Author Topic: SpaceX FCC filing for a 4425 satellite constellation providing Internet service  (Read 62714 times)

Offline feynmanrules

  • Member
  • Posts: 79
  • florida
  • Liked: 38
  • Likes Given: 71
Quote
[...]

So, for a 1600 sat, 32 plane initial constellation, you're looking at around 64 ASDS F9 launches, or 96 RTLS launches. Or maybe 32 FH launches at one per plane?

Staggering numbers for a vehicle that has yet to hit 30 launches, and that's just the initial setup.

If sats are built as inexpensively as rockets and they launch on reuseables, you can see why they're taking this approach and the end-user cost advantages they would/should have relative to most of other competitors.
« Last Edit: 11/16/2016 11:00 PM by feynmanrules »

Offline AncientU

  • Senior Member
  • *****
  • Posts: 4552
  • Liked: 2678
  • Likes Given: 3775
1,708 tonnes of satellite
>4 ISS

Launch starting in 2019?
Quote
...those satellites launched in 2019...
p55, last paragraph
« Last Edit: 11/17/2016 12:33 AM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline AncientU

  • Senior Member
  • *****
  • Posts: 4552
  • Liked: 2678
  • Likes Given: 3775
Quote
[...]

So, for a 1600 sat, 32 plane initial constellation, you're looking at around 64 ASDS F9 launches, or 96 RTLS launches. Or maybe 32 FH launches at one per plane?

Staggering numbers for a vehicle that has yet to hit 30 launches, and that's just the initial setup.

If sats are built as inexpensively as rockets and they launch on reuseables, you can see why they're taking this approach and the end-user cost advantages they would/should have relative to most of other competitors.

Also shows why they need to get launch rates up/turn-around times minimized.

Initial 800 sats before service initiation... 32 launches (1 year?) probably one launch to each plane.
Another 32 launches to complete initial deployment (end of year 2?).

With only 5-7 year lifetime, need to have this pace if you want to get reasonable service from first 800 sats.

Final deployment takes another 3-4 years, at which point you need to start replacing the initial 800...

So, fully deployed by 2025, assuming start in 2019.
32 launches of 25 sats per year -- indefinitely.
« Last Edit: 11/17/2016 12:51 AM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline M.E.T.

  • Full Member
  • ****
  • Posts: 401
  • Liked: 150
  • Likes Given: 12
Quote
[...]

So, for a 1600 sat, 32 plane initial constellation, you're looking at around 64 ASDS F9 launches, or 96 RTLS launches. Or maybe 32 FH launches at one per plane?

Staggering numbers for a vehicle that has yet to hit 30 launches, and that's just the initial setup.

If sats are built as inexpensively as rockets and they launch on reuseables, you can see why they're taking this approach and the end-user cost advantages they would/should have relative to most of other competitors.

Also shows why they need to get launch rates up/turn-around times minimized.

Initial 800 sats before service initiation... 32 launches (1 year?) probably one launch to each plane.
Another 32 launches to complete initial deployment (end of year 2?).

With only 5-7 year lifetime, need to have this pace if you want to get reasonable service from first 800 sats.

Final deployment takes another 3-4 years, at which point you need to start replacing the initial 800...

So, fully deployed by 2025, assuming start in 2019.
32 launches of 25 sats per year -- indefinitely.

Is it possible to do a rough profit model for this? With reasonable reusability we can estimate the launch costs for 32 launches per year. What is the potential revenue from such a fully installed satellite system? Does it dwarf the revenue from their existing business model?
« Last Edit: 11/17/2016 06:42 AM by M.E.T. »

Offline Welsh Dragon

  • Full Member
  • *
  • Posts: 196
  • Liked: 187
  • Likes Given: 25
And there's the dispenser mass to consider.
Surely they could try to design the satellites to not need a dispenser? Like the Boeing electric comm sat double stacks, or Orbcomm stacks, or the MMS sats on Atlas V? Might be a bit more tricky if you end up with more than one stack under your fairing though.

