Amazon and Spacex plan to launch thousands of satellites - sane or reckless?

[GeorgeOfTheJungle]

Right you are, a satellite can't orbit over a parallel! (except the Equator). So you'd need lots of them everywhere and in all directions. Cool!

[madires]

What's all this talk of moving satellites, won't they be geostationary like sat TV ones ? 

Geostationary satellites have an RTT of about half a second.

[tautech]

No, can't be geostationary because they want/need them to be much closer to ground.

Existing data providers ones are stationary.....why can't land based connections just use greater RF power and sensitivity ?

Actually they did that in the UK. We had a massive microwave backbone run by BT from the 1960s-1990s. This is mostly been supplanted by DWDM fibre.

Yes of course, in the name of speed !

We've had independent sat link providers for some years but the speeds are woefully slow and at times troublesome compared to an even decent copper DSL connection.
Yet satellite will be the most efficient way to reach the isolated and poorer areas of the globe without the need for costly ground based network infrastructure.
Locally we did an exercise with our national infrastructure provider and the cost for a 1.4km fiber backhaul link and new local cabinet and HW was some $200k. 
Now I have a 5 GHz wireless link to fiber 10km away for a service we only dreamed of just a couple of years ago.
A few $ for HW and installation plus some BW package ........done !

IMO this type of data service for those outside the cities is the way to go......get a fast service to a local tower, stick some sector panels on it and sign em all up !
If satellite can provide that speedy link then the rest can be done locally to serve just a targeted community.

[tautech]

What's all this talk of moving satellites, won't they be geostationary like sat TV ones ? 

Geostationary satellites have an RTT of about half a second.

We couldn't live with that but 3rd world countries and remote areas could.

[madires]

If there actually were 50,000 satellites and assuming a size of one square meter, and if they were all in the same orbit, and if there was no idea exactly where each one was but they are evenly spaced, there would be roughly one satellite every 42,000 meters.

Change any one of those conditions to something closer to what the likely scenario will be and I think the odds of an accidental collision will be close to zero.

Do you have a GPS receiver (module, smartphone)? Run a tracking tool and watch how many GPS satellites are seen over time. The number will change during the day. There are maximums and minimums. Satellites aren't evenly distributed. And the next problem is that you have to think four-dimensional, i.e. space and time. You have to make sure that your rocket won't hit any satellite during the whole flight.

[madires]

Geostationary satellites have an RTT of about half a second.

We couldn't live with that but 3rd world countries and remote areas could.

Of course, but with LEOs the enduser's units are easier to handle and cheaper. Rx and Tx with a GEO requires more effort.

[SiliconWizard]

I wonder if someday it will become so crowded up there that they will actually partly shield us from sunlight.

dyson swarm

It would help with global warming...

Oh, of course.
And then, it would give cheap Internet access even to the last few people that don't want it! The little buggers!
Two good reasons to do just that!

[StillTrying]

If there actually were 50,000 satellites and assuming a size of one square meter, and if they were all in the same orbit, and if there was no idea exactly where each one was but they are evenly spaced, there would be roughly one satellite every 42,000 meters.

I make them only 800 meters apart in low orbit.

I can simplify the the whole thing, - it's not going to happen.

[GeorgeOfTheJungle]

...

[rdl]

You're right, I re-did the calculations and get about 850 meters spacing at a 350 km orbit.

Now what would the spacing be if they were uniformly distributed across that sphere?
(Which they won't be but it would give a ballpark idea of the eventual density)

I make them only 800 meters apart in low orbit.

[GeorgeOfTheJungle]

sqrt(4*pi*((6378+350)^2)/50e3) = 106 km apart IIANM

[rdl]

That's about what I got, one every 10,000 square kilometers*. Not quite enough to blot out the Sun, and since they will be in well defined and controlled orbits by necessity, I can't see there being much of a collision problem. I somehow doubt there will ever be that many satellites needed anyway.


edit: fixed, kilometers not meters

[StillTrying]

106 km apart

But they're traveling at 7.5km meters per second so they'll be passing each other quite often.

I've guestimated it at at about 1000 near-misses per second.

[rdl]

Just having the satellites at slightly different altitudes would make near misses irrelevant. Without knowing the exact orbital geometry they will be using I don't really see much point in speculating, beyond just getting an overall feel for the actual density.

[james_s]

Radio waves propagate in the air faster than the light inside a fiber optic, right?

Yep, light travels with 2/3c in glass fiber.

And a submarine cable can have tens or hundreds of fibers, how many radio links with a decent bandwidth can be had simultaneously between a satellite and the ground station? There's a physical limit to that, isn't it? I mean the EM spectrum is what it is and only so many channels can fit, so to upgrade you can't just put 1000 fibers more in parallel and be done, right?

Most submarine cables have 8 fibers. Maybe we'll see some with 12 or little bit more but this is the current limit of what's feasible. The high data rates of many Tbps are achieved by DWDM, i.e. running several light colors each carrying 100 or 400Gbps. 20 years ago it was just 10Gbps per color. With radio links you have a limited frequency range assigned by some authority.

Why so few? I know nothing about submarine fiber cables but given fibers are quite small and crosstalk is not an issue, what is the limiting factor?

[madires]

Why so few? I know nothing about submarine fiber cables but given fibers are quite small and crosstalk is not an issue, what is the limiting factor?

Mostly costs. For the long distance EDFA/Raman amplifiers are placed in-line and they need to be powered. The cable is fed at both ends with a few kV (one side negative, the other positive). If you increase the number of fibers you have to provide more power for the additional amplifiers, i.e. more copper, thicker cable and larger bumps for the amps. A thicker cable makes everything much more expensive (not proportional). For example, the ship for laying submarine cable can load only a specific volume of cable. The thicker cable means less length.

[BrianHG]

No, can't be geostationary because they want/need them to be much closer to ground.

Existing data providers ones are stationary.....why can't land based connections just use greater RF power and sensitivity ?

You have limited transponder bandwidth on each satellite.  Without a focused dish pointing to single out a single satellite, you wont serve many people.  Remember, with LEO satellites, the antenna you can use will be on a cell phone, not a focused dish.  Also, the available geostationary locations based on signal selectivity per given viewing angle is already filled and reserved for mass TV broadcasting and dedicated over seas relays where data latency doesn't matter.

Besides at the geostationary distance, there is a huge 2 second round trip delay.  Talk about a lousy ping time.

[GeorgeOfTheJungle]

Besides at the geostationary distance, there is a huge 2 second round trip delay.  Talk about a lousy ping time.

Can't be 2 seconds RTT if it's ~= 36e3 kilometres above ground !

[tautech]

No, can't be geostationary because they want/need them to be much closer to ground.

Existing data providers ones are stationary.....why can't land based connections just use greater RF power and sensitivity ?

You have limited transponder bandwidth on each satellite.  Without a focused dish pointing to single out a single satellite, you wont serve many people.  Remember, with LEO satellites, the antenna you can use will be on a cell phone, not a focused dish.  Also, the available geostationary locations based on signal selectivity per given viewing angle is already filled and reserved for mass TV broadcasting and dedicated over seas relays where data latency doesn't matter.

Besides at the geostationary distance, there is a huge 2 second round trip delay.  Talk about a lousy ping time.

Yeah I get that as existing sat positioning was decided upon to get the best coverage and bang for buck plus for TV latency matters little.
Our Optus D1 (TV sat) is 37000km for me according to DishPointer which is a stupid long distance compared to what could be achieved with targeted placement over regions that could need data connectivity.
NZ is only 2000km long and two islands where 2 birds could serve us well and give everybody some sort of data connection were there are currently places where there is none.

Again, ping time means nothing to those that presently don't have internet.

[StillTrying]

NZ is only 2000km long and two islands where 2 birds could serve us well and give everybody some sort of data connection were there are currently places where there is none.

The 2 birds would still be 37,000km away, they could have done with building the geostationary orbit much lower really.

[Red Squirrel]

For those saying they should be geostationary keep in mind that orbital periods are dictated by distance.  You can't just put something in LEO and make it geostationary.  So for it to be geostationary it needs to be much further out.  Not sure if there's other factors like mass that also play into this, but essentially to get an orbit you need to be going a certain speed and that dictates how far you end up from the planet.  This of course changes between different planets because of gravity. (I think?)    Geostationary is that sweet spot where you just happen to be matching the Earth's rotation. You of course need to be at a 0 degree inclination, though if you're not you just end up doing a figure 8 pattern over an area.

That said I think they should do a compromise, do a higher up orbit that is not geostationary but further enough out that the satellites get a much larger LOS.  Then they don't need as many.  Of course you get higher latency, but still less than geostationary ones.

At least that's my understanding of it... I'm not an expert on orbital mechanics.  I know enough to get stuff in orbit in KSP.

I think their end goal though is to compete with even fibre optic internet.  In space, latency between satellites is actually LOWER than fibre on the ground, because it's in a vacuum.  Glass slows down the light a bit.

I'm now curious if anyone will ever come up with vacuum evacuated fibre optic cables at some point, it would basically be a tube with a mirrored inner surface that has all the air evacuated out.  That would be quite a feat of engineering though and probably not worth the cost.

[Nusa]

NZ is only 2000km long and two islands where 2 birds could serve us well and give everybody some sort of data connection were there are currently places where there is none.

The 2 birds would still be 37,000km away, they could have done with building the geostationary orbit much lower really.

Geostationary is a geosynchronous orbit that's in the plane of the equator. If you put it in any other plane (synchronous but not stationary), it's not going to be over the same arc of the earth all the time. The required distance to be geosynchronous is defined by physics, not convenience. It's the point where the centrifugal force of a 24-hour orbit exactly matches the gravitational force from the mass of the earth.

Of course you don't need geostationary if you have enough satellites to maintain coverage from any specific orbit.

There are, of course, other possible approaches that might work for a smallish place like New Zealand. Such as balloons/blimps up in the stratosphere. I think Google has a spin-off that's experimenting with that concept.

[tautech]

Thanks Nusa and others for some clues for further learning about Geo-whatever birds:
https://gisgeography.com/geosynchronous-geostationary-orbits/

[3roomlab]

I am just wondering
what is the (cost + maintenance)/(km sq cover * years in service)
starlink is stated to have a lifespan of 5 years

vs normal 200 -300 foot radio masts?

[duak]

The Iridium constellation was orbited in the late 90's and could be considered a precursor: https://en.wikipedia.org/wiki/Iridium_satellite_constellation  There were some 66 satellites plus extras and the altitude was 485 miles.

I worked with meteorlogical satellite groundstations for TIROS-N & NOAA-6/7 in the late 70s.  With the technology of the time we needed a 3 to 5 m tracking dish to achieve a 667 KBPS downlink data rate.  The downlink was simple PSK on an FM 2.3 GHz signal.  Altitude was something like 800 km.  Landsat had something like a 15 MBPS downlink but needed a much bigger tracking dish. Things have come a long ways since then.

It was kind of magical to point the antenna at the horizon where the satelite was supposed to rise from.  Within a few seconds of the calculated time, the receiver would show a signal, the antenna would slew to the exact position, the bit sync would indicate a lock then our equipment would show a frame lock and start recording on magnetic tape.  I think the longest transit was about 15 minutes.  It used a whole tape and was maybe 200 Mbytes.