireless) has ethernet ports, but you have to have bulk cat6 wire, the rj-45 connectors, and a crimper to "service" it. I don't know about you, i'm the only person i know that has all of those things in my area. the dongle to add an ethernet port to the starlink router, on the other hand, just requires removing one gasket-laden plug for another.
I personally use it as strain relief. It's cheaper to replace that dongle than it is to replace the antenna wire, especially the 100' one.
>their router can be submitted to the elements, outdoors. good luck doing that with an rj-45 on the device
Um, that's very, very easy to do actually. And indeed the original v1 Starlink cleanly separated the terminal and the router. There was zero need to ever use the included one, you could just treat the terminal as a pure WAN link and that's it. Works extremely well. The v2 was a significant and regrettable downgrade in that respect. There were some reasonable arguments for going away from ethernet since they weren't within the official spec which unfortunately maxes out too low for power [0]. But I wish that at least with the business one they'd simply separated out power and data at least as an option while keeping the terminal self contained.
>Now, CPE wireless stuff (like at&t fixed wireless) has ethernet ports, but you have to have bulk cat6 wire, the rj-45 connectors, and a crimper to "service" it.
Or, you know, just get a preterminated shielded patch cable? Or for stuff SFP use fiber, again preterminated is fine. There is no downside to having this be easy for those who need it and the same for those who don't.
----
0: Though I'll edit to add that I've had people tell me they've powered them just fine with standard 100W PoE++ injectors. It may be that either SpaceX is being generous with their claimed power needs, or the full power is only ever used for serious snow/ice melting with the heaters and is a worse case scenario that much of the planet will never need to think about. Of course that'd just remove even more justification for going proprietary.
>Or, you know, just get a preterminated shielded patch cable?
the rj-45 plastic part won't fit in the tiny hole for the cable. I'm unsure if you understood what i was talking about. outdoor ethernet requires a crimper, otherwise you can't guarantee water-resistance.
I don't know anything about the "old" starlink stuff, i have the rectangular dish. the power and data are separate - unless you mean the cable going to the dish - in which case, the dongle splits out the ethernet from that cable.
I haven't tried to "dumb WAN" the starlink because it's outside the scope of my use. I don't even see the point of getting rid of the NAT at that spot, because it's still NAT upstream, and you're gunna need NAT on your network anyway, probably. I had this issue with 4g/LTE CPE, too. Once something goes CGNAT, there's no reason to use your own equipment, as it's getting a private address anyhow, and now you're approaching quadruple NAT territory.
>I'm unsure if you understood what i was talking about. outdoor ethernet requires a crimper, otherwise you can't guarantee water-resistance.
No, it does not. I'm fairly sure you do not deal with outdoor networking of any kind, or you'd be more familiar with the myriad of solutions to this. The most basic of which are feeding a terminated cable through a tight rubber gasket of some kind, which then fits around the 8P8C port in some way, and is installed facing up. That is typically plenty for outdoor survivability short of full submersion, including driving rain storms, complete coverage in snow etc. Sometimes dielectric grease or the like may be added if it's going to be buried underground too. Like, this is a solved problem, for ages. Cheap PtP devices, outdoor WAPs, outdoor switches, ethernet extenders etc all handle this just fine. All of them use preterminated cables (of course you can make your own).
>the power and data are separate - unless you mean the cable going to the dish
Of course I meant the cable going to the terminal (dish). I want that in at least a pro/business terminal to have an option to use SFP directly so I can have the data go direct to fiber and power be separate, which as well as allowing far more range to the router has the added advantage of eliminating any surge or induced current considerations.
>I haven't tried to "dumb WAN" the starlink because it's outside the scope of my use. I don't even see the point of getting rid of the NAT at that spot, because it's still NAT upstream, and you're gunna need NAT on your network anyway, probably. I had this issue with 4g/LTE CPE, too. Once something goes CGNAT, there's no reason to use your own equipment, as it's getting a private address anyhow, and now you're approaching quadruple NAT territory.
I'd politely suggest you perhaps not be quite so confident about things you don't really use or understand.
Since you ignored pretty much everything i said for whatever reason, i'll just say it again: it may be "just fine" in places that don't have extreme humidity and temperature deltas between night and day. All i can tell you is i've installed PtP and PtMP sector antennas and "dishes", and if the power goes out for 2 nights, you're at least in for replacing some parts.
It isn't "the rain" that I/we worry about. Oddly, all the starlink stuff is heavily gasketed connectors. i wonder why they don't just use USB-C like some cameras, they could even face the connectors downward!
As far as my last point - i said i don't see the point since starlink is CGNAT. It's like putting a fiber cable to a cellphone for tethering. you get no benefit. In the starlink app you can turn off DHCP and put your own junk behind it if you want - that is, if you want triple or quadruple NAT.
