SpaceX Internet Satellite Network: Starlink

[Cosmacelf]

WSJ with a piece about SpaceX raising another $500M on top of the last $500M they raised late last year. Fairly negative, and splices Shotwell’s quotes up to make her sound like she has doubts on Starlink (which is rather unlikely).

SpaceX Is Raising $500 Million Amid Internal Questions Over Satellite Internet Business

 

[bxr140]

Great articles- clever ways to manoeuvre without fuel- genius stuff.

The good news is that renewable energy has always been a major element to spaceflight, and so these kinds of solutions that use electrical energy have been genius for quite a while. Propellant is a non-renewable resource (plus it’s mass, which is generally not preferred) and so it’s use is limited to maneuvering events that can’t otherwise use electrical equipment. In chemical form its also typically way too harsh for precision maneuvering--thanks to Newton's 3rd, (<shakes fist, looks upward>) some of the reaction from the thruster pulse gets absorbed by the structure and that can often translate into unwanted vibration that negatively impacts your pointing performance. Especially if you have big dangly bits (like reflectors or solar arrays) you can see some low frequency effects that you basically have to wait for them to damp out.

Regarding the high quantities of parts required, just about every satellite launched that’s bigger than a cubesat has wheels (and even many cubesats have wheels these days). Because space is three dimensional you need three wheels to have full control, and so most satellites fly 4 wheels in a wonkily oriented 4-for-3 redundancy scheme where if any one wheel craps out you still have 3 axis authority. Farmer's math on 4 wheels over 12000 satellites gets to the 45,000 WAG from above.

3 is often the number for torque rods too, but they're so basic that they are almost never redundant, and often torque rods aren't the primary control mechanism because their total force is pretty low (often their role is to just take momentum out of the wheels). ...and of course, torque rods only work on something that’s orbiting through the van allen belts, so GEO (and beyond) don’t get them.

 

[adiggs]

...and of course, torque rods only work on something that’s orbiting through the van allen belts, so GEO (and beyond) don’t get them.

Of course

 

[ecarfan]

Lower orbit granted for first SpaceX Starlink satellites

Quote: “The first batch of the small satellites—a total of 1,584, which are expected to begin launching as early as May 2019—has already arrived in Florida in preparation for launch. In the lead up to launch, the FCC has grantedSpaceX’s request to lower the orbits of the first group of satellites from the original 715 miles (1,150 kilometers) above the Earth to 340 miles (550 kilometers).”

Wow, that’s quite a change.

 

[mongo]

Lower orbit granted for first SpaceX Starlink satellites

Quote: “The first batch of the small satellites—a total of 1,584, which are expected to begin launching as early as May 2019—has already arrived in Florida in preparation for launch. In the lead up to launch, the FCC has grantedSpaceX’s request to lower the orbits of the first group of satellites from the original 715 miles (1,150 kilometers) above the Earth to 340 miles (550 kilometers).”

Wow, that’s quite a change.

Perhaps less delta-V needed, so more sats in one launch. Also, faster natural de-orbiting, so the first versions will not hang around as long nor take up the more favorable orbits. Also gives a smaller beam spot for the same beam width, along with higher signal levels. Does require faster beam steering though.

 

[ecarfan]

SpaceXNow app still shows the Starlink-1 launch as NET May 2019 but no exact date yet. Exciting to think the Starlink network will start to take shape so soon! Of course commercial service will not be available within weeks, more likely it is many months away. But, progress!

 

[Joel Weber]

Closer to the surface of Earth also has the potential to reduce the round trip time if there are enough satellites for one to be nearby at any given time.

Do we know what ASN Starlink is using?

 

[Cosmacelf]

Closer to the surface of Earth also has the potential to reduce the round trip time if there are enough satellites for one to be nearby at any given time.

Do we know what ASN Starlink is using?

If you knew the ASN, how would that help you find it? I know enough about Internet networking to be dangerous, but don't know how to track down an ASN.

 

[Joel Weber]

https://bgp.he.net/AS27277#_graph4

https://bgp.he.net/AS14593#_graph4

shows some information about what paths some of SpaceX's networking infrastructure uses to connect to various parts of the Internet.

The prefixes tabs indicate SpaceX announces three IPv4 /24s and one IPv6 /48. Arguably one IPv6 /48 isn't even as much as should be allocated for a corporate network for a corporation that has more than a single location. Starlink should be able to trivially get an IPv6 /32 by going to ARIN, stating that they're an ISP, and paying the fee, and if they expect to have more than 65,000 customers, getting an even bigger block to be able to have an IPv6 /48 per customer should be possible.