Offline AncientU

  • Senior Member
  • *****
  • Posts: 4552
  • Liked: 2678
  • Likes Given: 3775
And there's the dispenser mass to consider.
Surely they could try to design the satellites to not need a dispenser? Like the Boeing electric comm sat double stacks, or Orbcomm stacks, or the MMS sats on Atlas V? Might be a bit more tricky if you end up with more than one stack under your fairing though.

If ever there was a driver for a reusable second stage, this is it.  Even a tanker second stage would cost less than M1D vac, tankage, dispenser, fairing every ten days.  Simply checking out a new stage/dispenser every time will be costly.  (Building three sats per work day also isn't cheap, so maybe the dispenser and stage is a marginal cost.)
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline AncientU

  • Senior Member
  • *****
  • Posts: 4552
  • Liked: 2678
  • Likes Given: 3775
Reposting this great data dump:

Also twitter from @malderi

"Musk: Have to pay attention to security. Bad if hacked, either by AI or people."
    - very important IMO, this would get round state censorship, so attack by states possible.

"Musk: biggest concern about success: important to assume that competitors get better, too. Teledesic didn't."
 - it will be interesting to see how they build a system that can be expanded to more bandwidth and increased capabilities. This is one of the biggest system design problems IMO, get it wrong and the system is doomed in the long term.

"Musk: Hopeful that we can structure this to work with different countries. Don't want China shooting out our satellites."
 - setting up a global sales network will take time, need financing and other options that are sensitive to local circumstances.

"Musk: Plan on using Hall thrusters. Easy to make, no real production difficulty. Doesn't make sense to outsource."
- good choice.

"Musk: Cheaper to have a bunch of PCs on racks than a few mainframes, this is the same idea."
- distributed aspect more similar to PCs on desks

"Musk: Don't see bandwidth as difficult issue. Space to ground has plenty of usable spectrum."
- doesn't explain how he expects to get permission to use that spectrum

"Musk: Going to cost a lot to build. Ten or fifteen billion dollars, or more. But revenues fund city on Mars."
- SpaceX does not have the resources to fund this internally
- Such a high investment, with significant chance of failure risks taking down SpaceX as well.

"Musk: Smaller satellites, few hundred kg, but capability of much larger satellites."
- my guess is optimum size is somewhere in 300-600 kg range, driven mainly by antenna size.

"Musk: (timeline?) In the past, I've been optimistic on schedule. Recalibrating. Envision version 1 in about 5 years."
- my translation - I'm aiming for 3 years, but I've been optimistic

- planning on multiple versions from the start

"Musk: Talking about around 4000 satellites. 4025 exactly in current design."
- that is probably 100s of launches, maybe 50/year.

"Musk: Talking mostly around 1100km level. Space debris not much of a problem there."
- with that number of satellites, they will have to take great care to dispose of them at end of life.

"Musk: 20-30ms latency everywhere on earth. Expand tech to Mars, not much fiber there yet."
- a unique selling point, lower latency for long distances than earth bound networks

"Musk: Developing world, but also options for people stuck with comcast."
- big problem in developing world is that it needs to be very cheap both in cost of equipment and charges. Larger antennae on satellites mean smaller cheaper ground stations.
- significant numbers in developed world have poor internet speeds and few prospects of significant improvement, not enough of a market on its own but still maybe $1B/year.

"Musk: More than half of global long haul comm over this system."
- this is a biggie, global long haul market is massive. Competes directly with GEO telecom sats.

"Musk: Long term, rebuilding Internet in space. 10% of local business traffic."
- another biggie, business traffic is concentrated in a far smaller number of customers than consumer, but generally needs higher quality of service and grade of service guarantees.