>Since you ignored pretty much everything i said for whatever reason, i'll just say it again: it may be "just fine" in places that don't have extreme humidity and temperature deltas between night and day. All i can tell you is i've installed PtP and PtMP sector antennas and "dishes", and if the power goes out for 2 nights, you're at least in for replacing some parts.
What's your criteria for extreme deltas? I've had PtP/PtMP kit installed in areas that go from above 100 to 35 below zero over the course of a year, 60+ degree 24h gyrations on occasion, with heavy humidity, heavy rain, 90 mph winds, unprotected on top of buildings, mountains and in forests. None of it has had the trouble you're suggesting over the course of a decade. I've used those mikrotik extenders to join two terminated cables underground in wet earth buried for years and they're fine. I'm legitimately, honestly curious about what your environment is like, and how power going out for just a few nights, not even weeks, would mean any kind of permanent damage? Direct lightning strikes? That at least is fairly uncommon here, usually double grounding is plenty sufficient though I tend to break exterior links to core with fiber these days just in case.
>As far as my last point - i said i don't see the point since starlink is CGNAT. It's like putting a fiber cable to a cellphone for tethering. you get no benefit.
What? There's tons of benefit! First, Starlink, and failover cellular dedicated modems, PtMP/PtP etc, frequently would be ideally installed hundreds or more feet away from wherever the core rack and router are, which in turn are away from fanned out further switches/WAPs. And in high positions more prone to surge. Fiber breaks that link in and has effectively no range limits, while being cheap and easy to work with. What would CGNAT have to do with the tremendous limitations of their router? Say you've got a small LAN covering even just a million square feet, Starlink router won't cover that itself. What about nested traffic shaping of various kinds, supporting your 802.1x auth, or arbitrary VLANs (direct or stacked) or a million other things? How about automatic failover with another connection, or shaping, or automated VPN tunneling of given subnets? Which is indeed a trivial way to give even a residential CGNAT IPv4 a static IP, just get a basic VPS somewhere geographically appropriate for access patterns, setup Wireguard on it and then tunnel both ways. Totally transparent to clients, can use PPSK/MPSK so that even the most outdated pure wireless stuff with zero radius support can simply be given a password and it's on the right place.
And second fwiw, Starlink Business does offer a publicly routable IPv4 (though not static), and each Starlink Terminal gets a /56 IPv6. And I did mention that terminal specifically didn't I?
I mean, how is Starlink here any different then nearly any other classic ISP in existence going back decades? It's not like there's anything wrong with being happy with an ISP supplied default AIO combo deal, most of the population is. But it's also always been the case that sometimes there is a need for more. And for businesses that's particularly true. Starlink v1 supported that elegantly and near perfectly. Starlink v2 took that away, even for 5x more expensive business terminal. I'm just saying that's unfortunate, clearly unnecessary, and I hope they change course with a v3 or v4 or the like someday. Even if only on the high end one.
I wonder if the antenna is expensive or if it's the other electronics in the terminal? I assume it uses a phased array. I also assume most phased arrays are pretty cheap.
The antenna or the array antenna should be pretty cheap because they're technically passive devices. The expensive parts, however, are active circuitry namely the down conversion circuitry and signal processing for beam steering. The trick is to reliably down convert the signal to low frequency baseband by maintaining its amplitude and phase in which information is encoded since frequency is ephemeral (can be changed and manipulated without losing the information). According to ChatGPT-4, the user bands for Starlink satellites are mainly using Ku-band for downlink between 10.7 to 12.7 GHz (satellite to user) and 14 to 14.5 GHz for uplink (user to satellite).
They are 3 ways of doing the frequency conversions namely superheterodyne (Single-IF), Direct conversion (Zero-IF) and the latest Direct RF sampling. The complexity is decreasing, hence the price is decreasing as we go from former to latter. Due to the relatively high frequency of Ku-band, Starlink antenna systems probably also utilize multiple down conversions or multi superheterodyne which make the price to be rather expensive due to the increased complexity of the circuitry and signal processing.
Now due to the rapid improvement in RF SoC/chip design by company like ADI and TI, we now have RF transceiver chip that can perform Direct RF sampling utilizing high speed ADC/DAC operating in Ku-band. Thus we can expect the price to be going down even further. I suspect the decreased in manufacturing for the antenna price is not so much due to the their better manufacturing iterations, increase yield, etc but mainly due to the recent availability of this high speed Direct RF sampling transceivers but I can be wrong.
Do note, that any "satellite trains" you see are from freshly launched satellites. The "trains" are from still relatively densely clumped satellites that are still separating from each other. After they separate, they re-orient which makes them almost invisible from the ground. You need long camera exposures or a telescope to see them at this point. They then start gradually raising their orbits to a parking orbit. After they reach a parking orbit they begin to drift the plane of their orbit so as to separate a single launch's satellites into multiple different planes, after which they then raise their orbits again into the operational orbit.
.jpg){kind=link}
