Starlink Satellites Observed

I think I've finally seen them first hand.

Early this morning around 5am in early dawn twilight I could see them
transiting S to N at regular intervals of about 35s across my bedroom
skylight. I counted about six of them. Two of them were traveling within
just a degree or two of each other. Just happenstance. I've not tried on
purpose to observe them, but now I know best times and where to look.

I will probably be an early adopter if the service is not too expensive
as a backup to my cable Internet which goes down every time I lose
utility power, which is a problem when I'm working from home. I have a
backup generator and UPS's to keep things alive otherwise.

Dave

On 2020-04-23 10:52 AM, JF Mezei wrote:
On 2020-04-23 09:59, David Spain wrote:
I think I've finally seen them first hand.


Of the satellites that have been launched, what technology is used for
satellite to satellite communications? I read that the fancy lasers have
not yet been deployed.


From what I can see there isn't any. And apparently won't be until the
end of this year (2020). See Section "Satellite hardware" para. 2:

https://en.wikipedia.org/wiki/Starlink

The details of those lasers is apparently a Starlink trade secret.
According to this link:

http://licensing.fcc.gov/myibfs/down...nt_key=1210294

In a letter dated April 20, 2017 from an attorney acting as counsel to
SpaceX to the FCC's Satellite Division Chief entitled:

" Space Exploration Holdings, LLC, IBFS File No. SAT-LOA-201611215-00118"

In response to questions 7 and 8 from the FCC I quote (footnotes elided):

quote
Q7: For optical inter-satellite links, please provide the wavelength,
power, duty cycle, beam diameter at emitter, and beam divergence. In
addition, please provide the power margin at the receiver at maximum
operating distance.

A7: SpaceX has dedicated considerable resources to developing the
technology supporting its optical inter-satellite links (“ISLs�), and
this information is highly proprietary and commercially sensitive,
especially given that other NGSO applicants have also proposed use of
optical ISLs with their systems. Accordingly, SpaceX is providing the
requested information with respect to the current generation of that ISL
technology under separate cover with a request for confidential treatment.

Q8: Please indicate whether optical inter-satellite links will be
coordinated with other systems proposed in FCC applications and with the
DoD's laser clearing house, and, if such coordination has commenced,
please address the status of coordination.

A8: SpaceX plans to operate its optical ISLs at a frequency greater than
10,000 GHz. The Commission has consistently held that these optical
transmissions fall outside its jurisdiction over radio communications.
This conclusion is consistent with international norms as well, where
the ITU Convention defines the term “radio waves� as “electromagnetic
waves of frequencies arbitrarily lower than 3000 GHz, propagated in
space without artificial guide.� Indeed, the one time that the ITU
looked into the possibility of adopting procedures for free-space
optical links, the U.S. took the position that “interference between
inter-satellite links would also be rare due to directed and narrow
beamwidths, and the vast geometry of space,� and that therefore “there
is no evidence to suggest procedures for free space optical links are
needed.� The ITU agreed, and the underlying resolution to look into this
issue was deleted. Regulation would be especially unnecessary for the
optical ISLs SpaceX plans to use, which operate at a very low power
level – low enough to qualify as Class 1 laser, which cannot emit laser
radiation at levels that are known the underlying resolution to look
into this issue was deleted. to cause eye or skin injury during normal
operation and thus are not generally subject to regulation. These same
considerations effectively obviate the need for coordination of such
ISLs between satellite operators. And, unlike lower frequency emissions
within the Commission’s regulatory purview, beams at or near optical
frequencies exhibit virtually no emissions outside of that narrow beam.
In addition, because optical receivers are also highly directional,
interference on an optical link would only be likely if another
satellite (1) itself used ISLs, (2) passed through
the narrow optical beam from SpaceX, and (3) did so with its ISL
receiver aligned parallel to the SpaceX beam. This combination of
events is extremely unlikely and makes any formal coordination
unnecessary. Moreover, even if such an unlikely event were to occur,
orbital dynamics ensure that the conjunction would only last fractions
of second and could only be repeated through deliberate action.
Likewise, SpaceX is aware of no general requirement for users of optical
ISLs to coordinate with the Department of Defense (“DoD�) Laser
Clearinghouse. This clearinghouse is intended for DoD users and others
using lasers intentionally directed at DoD assets. SpaceX’s proposed
system does not meet either of these criteria. More importantly, as
noted above, the very low power level and extremely narrow beam
effectively mitigates any risk of harm. Notably, the DoD
Laser Clearinghouse is operated by JSpOC. As discussed in its
application, SpaceX works closely with JSpOC on a range of issues,
including planning for its proposed constellation. SpaceX would, of
course, coordinate with other operators to the extent any such
requirement is adopted or imposed.
end-quote