"Musk: Start by developing our own constellation. Comm, earth science, space science. Focus is global comm system."
- looks like long term they plan for more than just this initial 4025 satellite global internet system.
Bolding is mine

Lots of features from two years ago bang on.
« Last Edit: 11/17/2016 11:37 AM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Online meekGee

  • Senior Member
  • *****
  • Posts: 7500
  • N. California
  • Liked: 3829
  • Likes Given: 804
I've said it before - they must have a dedicated reusable second stage / dispenser for this
ABCD - Always Be Counting Down

Offline baldusi

  • Senior Member
  • *****
  • Posts: 7436
  • Buenos Aires, Argentina
  • Liked: 1433
  • Likes Given: 4468
There are 52 satellites in a plane (2 spares), so just over 20 tonnes, which is about what F9 can place in that orbit, but possibly not a F9 with ASDS landing.

Checking the NASA NLS II performance page, F9 FT + ASDS landing can put 11,805 kg into a 1150km 51.6 degree orbit. RTLS knocks that down to 8535 kg. That's different than the specified 53 degree inclination for the plane, so performance would likely be lower than that. And there's the dispenser mass to consider.

At 386kg per sat, two ASDS launches of 26 sats would be 10036kg leaving 1769kg worth of performance for the dispenser mass and 51.6 -> 53 degree inclination change.

Or maybe three RTLS launches of ~17 sats for 6562kg, leaving 1973kg worth of performance for the dispenser and inclination change.

So, for a 1600 sat, 32 plane initial constellation, you're looking at around 64 ASDS F9 launches, or 96 RTLS launches. Or maybe 32 FH launches at one per plane?

Staggering numbers for a vehicle that has yet to hit 30 launches, and that's just the initial setup.
The current F9 fairing has a 4.6m internal diameter, that starts to taper off at 6.7m with a 70deg angle.
If the satellites are 4m x 1.8m x 1.2m at launch, they are lucky if they can put four in a ring. At 4m long, they can't put two rings in the current fairing. Since the solar panels are 6m x 2m each, I would tend to think that the first dimensions are main body. And if they have some slope to enable two rings, this would only allow for 8 satellites within the current fairing.
This would, imply that they will launch no more than 8 satellites per launch,  or that they are planning for a longer fairing.

Offline wannamoonbase

  • Elite Veteran
  • Senior Member
  • *****
  • Posts: 2644
  • Denver, CO
    • U.S. Metric Association
  • Liked: 513
  • Likes Given: 880
The FH is going to be very busy, which is a good thing.

It would be a great way to push the envelop on fast turn around on reuseable launch vehicles.

Most if not all the launches would be VAFB correct?

Fully reuseable FH US  with a Raptor would be nice but adds development costs for a company already doing a lot of things.

If SpaceX can pull off this constellation it will be one of the biggest engineering accomplishments in history.

Excited to be finally into the first Falcon Heavy flow, we are getting so close!

Online envy887

  • Senior Member
  • *****
  • Posts: 2324
  • Liked: 1008
  • Likes Given: 676
The FH is going to be very busy, which is a good thing.
Since they appear to be volume and not mass-limited, and don't seem to be developing a larger fairing for FH, it doesn't look like any of these will launch on FH.

Quote
Most if not all the launches would be VAFB correct?
3200 out of 4425 birds are going to orbits they can hit with F9 RTLS from CCAFS or KSC (payload 8,535 kg to 51.6 deg 1150 circ LEO). Even with a dogleg to 53.0 or 53.8 degrees, that's likely more mass than they can fit in the fairing.


Offline Sam Ho

  • Full Member
  • ****
  • Posts: 463
  • Liked: 163
  • Likes Given: 1
Most if not all the launches would be VAFB correct?
3200 out of 4425 birds are going to orbits they can hit with F9 RTLS from CCAFS or KSC (payload 8,535 kg to 51.6 deg 1150 circ LEO). Even with a dogleg to 53.0 or 53.8 degrees, that's likely more mass than they can fit in the fairing.
CCAFS and KSC can reach as high as 57° without a dogleg.  VAFB, on the other hand, can only get down to 65° without a dogleg and requires a dogleg between 51° and 65°.

Offline old_sellsword

  • Full Member
  • ****
  • Posts: 517
  • Liked: 419
  • Likes Given: 374
The current F9 fairing has a 4.6m internal diameter, that starts to taper off at 6.7m with a 70deg angle.
If the satellites are 4m x 1.8m x 1.2m at launch, they are lucky if they can put four in a ring. At 4m long, they can't put two rings in the current fairing. Since the solar panels are 6m x 2m each, I would tend to think that the first dimensions are main body.

That 4 x 1.8 x 1.2 m set of dimensions is for calculating orbital decay characteristics. Since this is a communications satellite, it will have antennas that fold out, which are included in the "Satellite Body" dimensions for the orbital decay model. So the actual satellite bus in a stowed away launch configuration should be significantly smaller than 4 m in that one dimension, bringing the launch dimensions a lot closer to 1.5 x 1.8 x 1.2 m or something like that (we aren't given the completely folded up dimensions).

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 2921
  • Florida
  • Liked: 1246
  • Likes Given: 85
One of the key items on the business case is the power available on the sat. With 12m^2 of solar arrays that would be ~3.6kw. This would then be enough power to have 5 equal bandwidth links Ground-Earth,Forward, Aft, Port, and Starboard. Forward and Aft is the in-ring sat-t-sat. Port and Starboard is the ring-to-ring sat-to-sat. This is enough power to have a 500-1000GB Earth-Ground bandwidth per sat.
http://licensing.fcc.gov/myibfs/download.do?attachment_key=1158346
Quote
FREQUENCY BANDS REQUESTED FOR SPACEX SYSTEM
10.7-12.7 GHz
13.85-14.5 GHz
17.8-18.6 GHz
18.8-19.3 GHz
27.5-29.1 GHz
29.5-30.0 GHz
This is at 50Mhz per 1Gb spot/transponder [256QAM] * (2 transponders per frequency per spot [V/H]) / 2 (up down freq seperation) gives a possible 120 1Gb transponders per spot. If use frequency seperation between adjacent spots on same sat then it is 1/4 that at 30Gb per spot with a capability of:
 -  16 spots = .5Tb/sat
 -  64 spots = 2Tb/sat

Now at a cost per sat of $2M and cost of deployment of 8 sats per launch on a F9 at SpaceX cost of $32M per launch =$4M/sat for a cost of $6M/sat that is a bandwidth cost per sat of $3-12/Mb over the life of the sat.

This constellation will generate tremendous profit.

At any location on Earth with the ability to see as many as 128 sats and each sat is illuminating with 30Gb of bandwidth that is a total of ~3.9Tb of bandwidth. A single long haul station (Internet backbone) that uses 2Tb of that would generate $2.4M/yr($1200/Gbps in a year) in revenue.

For the initial 800 sat constellation cost of the establishment is $4.8B. Over the life (5yrs) you would need 400 long haul links to break even. Just US location to US location alone the system would be capable of a few thousand Internet backbone links. The system will generate many-many times in revenue to its initial costs just for the low cost Internet backbone rates.
« Last Edit: 11/17/2016 03:23 PM by oldAtlas_Eguy »

Offline gongora

  • Global Moderator
  • Full Member
  • *****
  • Posts: 1641
  • US
  • Liked: 1082
  • Likes Given: 862
Page 4 of the technical attachment has the breakdown of the intended use for each set of frequencies.  Sat-to-sat is optical.  SpaceX would need to get waivers to use some of their intended frequencies.

Offline VIY

  • Member
  • Posts: 22
  • Liked: 13
  • Likes Given: 13


This is at 50Mhz per 1Gb spot/transponder [256QAM] * (2 transponders per frequency per spot [V/H]) / 2 (up down freq seperation) gives a possible 120 1Gb transponders per spot. If use frequency seperation between adjacent spots on same sat then it is 1/4 that at 30Gb per spot with a capability of:

I am a  layman here, but how can one fit 1Gb in a 50MHz bandwidth? I would think 50MHz allows for maximum of 50 Mb? Thanks.   

Online Robotbeat

  • Senior Member
  • *****
  • Posts: 26622
  • Minnesota
  • Liked: 6488
  • Likes Given: 4704
Quote
[...]