Looking at Internet routing table information might give us some clues about how quickly Starlink is reaching a state of being a functioning network.

And knowing the ASN won't provide any information about exactly where in the sky the satellites are by itself, since each backbone operator usually has one ASN for its entire global network (although oftentimes when a single corporate entity buys multiple backbones, they don't bother to consolidate them into a single ASN, and so one can observe cases where traffic going from one part of a ``merged'' network corporation in Boston to another division of the same company in Boston has to travel through New York City...)

 

[Joel Weber]

https://bgp.he.net/AS14593#_whois
PeeringDB
both mention Starlink explicitly.

It'll be interesting to see whether they build ground stations with enough bandwidth to upgrade to 100 gigabit ports at places like Seattle Internet Exchange.

 

[ecarfan]

 

[mongo]

5 dozen to be exact.
Twitter

 

[bxr140]

5 dozen to be exact.

Fascinating. It’s quite likely they’re flat and literally stacked on top of one another. I counted 60 gold donut things on the right side of the stack and 30 on the left. Those gold vertical features on which the donuts are located seem like some kind of restraint bars and the black cylinder at the top of each of them is most likely a rod catcher for the release mechanism. The whole vertical gold feature probably falls away from the stack as a single thing when released. The hoops mid elevation could possibly provide some spring force to counteract bending in the long gold features.

There’s also repeating circular features along the close corner of the stack that could be some kind of aperture. There’s 30 of them visible; I’d speculate there’s another 30 on the far side of the stack.

One really radical idea that SpaceX could be using is a near passive attitude control system. Long thing things actually stabilize on their own; perhaps these are long and thin or perhaps they unfold into that form factor? With so many spacecraft you probably don’t need much need for actual control to maintain sufficient network coverage, you just need attitude/pointing knowledge so signals hop from satellite to satellite at the right moments. That would significantly reduce both hardware and associated control software.

Can’t wait to see more!

 

[mongo]

So it looks like two stacks of 30 side by side, each folded twice. Around 120x60x8 inches each. Maybe each with one fold and a half area section and full area section.
The lifting lugs are attached to those verticles, so they seem to serve as an exoskeleton. It will be interesing to see what the release/ staging mechanism is.

 

[Grendal]

I'll just give it a Wow.

I'll add, that is incredible use of space. They used the interior space of the fairing to its maximum.

 

[N5329K]

I'll just give it a Wow.

I'll add, that is incredible use of space. They used the fairing to its maximum.

+1
Great job.
Robin

 

[Grendal]

I find it fascinating that these satellites are nothing at all like the test articles already sent up. This design allows for a much quicker deployment of the constellation. Five dozen per launch on a plain F9 means there is no huge demand for Starship to be involved in the deployment. When the satellites were much larger then getting the number they intend to deploy up pretty much forced the need for Starship.

 

[mongo]

I find it fascinating that these satellites are nothing at all like the test articles already sent up. This design allows for a much quicker deployment of the constellation. Five dozen per launch on a plain F9 means there is no huge demand for Starship to be involved in the deployment. When the satellites were much larger then getting the number they intend to deploy up pretty much forced the need for Starship.

If they can deploy 60 sats per launch at all the different orbits, and maintain 60 per fairing with the inter satellite laser hardware: 12,000 satellites is 200 launches or one per week for 4 years. If it drops to 50 per, then 240 launches, one a week for 5 years. If 40 sats, 300 launches and 6 years. Heck of a cadence.

 

[bxr140]

So it looks like two stacks of 30 side by side, each folded twice.

Yeah, agree. Seems like the feature with 60 donuts is where the two stacks meet. It kinda looks like the ‘back’ stack could be offset a half spacecraft higher than the ‘front’ stack...?

I would guess that the decision to reduce altitude was done to a) reduce or possibly eliminate the need for on board propulsion and b) reduce the power requirements of the payload, both of which would be an enabler (or an after the fact necessity) for this very non-traditional form factor. Combine with a gravity gradient stabilization and you really don’t need a lot of volume in your satellite.

 

[Grendal]

If they can deploy 60 sats per launch at all the different orbits, and maintain 60 per fairing with the inter satellite laser hardware: 12,000 satellites is 200 launches or one per week for 4 years. If it drops to 50 per, then 240 launches, one a week for 5 years. If 40 sats, 300 launches and 6 years. Heck of a cadence.

No question that Starship would still be incredibly helpful for getting such a massive constellation in place. But the new design does make using a standard F9 a lot better.