Launching Starlink with Starship

[su27k]

I suspect that there will be a major redesign of Starlink sats to take advantage of Starship. I wouldn't be surprised for them to grow to 1000-2000kg and hundreds of GBps each.

SpaceX could also sell the bus and launch as a package deal too. No Geo-sat manufacturer would be able to compete with a bus + launch to GEO for a few million.

Incredibly unlikely that Starlink satellites will ever weigh more than a ton, not unless SpaceX can still affordably mass-produce thousands per year. For a constellation to exist below ~550 km while still maintaining uninterrupted coverage, you NEED ~1500 satellites. ~500 kg is conceivable. 1-2 tons is far less so while maintaining full demisability and massive economies of scale and keeping the per-satellite cost reasonable.

For example, if SpaceX can double satellite mass and, say, 4x or 6x throughput, it could accomplish with 7,000-10,000 satellites what it had planned for >40k and maintain production on the order of 1000 per year, which would sustain significant manufacturing efficiencies.

If we're talking about 2-ton satellites, SpaceX would presumably end up raising throughput 10-20x, reducing the number of satellites needed to ~2000-4000 and thus lowering the required annual output to maybe 200-300 satellites. At that point, Starlink would be more like boutique hypercar production, whereas 1000-2000 per year is closer to 'mass' production at Ferrari or McLaren. And it's worth remembering that SpaceX's proposed 40,000-sat constellation (with decent interlinking and 20 Gbps/sat) would be an 800 Tbps network likely capable of serving like half of the entire global market for fixed broadband.

Sorry, I may be misreading this, but it sounds like you're saying increasing Starlink satelite's weight would cause the network to have too much bandwidth? Is "too much bandwidth" even a thing? They need all the bandwidth they can get. The definition of "broadband" is not going to standstill, there's already talk about defining "broadband" as 1Gbps.

Also need to consider that Starlink may not be just for internet connection. Their trademark application included remote sensing as goods/services too, then there's hosted payload and other applications like direct to phone connection to consider as well.

[su27k]

Granted the cost have to be reasonable, which I think it would be.

gather all the old satellites in a new train near the orbit Starship is going to deploy the new satellites, this way Starship can pick them up one by one quickly.

Just saying something could be done isn't interesting.  Most anything can be done.  The question is whether it's practical.  And when I think of the list of things that must be accomplished, I can't see how it would have a reasonable cost. 

Starlinks just can't be picked up quickly.  That last bit is the thing that really raised my hackles.  Just think of how long it takes to dock Dragon.  That's the kind of speed things happen in space.  Yeah one can probably be picked up faster than Dragon but I don't think you are going to get it quick enough to be practical.  Could you get it to an hour?  IDK.  That would make a 60 hr loiter to pick up F9 batch reasonable (is that "quickly"?) but 200 hrs for SS load of double-sized sats.

If the garbage analogy was apt, then you'd have more of an argument.  But full demisability is a severe blow to it.  I suppose if a NEPA ruling goes the wrong way and there's really some legs that develop around the ridiculous concern about incinerated Starlinks polluting the atmosphere then it enters the realm of doing a cost-benefit analysis.  But I fail to see how the costs aren't huge.  But hey SpaceX can surprise you there.  The "benefit" side is really where I think practicality fails.  I just don't see much benefit. 

If you want to build bigger SL's, then do it in a way that retains demisability.  That should be infinitely easier than bringing them back.
If it's the NEPA incineration pollution thing, well that's a counterfactual that could only be evaluated once it raises its specter.  I personally think it's a joke but I think that about a lot of things these days.

Yes, I agree that pickup time is definitely a limiting factor, but I don't see this as a showstopper if they can turn around the vehicle quickly. Quick math: If they launch 60 satellites per launch, then they need 134 launches per year to maintain a 40,000 satellite constellation, that's about one launch every 65 hours, so it's just about right if you can pick up one satellite every hour then turn around the vehicle in an hour or so. What if it takes longer to pick up or turnaround need more time? No problem, since you can have multiple ships, using two ships would double the time you can spend on orbit, and you can add several more ships easily without breaking the bank.