So, for a 1600 sat, 32 plane initial constellation, you're looking at around 64 ASDS F9 launches, or 96 RTLS launches. Or maybe 32 FH launches at one per plane?

Staggering numbers for a vehicle that has yet to hit 30 launches, and that's just the initial setup.

If sats are built as inexpensively as rockets and they launch on reuseables, you can see why they're taking this approach and the end-user cost advantages they would/should have relative to most of other competitors.

Also shows why they need to get launch rates up/turn-around times minimized.

Initial 800 sats before service initiation... 32 launches (1 year?) probably one launch to each plane.
Another 32 launches to complete initial deployment (end of year 2?).

With only 5-7 year lifetime, need to have this pace if you want to get reasonable service from first 800 sats.

Final deployment takes another 3-4 years, at which point you need to start replacing the initial 800...

So, fully deployed by 2025, assuming start in 2019.
32 launches of 25 sats per year -- indefinitely.

Is it possible to do a rough profit model for this? With reasonable reusability we can estimate the launch costs for 32 launches per year. What is the potential revenue from such a fully installed satellite system? Does it dwarf the revenue from their existing business model?
Sure. A LOT. Yes. It would make SpaceX the largest satellite services company in the world by a large margin. Tens of billions of dollars of revenue per year. Eventually (in a few decades), potentially hundreds of billions.
« Last Edit: 11/17/2016 03:49 PM by Robotbeat »
Chris  Whoever loves correction loves knowledge, but he who hates reproof is stupid.

To the maximum extent practicable, the Federal Government shall plan missions to accommodate the space transportation services capabilities of United States commercial providers. US law http://goo.gl/YZYNt0

Offline JamesH65

  • Full Member
  • ****
  • Posts: 576
  • Liked: 327
  • Likes Given: 7
Fairing reuse would save an enormous amount of money here.....explains why they have been looking at it!

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 2921
  • Florida
  • Liked: 1246
  • Likes Given: 85
Page 4 of the technical attachment has the breakdown of the intended use for each set of frequencies.  Sat-to-sat is optical.  SpaceX would need to get waivers to use some of their intended frequencies.
Thanks.

So the numbers I am showing is the probable up end of capabilities.

The other items on revenues is that in the US alone there could be with 800 sats up to 360 backbone link locations (spot size of 100X100miles) with only ~25 sats visible with the smaller quantity the upper bound per link location is .7Tbps or a total revenue from just US Internet backbone of ~$300M but that 10% of direct to user link of 1Gbps shared channel with 200 customers sharing a channel and only for the US alone results in an upper bound for at $20/month charges a total of >$3B. That is supporting 14+Million customers just in US.

Even much less capable system/sat would still have a tremendous ROI.

Yearly revenue even for the 800 sat constellation will reach into the $Bs.

This then puts profits /yr into the $Bs.

Once the full constellation is deployed also expect rates to lower making terrestrial systems both backbone and direct to customer be more expensive!

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 2921
  • Florida
  • Liked: 1246
  • Likes Given: 85


This is at 50Mhz per 1Gb spot/transponder [256QAM] * (2 transponders per frequency per spot [V/H]) / 2 (up down freq seperation) gives a possible 120 1Gb transponders per spot. If use frequency seperation between adjacent spots on same sat then it is 1/4 that at 30Gb per spot with a capability of:

I am a  layman here, but how can one fit 1Gb in a 50MHz bandwidth? I would think 50MHz allows for maximum of 50 Mb? Thanks.
Oops! I am off by a factor of 2.

It should be 100Mhz per 1Gbps.
256QAM transmits 8bits/Hz (quite common). 1024QAM is 10bits/Hz. So 100Mhz would be 800-1000Mbps raw but with error correction bits (10-12bits for each 8 data bits) plus  a compression factor you get the 1000Mbps(1Gbps).

So max capability per spot area of full constellation would be 1Tbps (100X100mile sized spot) data rate density of 1Gbps/10square miles.

So divide revenue figures in half. But they still are in the $Bs per year even for the 800 sat constellation. (earlier post >$4B now >$2B).

Tags: