Author Topic: SpaceX - now a satellite manufacturer (Starlink)  (Read 599921 times)

Offline JH

  • Full Member
  • **
  • Posts: 282
  • Liked: 142
  • Likes Given: 60
Re: SpaceX - now a satellite manufacturer
« Reply #2080 on: 06/10/2018 12:57 PM »
Was there any hints at a potential merger between the Starlink side of spaceX and/or an established sat manufacturer (SSL ?) or more probably a sat operator with an experience in managing a large fleet (e.g Iridium or Telesat) ?

Will they poach people from established companies? Obviously.

But from their perspective, I doubt any company has meaningful expertise in either constellation management or satellite manufacturing. Iridium has the largest constellation ever launched and it is only 98 satellites. On the manufacturing side, I would imagine that they probably view the institutional knowledge and equipment of existing concerns as being tailored towards the production of massively expensive, bespoke satellites, which is anathema to them.

Offline groundbound

  • Full Member
  • **
  • Posts: 203
  • Liked: 161
  • Likes Given: 8
Re: SpaceX - now a satellite manufacturer
« Reply #2081 on: 06/10/2018 07:59 PM »
Was there any hints at a potential merger between the Starlink side of spaceX and/or an established sat manufacturer (SSL ?) or more probably a sat operator with an experience in managing a large fleet (e.g Iridium or Telesat) ?

Will they poach people from established companies? Obviously.

But from their perspective, I doubt any company has meaningful expertise in either constellation management or satellite manufacturing. Iridium has the largest constellation ever launched and it is only 98 satellites. On the manufacturing side, I would imagine that they probably view the institutional knowledge and equipment of existing concerns as being tailored towards the production of massively expensive, bespoke satellites, which is anathema to them.

More than that, there is a future growth opportunity there for SpaceX.

Once the satellite production line is fully up to speed and running without strain, they can potentially start a spinoff that offers the bus to startup customers for unique projects. Anyone with a new idea on how to commercialize space gets a tremendous bootstrap by using a proven and inexpensive design. This also likely improves access to VC funding.

Initially that may only add a small amount to SpaceX revenue, but the eventual growth in launch customers for cheap SpaceX launch gets expedited.

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2082 on: 06/10/2018 08:14 PM »
So what is the likely max amount of uplink/downlink capacity of any given gateway station on Earth surface?

Each sat will be able to produce several frequency channels per spot: 8 to 16 channels X2 for left and right antenna circular polarization.

Each sat will be able to produce several spots: ~16.

Each channel is 1Gbps.

Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

That puts the U/D throughput per sat at ~256Gbps to 512Gbps. With each of the laser links added the total throughput of the sat (what it has to provide switching on) is ~1.3 to 2.6Tbps.

Now multiple sats illuminate that same location on the ground simultaneously: as many as 128 for an elevated (hilltop) location.

So the max backhaul U/D linkage from any given station could be as high as 2 to 4 Tbps (128 sats * 16 to 32 1Gbps channels per sat).

NOTE: Even though latter generation sats will have more spots of smaller size and a higher per sat throughput the max U/D throughput for any given backhaul station will not change. What would change is minimum distance between locations of backhaul stations. Such as a 200km seperation distance down to a 50km seperation distance.  Such that the backhaul capability for a 200x200km area would go from ~2Tbps to 32Tbps on a later generation sat system.

Another NOTE: Starlink Backhaul is not a true physical implementation. It can be reconfigured in seconds for transport direct from any station on Earth to any other station on Earth so long as the total throughput of the given station is not exceeded. Physical "buried" cable cannot do this and represents a gamble by the investors that  the path to be used for the clients will be the one that they are installing. Starlink just needs to collocate next to where the clients are and where they may be connecting to usually other clients so the financial risks for SpaceX/Starlink is low since even though the specific path may not come to be there will be some path used to some other location.

From a business case standpoint backhaul is a very reliable revenue stream. It will continue to increase in demand but that demand may shift in the point to point locations used. Starlink can accomodate the shift easily and without great expense, "cable" backhaul cannot.