As for full demisability, I'm not sure it would be so easy to achieve if you increase the weight of the satellite to 1 to 2 ton range.

[vaporcobra]

I suspect that there will be a major redesign of Starlink sats to take advantage of Starship. I wouldn't be surprised for them to grow to 1000-2000kg and hundreds of GBps each.

SpaceX could also sell the bus and launch as a package deal too. No Geo-sat manufacturer would be able to compete with a bus + launch to GEO for a few million.

Incredibly unlikely that Starlink satellites will ever weigh more than a ton, not unless SpaceX can still affordably mass-produce thousands per year. For a constellation to exist below ~550 km while still maintaining uninterrupted coverage, you NEED ~1500 satellites. ~500 kg is conceivable. 1-2 tons is far less so while maintaining full demisability and massive economies of scale and keeping the per-satellite cost reasonable.

For example, if SpaceX can double satellite mass and, say, 4x or 6x throughput, it could accomplish with 7,000-10,000 satellites what it had planned for >40k and maintain production on the order of 1000 per year, which would sustain significant manufacturing efficiencies.

If we're talking about 2-ton satellites, SpaceX would presumably end up raising throughput 10-20x, reducing the number of satellites needed to ~2000-4000 and thus lowering the required annual output to maybe 200-300 satellites. At that point, Starlink would be more like boutique hypercar production, whereas 1000-2000 per year is closer to 'mass' production at Ferrari or McLaren. And it's worth remembering that SpaceX's proposed 40,000-sat constellation (with decent interlinking and 20 Gbps/sat) would be an 800 Tbps network likely capable of serving like half of the entire global market for fixed broadband.

Sorry, I may be misreading this, but it sounds like you're saying increasing Starlink satelite's weight would cause the network to have too much bandwidth? Is "too much bandwidth" even a thing? They need all the bandwidth they can get. The definition of "broadband" is not going to standstill, there's already talk about defining "broadband" as 1Gbps.

Also need to consider that Starlink may not be just for internet connection. Their trademark application included remote sensing as goods/services too, then there's hosted payload and other applications like direct to phone connection to consider as well.

Absolutely not saying that there is such a thing as too much bandwidth. The more the merrier. But the application of unprecedented economies of scale is a keystone feature of Starlink and part of what distinguishes it and might make it viable in contrast to a graveyard of failed, bankrupted LEO internet efforts. And it's just a simple fact that SpaceX would need tens or even hundreds of millions of customers for that ~800 Tbps constellation to be worthwhile, so a need for massive growth in per-unit bandwidth would require some very specific conditions be met on top of immense commercial success.

[oldAtlas_Eguy]

Based on what they learn, I think they will scrap the plans for 40k constellation and keep a few thousands ( 4k ? ) much more capable ones and with longer planned life ( >10+ years ) at around 550km. Double the weight/fuel, etc.

1/I think they  will go for 40K and every 2 years replace them to  push technology of satellite.
2/End game is not just global internet to anybody even in populated area but I will be not surprise if mobile service will be available later.

In order for a ground UT to be able to do multiple Gbps data rates as in 10Gbps. It would take actually that UT on the ground connecting to multiple sats as in about 20+ of them simultaneously.

So in the long run more sats gets more than fewer larger ones for a overall lower cost due to higher economies of scale. Sats will get larger but not in the sizes of the hugh GEO sats. Or at least any time soon.

[joek]

In order for a ground UT to be able to do multiple Gbps data rates as in 10Gbps. It would take actually that UT on the ground connecting to multiple sats as in about 20+ of them simultaneously.
...

Do not expect that to be an issue for consumer end-user terminals. For multi-gig bandwidth, expect dedicated ground stations or similar which have broad-based capabilities well beyond an induvial UT, such as with Google or MSFT. Many more references; expect others.

p.s. And that is a direct threat to OneWeb's pivot to a more B2B orientation.

[Asteroza]

Based on what they learn, I think they will scrap the plans for 40k constellation and keep a few thousands ( 4k ? ) much more capable ones and with longer planned life ( >10+ years ) at around 550km. Double the weight/fuel, etc.

1/I think they  will go for 40K and every 2 years replace them to  push technology of satellite.
2/End game is not just global internet to anybody even in populated area but I will be not surprise if mobile service will be available later.

In order for a ground UT to be able to do multiple Gbps data rates as in 10Gbps. It would take actually that UT on the ground connecting to multiple sats as in about 20+ of them simultaneously.

So in the long run more sats gets more than fewer larger ones for a overall lower cost due to higher economies of scale. Sats will get larger but not in the sizes of the hugh GEO sats. Or at least any time soon.

Mass-wise sure, less than monster GEO birds, but in terms of antenna size, they might go larger if various things align. Might get super large if the economics and timing align such that on orbit assembly of antenna reflectors or perhaps just ganging multiple phased array panels becomes a thing. I remember old concepts that had tiled hex panels assembled into a partial dome structure, which allows for low to the horizon pointing of UT's.

There's pic of such a hex panel shallow dome sat on the last page of Dr. Hoyt's presentation for a FISO telecon, showing a nominal GlobalFi satellite

http://fiso.spiritastro.net/telecon16-18/Hoyt_12-13-17/

[oldAtlas_Eguy]

The primary thing I was trying to get at was that the item that most influences the total amount of bandwidth data available in a small area on Earth (100 sq km  a 10kmX10km) is the number of satellites. That is is because there is only so much frequency available in the 3 bands assigned to Starlink: Ku, Ka and V bands. It requires a physical point direction difference as in a different satellite to increase the total data capability in a small area. The first is to increase the number of sats. Then the next is to increase the number of spots on each sat to optimize between the size of the spots from a sat and the number of sats to maximize the total density of data bandwidth in a small area. With V1.0 sats and 4400 sats there is an approximate 20 sats that are illuminating any one area with channels of data. This is about 4 X 1Gbps just in Ku band from each sat X the 20 sats for a total data throughput capable in that small area of 80Gbps. At 0.1Gbps per active users that is 800 users all actively downloading 100Mbps of data simultaneously in a 10kmX10km area as a theoretical max. The problem is though with large field spots of as large or larger than 100kmX100km then that is not that mind blowing. Now add 100 X the number of spots on a single sat such that the spot sizes are shrunk to 10kmX10km. That is what I was trying to get at that it is not just the size of the sats but also the number of them that is so significant in how much user density and user data rates that can be supported. You increase both and the supportable user density and higher data rates become X the number of sats and 100X the number of spots per sat gives an increase in total system data bandwidth increase and capability of 800 which is a factor that can be more easily visualized as with the 4400 sat system V1.0 sats can support 800 simultaneous downloading users in a 100kmX100km area each downloading 100Mbps of data. But with the increased system that would become 640,000 simultaneous downloading users in a 100kmX100km area each downloading 100Mbps of data. If the limitation on number of subscribers in the first system is 10X oversubscription for that 100kmX100km area would be 8,000 subscribers vs 6.4 million subscribers possible in the same area in the upgraded system.

This is what Starship can do for Starlink. It can make the Starlink system be able to saturate into any market and compete 1 fro one on price and data speeds with any other service.

[libra]

. Quick math: If they launch 60 satellites per launch, then they need 134 launches per year to maintain a 40,000 satellite constellation, that's about one launch every 65 hours, so it's just about right if you can pick up one satellite every hour then turn around the vehicle in an hour or so.

Remarquable. 134 launches per year is roughly twice the annual numbers of launches, worlwide, in the 2000's / 2010's.