Satellite-to-satellite isn't needed if you are implementing the
traditional "bent-pipe" architecture. Of course these satellites aren't
that traditional analog model either. But let's say you are only
switching to one ground station from any given satellite, which follows
essentially the same data path architecture as "bent-pipe". I am going
to speculate that there are enough smarts in both the satellite and
"user terminal" ground station/antenna, to track signal integrity (not
just strength) to be able to maintain constant contact with a specific
Starlink ground station, even should that require automatic switchover
to another satellite. In other words I suspect the intelligence to do
that already exists in the existing
satellite/user-terminal/ground-station hardware, otherwise why waste 460
satellites w/o that capability?

Where sat-to-sat gets interesting is when you want to optimize a path to
a *particular* ground station. It might be far more efficient to do that
directly using a sat-to-sat connection before beaming a signal back down
to Earth. Or to bypass a ground station that is temporarily
out-of-service. Or to bypass a congested terrestrial network, etc. Also
multicast (point-to-many) might be better served using the inter
satellite connections rather than broadcasting across a terrestrial
network which also consumes the bandwidth of multiple ground stations.

Again, these are optimizations and not required to get the base system
up and running. IMO.

Dave

On 2020-04-24 3:50 AM, JF Mezei wrote:
On 2020-04-23 11:59, David Spain wrote:

From what I can see there isn't any. And apparently won't be until the
end of this year (2020). See Section "Satellite hardware" para. 2:

Thanks. Interesting that SpaceX made lasers appear like a done deal and
the "piece de résistance" of their new satelite constellation that would
solve the latency problem, but didn't actually implement it.


The only way this improves latency is if they can route sat-to-sat to a
specific ground station that avoids terrestrial congestion. In most
cases the best solution to latency is "bent-pipe" which routes directly
from your ground antenna to the satellite back down to the nearest
Starlink ground station. I suspect that this is going to be set up to
talk directly to the nearest neighbor satellites that are in the same
(co-planar) orbit. Either "ahead" or "behind" (or both). Aiming lasers
any other way (to non-planar satellites) would require incredible slew
capability of the emitter that I suspect doesn't exist. So the ground
station being routed to would have to be within the beam width of one of
the satellites in co-planar orbit. To push traffic across non-planar
satellites would require a ground station to relay across the planes.

Initially, I though it was just for the first test batch which would be
a pilot to test the satellite, their launch, orbit keeping etc. But now
that they have launched hundreds.

Is SpaceX launching incomplete satellites just to occupy the orbital
space allocated to them to avoid losing it? (knowing LEO satellites have
short lifespan it becomes easier to insert fiully fitted satellites when
ready) ?


No I seriously doubt that. There are far too many for that. No as I
explained previously, sat-to-sat is not necessary for the traditional
"bent-pipe" relay. It should work just fine without this. As long as
each satellite can "talk-to" both your ground antenna and the Starlink
base station simultaneously.


So for the current satellites launched, does this mean that you can only
receive service if you are in the "vicinity" of a ground station such
that the satellite that you see must also see a ground station to bounce
signals to/from ?

Or do they have inter-satellite relay via radio waves as backup to lasers?


As far as I know there isn't inter-satellite relay via radio at least
for traffic. Maybe command and control. I don't know, but I doubt it.

As for your first question, remember these satellites are 300+ *miles*
up. That gives them quite an aperture. No only is your ground station
essentially line of sight but also a good piece of the continental
United States. Not as good as GEO*, which sees ALL of the US, but
probably good out to about 1000 miles in all directions. So a direct
relay between a customer station in say Boston and a Starlink ground
station in say Maryland should be no problem. The issue is the beam
width and orbital position. Since, unlike GEO, the Earth terrain is
moving relative to the satellite, eventually it will pass outside the
cone in which the satellite can communicate with either or both ground
stations. At this point communication will have to be passed to another
sat. This system uses phased array antennas which are electronically
steered and with spacing between satellites in co-planar orbit to be
within a few degrees of each other, the system can automatically switch
between moving satellites electronically to keep your antenna in
constant contact with any ground station within the beam cone of the
relaying satellite.

None of this technology uses or needs lasers. It is all done via radio.
Thus the first gen satellites without lasers can still implement the
system. As a said before, lasers are an optimization not a requirement.

Dave

*GEO: Geosynchronous Earth Orbit

On 2020-04-24 1:06 PM, JF Mezei wrote:
On 2020-04-24 09:10, David Spain wrote:

No I seriously doubt that. There are far too many for that. No as I
explained previously, sat-to-sat is not necessary for the traditional
"bent-pipe" relay. It should work just fine without this. As long as
each satellite can "talk-to" both your ground antenna and the Starlink
base station simultaneously.

The "as long as" is the key restriction here. With LEO, the "view" each
satellite has is fairly limited.


I think my 1000 mile circle is pretty accurate. With that in mind how do
you see this as "fairly limited"? You do know that 4GLTE range is 3
miles from cell tower in most cases? 5G millimeter is going to be far worse.

If a ground station has full internet
connectivity, then the odds are that people without that footprint also do.

I'm not sure I understand what you are saying. That footprint is ~1000
miles. If a station is serving multiple satellites in multiple planes
the footprint is restricted simply by signal integrity vs azimuth. A
footprint measured in 1000's of miles. Terrain and population density
greatly effects terrestrial bandwidth in terms of obtaining quality
service. For ground based providers it's far more complicated than
installing a pizza box sized antenna at a residence.

If the goal is to serve unserved communities, then a satellite that is
flying over such unserved community is far less likely to also have line
of sight to a ground uplink station.

Huh? How do you reason that? There are multiple planar orbits with
multiple overlapping coverage circles. There isn't going to be ONE
Starlink ground service station. I suspect there is a formula for
placing these at certain deltas of longitude to provide some degree of
overlap. And these stations can and will talk to each other via ground
service. How do you even figure that?

On this group, I was told over and over that the magic lasers would
allow SpaceX to make its money linking London traders with Wall street
with less latency with light tracveling in space than any fibre optic
glass cables (where light travels slower) cable traveling under the ocean.


Magic seems to be the thinking here. No way Starlink latency "across the
pond" is going to be better than trans Atlantic fiber optic. No way.
It's physically impossible. If anyone told you that they're lying.
Remember it's 300+ miles up and 300+ miles down MINIMUM of signal travel
required that you DON'T have with fiber optic. That's 600+ miles.

So if this isn't happening and that london trader will have latency from
his office up to satellite 400 miles up and 400 miles back down to a
london groud station before getting onto the wireline Internet to Wall
Street then he has better service connecting directly to wirreline
Internet as they do now.


Given your scenario, yes.

Dave

On 2020-04-24 1:52 PM, David Spain wrote:
On 2020-04-24 1:06 PM, JF Mezei wrote:

If the goal is to serve unserved communities, then a satellite that is
flying over such unserved community is far less likely to also have line
of sight to a ground uplink station.

Huh? How do you reason that? There are multiple planar orbits with
multiple overlapping coverage circles. There isn't going to be ONE
Starlink ground service station. I suspect there is a formula for
placing these at certain deltas of longitude to provide some degree of
overlap. And these stations can and will talk to each other via ground
service. How do you even figure that?


This will undoubtedly be true within the US. However I will concede your
point is well taken for regions outside the US. There will be vast
regions of the planet that will have no coverage due to lack of ground
stations until sat-to-sat is established. It's more than just a question
of the cost of installing ground stations. It's also a bit of a
regulatory headache.

Once sat-to-sat capable birds start flying it will far easier to just
install a pizza-box antenna at the user's site to get going. Your
latency will no doubt vary, depending upon destination.

Being a resident of the US, this is not particularly a problem of
concern to me.

It is interesting to consider power consumption requirements of each
satellite. There is no question that there will be times of intense
traffic and sparse traffic as the satellite orbits. And it won't be at
uniform intervals either, since the orbital plane traverses the planet
as it rotates.

Dave

On Apr/24/2020 at 13:52, David Spain wrote :
On 2020-04-24 1:06 PM, JF Mezei wrote:


On this group, I was told over and over that the magic lasers would
allow SpaceX to make its money linking London traders with Wall street
with less latency with light tracveling in space than any fibre optic
glass cables (where light travels slower) cable traveling under the
ocean.


Magic seems to be the thinking here. No way Starlink latency "across the
pond" is going to be better than trans Atlantic fiber optic. No way.
It's physically impossible. If anyone told you that they're lying.
Remember it's 300+ miles up and 300+ miles down MINIMUM of signal travel
required that you DON'T have with fiber optic. That's 600+ miles.

Speed of light is about 30% slower in a fiber optic cable than in
vacuum. It isn't physically impossible that the latency of Starlink
would be lower than a transatlantic cable. I'm not saying that it is,
just that it is possible.


Alain Fournier

On 2020-04-24 2:30 PM, Alain Fournier wrote:
On Apr/24/2020 at 13:52, David Spain wrote :
On 2020-04-24 1:06 PM, JF Mezei wrote:


On this group, I was told over and over that the magic lasers would
allow SpaceX to make its money linking London traders with Wall street
with less latency with light tracveling in space than any fibre optic
glass cables (where light travels slower) cable traveling under the
ocean.


Magic seems to be the thinking here. No way Starlink latency "across
the pond" is going to be better than trans Atlantic fiber optic. No
way. It's physically impossible. If anyone told you that they're
lying. Remember it's 300+ miles up and 300+ miles down MINIMUM of
signal travel required that you DON'T have with fiber optic. That's
600+ miles.

Speed of light is about 30% slower in a fiber optic cable than in
vacuum. It isn't physically impossible that the latency of Starlink
would be lower than a transatlantic cable. I'm not saying that it is,
just that it is possible.


Alain Fournier

Yeah, but this isn't a straight linear horse race.
Since I highly doubt that the laser sat-to-sat links are not co-planar
that implies the laser signal has to travel within the orbital arc. If
that's the case, it's a physical impossibility.

I will concede the possibility if your destination is within 1000 miles
of the same longitude line. The posited scenario was London/New York.

In any other case it has to transit a ground station to transfer to a
satellite in another orbital plane. Even if you assume the ground relay
station has a large azimuth to cross planes, it's still bent pipe radio
relay and in the best case that's two hops so 1200+ miles. I also
seriously doubt it would be just two hops.

Lets ignore the difference in speed of light between our atmosphere and
open space, it's really small. Averaging the last 6 non-5 digit trans
Atlantic cable lengths*, I get a number of 3702 miles. 30% of which is
1111 miles (rounded). So I still don't see it unless the laser crosses
planes, or the satellites are lower than 277 miles. The first I highly
doubt, the second conflicts with what I have read on-line** (340 miles).

Dave


* https://en.wikipedia.org/wiki/Transa...ications_cable
** https://en.wikipedia.org/wiki/Starlink (see list of launches)

On 2020-04-24 4:03 PM, David Spain wrote:
I will concede the possibility if your destination is within 1000 miles
of the same longitude line. The posited scenario was London/New York.

Let me add to that, I was thinking in terms of the path of the
satellites in the same plane. As the Earth is rotating underneath the
satellites evenutally the two ground stations will eventually also move
out of the range of the circular orbit and have to be passed to a series
of satellites in the next plane.

Dave

On 2020-04-24 11:43 PM, JF Mezei wrote:
On 2020-04-24 13:52, David Spain wrote:

I think my 1000 mile circle is pretty accurate. With that in mind how do
you see this as "fairly limited"? You do know that 4GLTE range is 3
miles from cell tower in most cases? 5G millimeter is going to be far worse.

Unserved areas are often more than 1000 miles from any optilal fibre
where a ground station can be erected. (Think Canada's arctic). And not
sure "horizon to horizon" footprint applies because the downlink MAY
have to be more vertical since passing over more clouds will greatly
limit availability of beam and its throughput.

Yes there are going to be ill served areas without sat-to-sat
communication due to lack of ground station coverage. There are even
worse situations than empty parts of Canada, because there it's
primarily an issue of cost to build the ground stations. It other
countries there will be not only cost, but potentially hostile
governments to contend with. To truly fulfill the promise worldwide it
will require sat-to-sat communications. But not for the most economical
markets.

Once sat-to-sat, even if it's co-planar, you'll be able to relay around
the arc to a ground station within the spot of one of the sats.

(and the antaenna itself may have a limit on how much it can "move"
(either physicallt or virtually) to point to a distant ground station.


I suspect there will be recommended best practice on how to mount the
pizza box antenna. FCC regs may require licensing of installers, I don't
know if that has been determined yet. There will definitely be an
aperture limit and it won't be horizon to horizon. Think of the antenna
shape for a clue. But it won't need to be that, there are enough
satellites in co-orbit within each plane for the system to automatically
track and switch electronically as needed, no gaps.


Magic seems to be the thinking here. No way Starlink latency "across the
pond" is going to be better than trans Atlantic fiber optic. No way.

I had argued that very same thing many times here only to be constently
shot down by the "experts" who beleive all the snake oil produce by SpaceX.


The only way not to be co-planar requires some fancy tracking emitter
gear or the ability to shine radially (power hungry). I don't see it,
but if I'm wrong, great! It would render the rest of what I'm about to
say moot. Otherwise, it'd be far simpler to relay across the planes
using a ground station. Now where this works almost ideally is crossing
hemispheres. There isn't nearly as much cable running those ways and
sat-to-sat will beat that in terms of latency easily. East/West not so
much. Of course if you relay across ground stations closer to the poles
that helps.


It's physically impossible. If anyone told you that they're lying.
Remember it's 300+ miles up and 300+ miles down MINIMUM of signal travel
required that you DON'T have with fiber optic. That's 600+ miles.

Their argument was that because light in vacuum travels fasted that
light in a glass cable, that having a signal from London hop from
satellite to satellite via lasers between satellites woudl beat the
latency. Even Musk made the argument his system would beat latency of
undersea cables.

Have had that discussion with AF already. See para above and previous post.

(The SpaceX fan club of course omitted latency within each satellite as
it processes and routes the packet (initially there would have been
multiple lasers so that one satelite could receive data from others and
choose th path to a satellite that is over a ground station). (and that
meants constantly updating BGP tables or whatever equivalent dynamic
routing they woudl use).

That won't be much in terms of latency. I suspect the routing is nearest
neighbor or ground. Which in case of co-planar laser is only two
targets, leading and lagging. Maybe you send to both and let them figure
it out. But that would waste bandwidth across the system so probably
unlikely. In any case the equipment on-board isn't likely going to be
significantly worse in performance than ground based, power consumption
and cooling of course being the limiting factors here.

If you donwgrade the system to a simple, "go up to satellite, go down to
ground station" then you travel much more compared to just connecting
directly via wireline to that ground station.


Yes. Welcome to satellite, all other things being equal. As I pointed
out earlier. However, when it comes to spanning hemispheres and oceans,
deserts and tundra or even power outages, all things are not equal.

Dave

On 2020-04-25 3:20 AM, David Spain wrote:
The only way not to be co-planar requires some fancy tracking emitter
gear or the ability to shine radially (power hungry). I don't see it,
but if I'm wrong, great! It would render the rest of what I'm about to
say moot. Otherwise, it'd be far simpler to relay across the planes
using a ground station. Now where this works almost ideally is crossing
hemispheres. There isn't nearly as much cable running those ways and
sat-to-sat will beat that in terms of latency easily. East/West not so
much. Of course if you relay across ground stations closer to the poles
that helps.

Well it's fun to speculate and that last statement got me to thinking.
Doing cross plane tracking is tricky until you get into the vicinity of
the poles. There you might be able to use a bi-directional laser (using
a beam-splitter) inclined at X degrees relative to your orbital path to
hit a satellite that is in either a slightly higher or slightly lower
orbit at the polar crossover. Slight is of course measured in 10's of
miles. A goodly distance for these small sats but a smidgen in terms of
orbital distances. If the timing of the orbits is tightly controlled you
can control the degree of slew needed in the emitter to a single axis.
The idea being the laser can cross planes at the poles during the
approach and departure interval for a short period of time. If that
period is long enough to allow overlap between sats in different planes
then you have established continuous contact via switch over for the
entire network, no ground station needed. Beam dispersion would allow
cross linkage to many different orbital planes. Interference could be
eliminated by using different frequencies between non-planar birds.
Regrettably (for us) that info is proprietary to SpaceX.

The ISL could work in two modes. Continuous for your two nearest
neighbors in co orbit (which appear motionless to you) and spot when
hitting a cross plane bird in the polar regions. But for the entire
network it appears continuous since any bird can relay to a polar bird
in its orbital plane. Latency vs ground cable becomes an issue then of
path length vs deltas in the speed of light. You'll win some and lose some.

Dave