Offline JH

  • Full Member
  • **
  • Posts: 282
  • Liked: 142
  • Likes Given: 60
Re: SpaceX - now a satellite manufacturer
« Reply #2083 on: 06/10/2018 09:21 PM »
So what is the likely max amount of uplink/downlink capacity of any given gateway station on Earth surface?

Each sat will be able to produce several frequency channels per spot: 8 to 16 channels X2 for left and right antenna circular polarization.

Each sat will be able to produce several spots: ~16.

Each channel is 1Gbps.

Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

That puts the U/D throughput per sat at ~256Gbps to 512Gbps.

Even ignoring the backhaul capacity, that is a beastly number.

(512 Gbps/satellite)*(4425 satellites)*(30 days) gives ~730 EB/month. Assuming that >90% of that capacity is wasted by time spent over the ocean, deserts, etc. it is still a large fraction of total global usage, which is estimated to be ~150 EB/month currently.

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2084 on: 06/10/2018 10:27 PM »
So what is the likely max amount of uplink/downlink capacity of any given gateway station on Earth surface?

Each sat will be able to produce several frequency channels per spot: 8 to 16 channels X2 for left and right antenna circular polarization.

Each sat will be able to produce several spots: ~16.

Each channel is 1Gbps.

Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

That puts the U/D throughput per sat at ~256Gbps to 512Gbps.

Even ignoring the backhaul capacity, that is a beastly number.

(512 Gbps/satellite)*(4425 satellites)*(30 days) gives ~730 EB/month. Assuming that >90% of that capacity is wasted by time spent over the ocean, deserts, etc. it is still a large fraction of total global usage, which is estimated to be ~150 EB/month currently.
But by the time those 4425 sats get deployed April 2027 (9 years from license date of April 2018). The total global usage could be greater that the max capability of Starlink 1st generation sat constellation design.

Now if you use a simple technique of having the sat create and extinguish spots so that when it creates a spot that spot does not move its Earth illuminated location even as the sat moves across the sky then latter sat generations which would create many more spots of much smaller size would still offer ground terminals exactly the same connection time per spot/sat as the 1st gen sats. This allows older ground terminals unmodified to work easily with newer sats.

If then a new sat generation is introduced every 3 years with increased throughput such that spot sizes are decreased by a factor of 2 and number of spots per sat increases by a factor of 4. In 4 generations this 4425 sat constellation (15 years from now) could have an max EB/month value of 16X the value you estimated. But interestingly enough Starlink's capability would only be keeping up with the global data demand and not even that far out front of it and maybe even following it. Hence a probable deployment implementation of the additional 8000 VLEO sats using the much unused higher frequency bands.

In 10 years all of our media will likely be delivered as on-demand and not broadcast. Where contracts for access is by either a specific content holder or by a general content provider reselling access from the content holders/originators. Existing smart TV's could fit into such a system without even requiring new hardware. Later versions of TV's will be more like smart phones than TV's only much larger screens.

The data world is evolving rapidly and the evolvement is even accelerating. Starlink is chasing not leading future. It is only that it's implementation is massive where others have been minor. Look out terrestrial infrastructure data carrier providers.

Offline Ludus

  • Full Member
  • ****
  • Posts: 1113
  • Liked: 505
  • Likes Given: 254
Re: SpaceX - now a satellite manufacturer
« Reply #2085 on: 06/11/2018 12:15 AM »
Are you proposing two kinds of Starlink satellites, one for optical and one for RF?

What we "know" so far indicates that each satellite would have both RF to ground and optical for sat-to-sat, and there would be only "one" type (perhaps specialized for LEO vs VLEO and both surely evolved over time, but otherwise "the same")

No, the difference would all be on the RF side. A sat specialized to link the orbital network to the ground internet would have different design priorities (a few very high capacity links) vs the common sat (many moderate capacity links). Both would have the same optical sat to sat.