It's that kind of "nugget" of information that makes one realize how much a revolution SpaceX is bringing. Also how musk Mars plan is a finely tuned machine (if not steamroller).

I already said that in others posts and others threads, but - having grown in the 90's, a good case could be made that Musk created his own "killer app" to make "access to space" happens.

Starlink feeds BFR-Starship launch manifest, and also feeds the Mars program via a cash cow.

https://www.amazon.fr/Rocket-Company-Patrick-Stiennon/dp/1563476967

Time to read again that little gem of a fiction...

[Ludus]

Based on what they learn, I think they will scrap the plans for 40k constellation and keep a few thousands ( 4k ? ) much more capable ones and with longer planned life ( >10+ years ) at around 550km. Double the weight/fuel, etc.

1/I think they  will go for 40K and every 2 years replace them to  push technology of satellite.
2/End game is not just global internet to anybody even in populated area but I will be not surprise if mobile service will be available later.

In order for a ground UT to be able to do multiple Gbps data rates as in 10Gbps. It would take actually that UT on the ground connecting to multiple sats as in about 20+ of them simultaneously.

So in the long run more sats gets more than fewer larger ones for a overall lower cost due to higher economies of scale. Sats will get larger but not in the sizes of the hugh GEO sats. Or at least any time soon.

Starship cargo hold seems to fit 4 60 sat flat stacks pretty well. If they went to regular Starship launches on the same schedule as F9 Starlink they could launch 4X as many sats as F9 and they’d only weigh 62,400 kg. A 100 ton payload would still have 37,400 kg and more volume than an F9 faring left for other customers to ride share.

Starship though is intended to be a lot cheaper than F9 and more rapidly reusable. SpaceX could ramp up scheduled Starlink-Rideshare launches to twice a week rather than every two weeks. Then it could launch 4X as many times with 4X the sats.

It would seem a natural step to just link the Sats in the 4 stacks so they’re single 1040kg sat with 4X the power, antennas, etc - and merely launch 4X as many of the big ones, not 16X.

SpaceX doesn’t need to choose. Starship can launch tens of thousands of much larger, much more capable satellites. 1.0 is just a MVP.

[Asteroza]

Based on what they learn, I think they will scrap the plans for 40k constellation and keep a few thousands ( 4k ? ) much more capable ones and with longer planned life ( >10+ years ) at around 550km. Double the weight/fuel, etc.

1/I think they  will go for 40K and every 2 years replace them to  push technology of satellite.
2/End game is not just global internet to anybody even in populated area but I will be not surprise if mobile service will be available later.

In order for a ground UT to be able to do multiple Gbps data rates as in 10Gbps. It would take actually that UT on the ground connecting to multiple sats as in about 20+ of them simultaneously.

So in the long run more sats gets more than fewer larger ones for a overall lower cost due to higher economies of scale. Sats will get larger but not in the sizes of the hugh GEO sats. Or at least any time soon.

Starship cargo hold seems to fit 4 60 sat flat stacks pretty well. If they went to regular Starship launches on the same schedule as F9 Starlink they could launch 4X as many sats as F9 and they’d only weigh 62,400 kg. A 100 ton payload would still have 37,400 kg and more volume than an F9 faring left for other customers to ride share.

Starship though is intended to be a lot cheaper than F9 and more rapidly reusable. SpaceX could ramp up scheduled Starlink-Rideshare launches to twice a week rather than every two weeks. Then it could launch 4X as many times with 4X the sats.

It would seem a natural step to just link the Sats in the 4 stacks so they’re single 1040kg sat with 4X the power, antennas, etc - and merely launch 4X as many of the big ones, not 16X.

SpaceX doesn’t need to choose. Starship can launch tens of thousands of much larger, much more capable satellites. 1.0 is just a MVP.

Hrm, that made me think for a moment. What happens if starlinks dock corner to corner to create a larger aggregate sat? How penalized would you be by solar array shadowing?

[Zed_Noir]

Think there will be a new verison of the Starlink comsats for deployment from the Chomper variant of the Starship. SpaceX will replace the current rectangular flatpack with a hexagonal flatpack, IMO. It would make for more efficient use of the Starship payload volume.


Could fitted 420 Hexagonal flatpacks in 7 stacks of 60 with 6 stacks around the center stack. With each Hexagonal flatpack have a diameter of about 2 meters.

[oldAtlas_Eguy]

Think there will be a new verison of the Starlink comsats for deployment from the Chomper variant of the Starship. SpaceX will replace the current rectangular flatpack with a hexagonal flatpack, IMO. It would make for more efficient use of the Starship payload volume.


Could fitted 420 Hexagonal flatpacks in 7 stacks of 60 with 6 stacks around the center stack. With each Hexagonal flatpack have a diameter of about 2 meters.

Even at current mass / sat it would be too heavy. Think of a 30 to 50% growth in mass from it's current about 260kg to 320 to 400kg. So counts would be 245 to 315. For a set of 7 stacks -> 35 to 45 high * 7 stacks. The mass increase plus surface area increase point toward Earth would enable a 2X or 3X bandwidth throughput /sat increase.

[su27k]

Another thing to consider is that each plane only takes a small # of satellites, 20 to 58 in the 4,400 constellation. This means if you launch more satellites than this in a single launch, the rest of the satellites will need to drift to nearby planes, and this takes time, from weeks to months. This will cut into satellite's useful life time on orbit, and it slows down the deployment. This is why I'm skeptical that they'll launch hundreds of Starlink on a Starship. Seems to me the best way to take advantage of Starship is to launch much bigger and powerful satellites.

[Zed_Noir]

Another thing to consider is that each plane only takes a small # of satellites, 20 to 58 in the 4,400 constellation. This means if you launch more satellites than this in a single launch, the rest of the satellites will need to drift to nearby planes, and this takes time, from weeks to months. This will cut into satellite's useful life time on orbit, and it slows down the deployment. This is why I'm skeptical that they'll launch hundreds of Starlink on a Starship. Seems to me the best way to take advantage of Starship is to launch much bigger and powerful satellites.

Formation fly several small Starlink comsats to emulated a large Starlink comsat to increase the bandwidth available in a geographical cell area with the small Starlink comsats have more narrow coverage footprint. Will avoid the need to developed a bigger and more complex Starlink comsat.

[thirtyone]

Another thing to consider is that each plane only takes a small # of satellites, 20 to 58 in the 4,400 constellation. This means if you launch more satellites than this in a single launch, the rest of the satellites will need to drift to nearby planes, and this takes time, from weeks to months. This will cut into satellite's useful life time on orbit, and it slows down the deployment. This is why I'm skeptical that they'll launch hundreds of Starlink on a Starship. Seems to me the best way to take advantage of Starship is to launch much bigger and powerful satellites.

How much prop does changing planes quickly cost? There's got to be so much extra delta V available on an SS at low cost they could probably just get things where they want sooner

[Zed_Noir]

Think there will be a new verison of the Starlink comsats for deployment from the Chomper variant of the Starship. SpaceX will replace the current rectangular flatpack with a hexagonal flatpack, IMO. It would make for more efficient use of the Starship payload volume.


Could fitted 420 Hexagonal flatpacks in 7 stacks of 60 with 6 stacks around the center stack. With each Hexagonal flatpack have a diameter of about 2 meters.

Even at current mass / sat it would be too heavy. Think of a 30 to 50% growth in mass from it's current about 260kg to 320 to 400kg. So counts would be 245 to 315. For a set of 7 stacks -> 35 to 45 high * 7 stacks. The mass increase plus surface area increase point toward Earth would enable a 2X or 3X bandwidth throughput /sat increase.

Was thinking of a lighter comsat with target mass of 225 kg each. So the 7 stacks of 60 comsats will be under 100 tonnes. Rational for lighter comsat is cheaper unit cost and shorter production time while maintaining the unit capabilities of current comsats.

[STS-200]

Formation fly several small Starlink comsats to emulated a large Starlink comsat to increase the bandwidth available in a geographical cell area with the small Starlink comsats have more narrow coverage footprint. Will avoid the need to developed a bigger and more complex Starlink comsat.

It doesn't work like that. You can't have several satellites in the same position (or very close), as their uplink and downlink beams will overlap.
Each satellite has its own place in the sky, and the spacing between them cannot be reduced below a certain level because of the beamwidth of the ground station's antenna (looks like around 100km in the case of Ku-band Starlink).
If you try to put multiple satellites in the same place, they will all receive the same signal from the ground, which doesn't increase bandwidth. Vice versa for downlink.

To increase the data rate at a specific point on the ground, you need more bandwidth (a wider range of frequencies), or you need to signal different points in the sky. You can then re-use the same frequencies, but with the signals travelling in different directions so they don't interfere excessively.

Starlink is using both those techniques; planning to use E and V bands on later spacecraft (which also allows beamwidth to shrink), and increasing the number of satellites in various orbits, so each ground site can see many different spacecraft.

[STS-200]

I would think that future satellites for Starlink launch might also become more massive as they try to prolong their lifetimes - increases in redundancies, propellant and solar cell capacities (for EoL power reasons) might allow, say, a 10-year lifetime instead of 5.
Replacing many thousands of satellites every 10 years is likely to sound better for their investors than doing it every 5.

But back on topic; launching Starlink with Starship?
Yes, if Starship can do it more cheaply, which is their plan.

[tbellman]

I would think that future satellites for Starlink launch might also become more massive as they try to prolong their lifetimes - increases in redundancies, propellant and solar cell capacities (for EoL power reasons) might allow, say, a 10-year lifetime instead of 5.

Responding to this in the "Starlink : General Discussion" thread: https://forum.nasaspaceflight.com/index.php?topic=48297.msg2256225#msg2256225

[Asteroza]

Formation fly several small Starlink comsats to emulated a large Starlink comsat to increase the bandwidth available in a geographical cell area with the small Starlink comsats have more narrow coverage footprint. Will avoid the need to developed a bigger and more complex Starlink comsat.

It doesn't work like that. You can't have several satellites in the same position (or very close), as their uplink and downlink beams will overlap.
Each satellite has its own place in the sky, and the spacing between them cannot be reduced below a certain level because of the beamwidth of the ground station's antenna (looks like around 100km in the case of Ku-band Starlink).
If you try to put multiple satellites in the same place, they will all receive the same signal from the ground, which doesn't increase bandwidth. Vice versa for downlink.

To increase the data rate at a specific point on the ground, you need more bandwidth (a wider range of frequencies), or you need to signal different points in the sky. You can then re-use the same frequencies, but with the signals travelling in different directions so they don't interfere excessively.

Starlink is using both those techniques; planning to use E and V bands on later spacecraft (which also allows beamwidth to shrink), and increasing the number of satellites in various orbits, so each ground site can see many different spacecraft.

If I remember correctly, that's not entirely true though. In the old classic single antenna sense, yes that becomes a problem. Then sats moved to stuffing the equivalent of antenna cones like a beehive to create spot beam send and receive zones on the antenna. But Starlink is an all phased array setup with the sat managing the link, so no physical cones but logically separating the spot beams at the antenna. There are 6 antenna plates on a Starlink currently that are jointly managed. It does help that the plates are known distances and tightly aligned to be cooperative, so extending that to antennas on close formation neighbors is a bit of an issue but functionally the same. A phased array antenna is really made up of thousands of tiny antennas logically controlled, so adding some more is a software and timing problem (typically made easier when the antenna distances are physically known/fixed and directly managed on the same electrical bus).