Online gongora

  • Global Moderator
  • Senior Member
  • *****
  • Posts: 3666
  • US
  • Liked: 3013
  • Likes Given: 1801
Re: SpaceX - now a satellite manufacturer
« Reply #2086 on: 06/11/2018 01:50 AM »
The Starlink sats use Ku-band for user links and Ka-band for gateway links.  The Ku beam communicates with multiple users but a Ka beam only communicates with a single gateway at a time (and up to four satellites can communicate with a gateway).  For the Ku beams they use different polarities for uplink and downlink, for Ka downlink beams they can use both polarities at the same frequency.  This is all in the documentation.
« Last Edit: 06/11/2018 01:50 AM by gongora »

Offline Asteroza

  • Full Member
  • ****
  • Posts: 626
  • Liked: 81
  • Likes Given: 2
Re: SpaceX - now a satellite manufacturer
« Reply #2087 on: 06/11/2018 07:24 AM »
Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

Of note, freespace optics lasercomm is not much higher than 2-3Gbps for near COTS stuff right now, so unless you are doing some impressive multiplexing you are not going to see 100Gbps+ unless somebody has been hiding some innovations...

Offline Bananas_on_Mars

  • Full Member
  • *
  • Posts: 142
  • Liked: 83
  • Likes Given: 14
Re: SpaceX - now a satellite manufacturer
« Reply #2088 on: 06/11/2018 08:20 AM »


Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

Of note, freespace optics lasercomm is not much higher than 2-3Gbps for near COTS stuff right now, so unless you are doing some impressive multiplexing you are not going to see 100Gbps+ unless somebody has been hiding some innovations...

Those 2-3 Gbps are for a single wavelength, right? So you mean multiplexing by using lasers with different wavelengths for the same beam?

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2089 on: 06/11/2018 07:56 PM »
The Starlink sats use Ku-band for user links and Ka-band for gateway links.  The Ku beam communicates with multiple users but a Ka beam only communicates with a single gateway at a time (and up to four satellites can communicate with a gateway).  For the Ku beams they use different polarities for uplink and downlink, for Ka downlink beams they can use both polarities at the same frequency.  This is all in the documentation.
Thanks.

Do you have a modulation specification 16QAM, 256QAM, 1024QAM? Or something else? The modulation method will determine how many 1Gbps channels can be created per the ~1.1 GHz bandwidth available for each Uplink/Downlink at the Ka band for the use by the gateways. Next is whether the Ka band will have multiple spots or just a single spot per sat. Multiple spots means that the max data rate/sat-gateway link U/D is 1/4 that of a single spot. But that the spacing between gateway locations would have to be large in the >200km distances for a single spot.

So the gateway throughput bandwidth equation looks something like this:
 max data throughput Gbps for each gateway = 4(number of sats linked) * 2 (left right polarization) * Ch(number  of 1Gbps channels for the allocated 1.1GHz frequency bandwidth) / (1 [single spot] or 4 [multiple spots])

EXAMPLES:
16QAM gives 16GBps for per Gateway throughput (single spot and 50% overhead for FEC at transmission layer [1Gbps data uses 1.5Gbps raw bit transmission])

256QAM gives ~40Gbps for per Gateway throughput (single spot and 50% overhead for FEC at transmission layer [1Gbps data uses 1.5Gbps raw bit transmission])

1024QAM gives ~56Gbps for per Gateway throughput (single spots and 50% overhead for FEC at transmission layer [1Gbps data uses 1.5Gbps raw bit transmission])

256QAM is a mature level.

1024QAM is just entered usage.

Likely modulation method will be more like a 256QAM than a higher bit to Hz.

In order to up the per gateway throughput they will need to connect simultaneously to more sats than just 4.

Offline AncientU

  • Senior Member
  • *****
  • Posts: 6217
  • Liked: 4009
  • Likes Given: 5547
Re: SpaceX - now a satellite manufacturer
« Reply #2090 on: 06/15/2018 05:17 PM »
Excellent new article by vaporcobra:
Quote
SpaceX’s ultimate ace in the hole is its Starlink satellite internet business
Quote
Just by sheer numbers alone, stepping from launch vehicle and spacecraft production and operations into the satellite manufacturing, services, and connectivity industries is a no-brainer. Bluntly speaking, the market for rocket launches makes up barely more than one-sixtieth – less than 2% – of the entire commercial satellite industry, while services (telecommunications, Earth observation, science, etc.) and equipment (user terminals, GPS receivers, antennae, etc) account for more than 93%. Even the satellite manufacturing industry taken on its own is more than three times as large as the launch industry – $15.5b versus $4.6b in 2017.
https://www.teslarati.com/spacex-starlink-satellite-market-value-launch-business/
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2091 on: 06/15/2018 09:16 PM »
Excellent new article by vaporcobra:
Quote
SpaceX’s ultimate ace in the hole is its Starlink satellite internet business
Quote
Just by sheer numbers alone, stepping from launch vehicle and spacecraft production and operations into the satellite manufacturing, services, and connectivity industries is a no-brainer. Bluntly speaking, the market for rocket launches makes up barely more than one-sixtieth – less than 2% – of the entire commercial satellite industry, while services (telecommunications, Earth observation, science, etc.) and equipment (user terminals, GPS receivers, antennae, etc) account for more than 93%. Even the satellite manufacturing industry taken on its own is more than three times as large as the launch industry – $15.5b versus $4.6b in 2017.
https://www.teslarati.com/spacex-starlink-satellite-market-value-launch-business/
Yes.

The satellite and launch business industries are tightly tied such that an increase of 20% in satellite industry would have a near equal 20% increase in the launch industry. The LEO comm constellations are set to create an 20%+ increase in the launch rates. So this should also increase the satellite industry side by the same expansion. But the these new sats are mostly not built by the same existing sat manufactures but new kids on the block. So the ratio of 3 to 1 sat to launch may not apply to this growth.

But the ground terminal market is much larger. 10,000,000 ground terminals at ~$1,000 per terminal (a 2 year contract pays off the terminal cost at a charge to customer of $50 per month) is a $10B market for the LEO constellation. If you make 10% profit on the sale of this hardware then that is a potential profit for ground terminals alone of $1B for every 10 million terminals. The cost of ground terminals is very similar to that of a high end smart phone. With contract structures like that of smart phone services contracts the customer does not buy the terminal initially outright but on a payment plan. This makes the monthly high speed internet access through a LEO constellation sat network at a price of $100 to $150 /month. That is a yearly revenue for every 10 million subscribers of from $1.2B to $1.8B.

Starlinks backhaul at 50% of 1% of the complete network capability over land area results in a bit rate of ~680Tbps transport capability. At $1500/month/10 Gbps the revenue is for 1 year is ~$1.2B.

This puts a fully operational Starlink constellation yearly revenue at ~ $2.5B/year. A 800 sat initial constellation would have ~ a revenue of $500M/year. At initial 50% deployment at April 2024 that would be revenue of ~$1.2B per year.

If every 3 years the next gen sats double the troughput per sat (generally an increase in number of spots by factor of 2), then by 10 years the revenue will have grown to 8X or ~$20B/year. Also that puts the backhaul capability at 4,800Tbps if at a reduced price of $100/Gbps/month that is yearly revenue just from backhaul of $5.7B/year.

For ROI (profit vs investment) over 10 years of operation results in $10B in profit at a profit margin of just 10% over costs at a initial investment for the constellation of $10B is a 100% ROI or doubling of the investment in 10 years. Investors look at these numbers and the risk levels and end up being eager to invest when well established players are well advanced in the production and capability demonstrations such as OneWeb and SpaceX's Starlink.

Offline AncientU

  • Senior Member
  • *****
  • Posts: 6217
  • Liked: 4009
  • Likes Given: 5547
Re: SpaceX - now a satellite manufacturer
« Reply #2092 on: 06/15/2018 10:35 PM »
One comment on:
Quote
The satellite and launch business industries are tightly tied such that an increase of 20% in satellite industry would have a near equal 20% increase in the launch industry. The LEO comm constellations are set to create an 20%+ increase in the launch rates. So this should also increase the satellite industry side by the same expansion. But the these new sats are mostly not built by the same existing sat manufactures but new kids on the block. So the ratio of 3 to 1 sat to launch may not apply to this growth.

We need to add launch and satellite manufacturing revenues into the international pool, even if companies like Spacex do those things in house... no different than Boeing or LM or Orbital (NGIS) manufacturing a satellite for a third party and then launching it.  Someone will manufacture Starlink satellites just as someone manufactures OneWeb's.  Same applies to launch services... doesn't matter if a company is launching its own sats -- still counts as a launch.
« Last Edit: 06/15/2018 10:36 PM by AncientU »
"If we shared everything [we are working on] people would think we are insane!"
-- SpaceX friend of mlindner

Offline Asteroza

  • Full Member
  • ****
  • Posts: 626
  • Liked: 81
  • Likes Given: 2
Re: SpaceX - now a satellite manufacturer
« Reply #2093 on: 06/18/2018 04:42 AM »


Each sat has 5 laser links to its neighbors. Each of those should be equal in throughput to the U/D link.

Of note, freespace optics lasercomm is not much higher than 2-3Gbps for near COTS stuff right now, so unless you are doing some impressive multiplexing you are not going to see 100Gbps+ unless somebody has been hiding some innovations...

Those 2-3 Gbps are for a single wavelength, right? So you mean multiplexing by using lasers with different wavelengths for the same beam?

Generally there are wavelength options, as well as perhaps some polarization options. It helps that the lasercomm appears to be strictly in vacuum, so the usual atmospheric window limits are not present.

Offline biosehnsucht

  • Full Member
  • **
  • Posts: 288
  • Liked: 94
  • Likes Given: 146
Re: SpaceX - now a satellite manufacturer
« Reply #2094 on: 06/18/2018 07:45 PM »
I doubt the user terminals will cost $1000 - though I would not be surprised if the actual service (possibly including the terminal cost) was $50/mo.

Offline Bananas_on_Mars

  • Full Member
  • *
  • Posts: 142
  • Liked: 83
  • Likes Given: 14
Re: SpaceX - now a satellite manufacturer
« Reply #2095 on: 06/18/2018 08:07 PM »
I think at some point they said they want to bring the COST down to 200-300$. Can't find a source for that though.
What the PRICE will be - who knows. That will largely depend on their local partners.

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2096 on: 06/24/2018 04:47 PM »
C-COM believes that they can manufacture the phased array terminals for LEO constellations at under $1000/unit in high quantities.
http://spacenews.com/c-com-university-of-waterloo-test-hardware-for-new-phased-array-antenna/

The quantities being discussed are >1,000,000 units. At $500 per unit that is a market just for one provider much less the emerging three or four LEO constellations with hopes of the multi-million subscriber counts, is 1 million units at $500 is $500M in ground terminals which is not a small market. Eventually this market could be as large as 100 million units for a market size of $25B to $50B with a unit replacement rate of every 7 years. Making the average yearly sales of ground terminals around $3.5B to $7B.

An example of a invisible source of revenue/profit for Starlink is the rental/buyout of subscriber ground terminals. At 10 million terminals replaced every 5 years (2 million /year) and at a price per terminal (cost plus a 20% profit margin) of $500, this yields a revenue of  $1B/yr and a profit of $200M/yr. If units are purchased or rented at a rate where the units are paid off in 3 years that is a monthly bill to the consumer of $13.89.

When the ground terminals become this cheap then expect the consumers to continually upgrade their hardware every few years (based on twice the length of time than cell phone turnover rates) this is about 5 to 7 years.

Even though 3rd party vendors for ground terminals will exist at some point, the software that runs on each of these terminals will have a separate software license ID and revenue for it from SpaceX/Starlink. So SpaceX/Starlink even though they may not make any more ground terminals after a certain point would still make money for every Starlink terminal produced for the Starlink software license.

Added:
A side note is ruggedized ground terminals for industry and military use at a price of $2,000 to $5,000 per unit at the yearly sales rate of 1 million units for an additional market for such units of the size across all providers of ~ $2B to $5B  per year.

Expect the LEO broadband ground terminal market size in $'s to be about <10% that of the new cell phone market but >1%. With the yearly new cell phone market size of $478B (1.54 billion units) that is a yearly market size for LEO broadband ground terminals of $4.8B to $48B.
« Last Edit: 06/24/2018 05:35 PM by gongora »

Offline DistantTemple

  • Full Member
  • **
  • Posts: 270
  • England
  • Liked: 156
  • Likes Given: 287
Re: SpaceX - now a satellite manufacturer
« Reply #2097 on: 06/24/2018 05:19 PM »
A large percentage of the population carry around $600 iPhones that have a service life of 2 years, (and a 3rd passed onto a poorer relative! d child.)
Will the terminals all be individual, or will there be a tendency, or plan, to use them as "hot spots"?
We can always grow new new dendrites. Reach out and make connections and your world will burst with new insights. Then repose in consciousness.

Offline oldAtlas_Eguy

  • Senior Member
  • *****
  • Posts: 3459
  • Florida
  • Liked: 1864
  • Likes Given: 223
Re: SpaceX - now a satellite manufacturer
« Reply #2098 on: 06/24/2018 06:51 PM »
A large percentage of the population carry around $600 iPhones that have a service life of 2 years, (and a 3rd passed onto a poorer relative! d child.)
Will the terminals all be individual, or will there be a tendency, or plan, to use them as "hot spots"?
US probably individual other locations (non Europe) most likely hotspots. OneWeb's marketing to 3rd world customers is as hotspots for groupings of 100 to 1000 users that share the cost of the hotspot. 1000 users sharing the cost is ~$0.20/month per user. OneWeb is still likely to sell and provide services to a probable up to 1 million hotspots as well as a few million individual terminals. Same is likely of Starlink initially but more likely as a service provider of businesses where they need the higher bandwidth and have the funds to pay the ~$200/month charges. Starlink will expand to even a much larger individual user base than OneWeb as time goes by because of the higher bandwidth capability offered. Expect OneWeb to be offering similar bandwidth through their second or third generations sats after 5 years of operation as competition forces them offer to similar services for similar prices. The costs of the ground terminals forces the bottom prices for broadband to be at or above $50/month irregardless of actual bandwidth. So if the top end provider (1Gbps access) is $200/month a service offering a $50/month prices must provide 250Mbps service to be seen as competitive. Same $/bps.

Even though initially the higher bps rate ground terminals will be significantly more expensive. Very quickly the cost of the ground terminals for any bit rate will merge to a singularly low cost because of technology/manufacturing will advance on GT's faster than for the sats because of the quantities involved and the large number of competitive sources for hardware innovations.

Offline LouScheffer

  • Full Member
  • ****
  • Posts: 1866
  • Liked: 2420
  • Likes Given: 269
Re: SpaceX - now a satellite manufacturer
« Reply #2099 on: 06/24/2018 09:17 PM »

Even though initially the higher bps rate ground terminals will be significantly more expensive. Very quickly the cost of the ground terminals for any bit rate will merge to a singularly low cost because of technology/manufacturing will advance on GT's faster than for the sats because of the quantities involved and the large number of competitive sources for hardware innovations.
The physics of forming phased transmit and receive beams does not depend on the vendor.   The modulation and packet schemes do, but they can be done in software or at worst an FPGA.   So the same ground terminal might be able to work with any of the Ka band satellite vendors.  This would further increase volumes and lower prices.

Tags: