Starlink terminal revision 4: overview and tests
olegkutkov.me
olegkutkov.me
Parameter REV3 REV4
Download speed, Mbps 199 307
Upload speed, Mbps 10 15
Average ping, ms 93 88
Jitter, ms 111.9 9.2
Max power, W 55 100Indeed - significant quality of life improvement for drone operators !
Now its worse, the Russians also use them [1].
[1] https://www.aljazeera.com/news/2024/2/11/ukraine-intelligenc...
They are a bit upset that Musk refused to allow them an offensive use of Starlink which would have, in his opinion, escalated the war.
https://www.reuters.com/world/europe/musk-says-he-refused-ky...
Your comment seems to contradict itself. The technology is so valuable that the GUR is upset Musk didn't provide it to them, but not so valuable that deployed units and operators would be targeted?
I think this war has demonstrated that Starlink is priceless.
"The first casualty when war comes is truth"
[1] https://www.google.com/search?client=safari&rls=en&q=starlin...
Russia has publicly mentioned having 2.4 and 5 Ghz truck mounted scanners
(Tobol)
I'm thinking something like 100m but maybe that doesn't make much difference under shelling.
You certainly can't be mobile anymore with this setup.
We use optical fiber instead but this requires a very large, heavy and warm battery to stay very close to the Starlink dish while the operator can now be kilometers away. You can now disable the Starlink Wifi router but only if you patch the encrypted firmware and spoof the account credentials. SpaceX calls that illegal and will disable and brick your hardware as soon as they find out. Doing this extensive conversion is only for very capable hackers and has been done by only three people on our planet, as far as I am aware [2][3]. If you need similar patching and optical wireing, we will be happy to eductate you, for a small fee.
[1] https://en.wikipedia.org/wiki/Power_over_Ethernet
[2] https://hackaday.com/2023/08/31/diving-into-starlinks-user-t...
[3] https://www.wired.com/story/starlink-internet-dish-hack/
I also think UWB radio is going to be big in the future for communicating with nearby devices in warzones.
In the end, any sufficiently efficient signal is indistinguishable from noise to anybody but the intended receiver(s), no?
As I understand it, the same is true for e.g. DSSS (for receivers not knowing the spreading code), and the primary advantage of UWB is its precise ranging capability (due to its ability to reject multipath propagation errors and the generally high bandwidth, yielding better spatial resolution).
A transceiver could use attosecond 10^−18, femtosecond 10^−15 picosecond 10^−12 or nanosecond 10^−9 pulses at very irregularly intervals. That seems (almost) random to any observer but not to a receiver which has pre-agreed those irregularly intervals with the transmitter (for example with quantum key distribution, with entangled particles). The receiver measures if there was a signal or not. It does not use power level, frequency, or phase (or a combination of these) of a sinusoidal wave but by generating radio energy at specific time intervals and occupying a large bandwidth, thus enabling pulse-position or time modulation.
Not just spacial distribution can be used. You could use polarised photons, electron spin, etc.
In my wafer scale integration [2] I use very few free space photos to flip a 1 v transistor in an ultra wideband mode.
I refer you to [3] for [4] for a better explanation, even though ultra wideband is not mentioned specifically.
Contact me directly, I'll be happy to lecture for a few hours to answer your question.
[1] https://scholar.google.com/scholar?hl=en&as_sdt=0,5&q=ultra+...
[2] Smalltalk and Self Hardware https://vimeo.com/731037615
[3] Stanford Seminar - Saving energy and increasing density in information processing using photonics https://www.youtube.com/watch?v=7hWWyuesmhs&t=272s
[4] Attojoule Optoelectronics for Low-Energy Information Processing and Communications: a Tutorial Review https://arxiv.org/abs/1609.05510
For UWB, the energy of the signal is spread over a much wider frequency band, of at least a few GHz and up to tens of GHz. Therefore UWB behaves like a white noise with a much greater bandwidth than DSSS.
It is not difficult to jam the entire band that can be used by certain kinds of DSSS signals or to receive all of it and process it digitally to search for signals within it.
Such operations are much more difficult for the much greater bandwidth used by an UWB signal.
And regarding jamming resistance: Couldn't a jammer just transmit random pulses in the frequency band in question? Why is jamming harder than just transmitting a lot of useless data on the same channel using whatever modulation the targeted transmitter uses as well? Or is there just too much space (or rather, time) to cover with reasonable transmit power?
Isn't it possible out of the box these days in "bypass mode"?
https://www.starlinkhardware.com/how-to-bypass-the-starlink-...
And no, a GUI check box has little relation to what the hardware actually does.
You're moving goal posts now. If the stock firmware/software now allows disabling Wi-Fi, you don't need to hack the hardware.
And if you can't trust the stock firmware, i.e. you have reason to assume that it's actively malicious (e.g. by turning the Wi-Fi back on randomly), how can you trust the hardware? Who's to say that it doesn't, e.g., have an undocumented transponder feature that replies to a specific type of interrogation?
>You're moving goal posts now.
I'd say you where interpreting my comments on hacking too narrowly.
It would be dwarfed by the emanations from the, you know, space transmitter co-located with it. Locating the dish isn’t the issue, the proposal was using fiber to make the dish (detectable) and the operator (vulnerable) separated by distance.
For example, do you trust SpaceX to not relay your location to your enemy? Do you want to broadcast your identity at all if there can be a man-in-the-middle-attack? A dozen other reasons....
A military service would probably use wider beams coming from the satellite, as well as different encryption (making it impossible to distinguish which terminal is being transmitted to on which beam and not using correlatable identifiers on unencrypted protocol layers at all).
Using civilian communications technology like cell phones has been a bad idea, as evidenced in the same conflict.
Use Starlink dishes as phased array point to point links across 500 km without the v2 Satellites?
There are 200 Ukrainian drone startups alone, many more in other countries. They all hack firmware. Hacker News readers should not be surprised.
If the protocol requires you to reveal your precise location to the network (so that satellite beams can correctly target you), how would you get around that using your own open firmware/hardware?
The same goes for authentication using (potentially) long-term stable or even public identifiers, as is the case for e.g. the IMSI with GSM (not sure if later 3GPP generations finally fixed that): A GSM network won't let you attach without revealing your IMSI over the radio interface at least once. You can spoof it, but then you won't get any service.
> Use Starlink dishes as phased array point to point links across 500 km without the v2 Satellites?
That's a completely different scenario than the one we're talking about here ("why can't military users relatively easily put some distance between the Starlink terminal and its users in a warzone?"), and one which I'd assume the military to have existing solutions for.
Are Starlink terminals really directional enough for EMI from an Ethernet port to be a bigger omnidirectional concern than the many watts of power they transmit skywards as part of their normal operation?
Different wavelengths, infrared its already a target in a warzone.
>10Ghz is another type receiver.
The Starlink terminal is already a giant transmitter, after all.
Mostly I'm curious how the hardware is the limiting factor here, I always assumed it was a combination of queueing in the Starlink infrastructure plus retransmits. Maybe a better antenna lowers the number of retransmits needed?
I don't have a proper jitter measure, but FWIW my Gen 1 Starlink in Grass Valley, CA has a round trip time (measured with IRTT) that varies from 30-60ms over minutes. That improved significantly in January 2024 after some change Starlink made. Previously it was more like 40-100ms.
That said they've been evolving it rapidly, dropped the price, and it's really Just Worked. For all fixed locations I really wish there was fiber too, but until that happy day arrives sometime in the distant future I'm really grateful for what Starlink has enabled for rural internet.
Starlink is doing value engineering so there's a company tomorrow.
They hope to be the first constellation to recover lifetime costs and profit. Avoiding bankruptcy
Literally nothing you wrote had anything to do with my comment, just generic LLM type shlock. Try to do better.
There's new board revisions all the time.
Until Q2 2023 SpaceX was losing money on Terminals in the US. The discounts are still there outside the US.
Do you think the cost of manufacturing is now under $250? How much does air shipping to Europe in bulk cost for the dishes?
Profits matter and they are on the way
>For all [rural] fixed locations I really wish there was fiber too
Happy to help you out here. Think bundeling many patched firmware Starlinks up to 10 Gbps to jump-start the rural backbone, then extend it for 100km for $10K. 36 years experience.
Our Fiberhood speciality is rural, easier, cheaper, DIY deployable and already in more locations than you think [1]. Guerilla fiber plus instant Starlink.
[1]https://www.researchgate.net/publication/309254511_Fiberhood...
The jitter/etc fixes look great, but we are off-grid here on limited solar power (we have to unplug the fridge every night or we won't have any battery left in the morning). So a little disappointed to see increased power consumption (glad that we would never need the snowmelt feature here).
In a way, it's great not to have the extra moving parts and complications that come with articulation. Performance-wise, REV4 seems to work fine even when the app complains about misalignment. On anchor, a boat can easily swing +/- 20° within a minute while bobbing around a few degrees, which means any increase in range and power helps considerably.
My only disappointment is the power consumption that is likely inevitable with the wider-range static terminal. The possibility of cutting power draw by 30-50% makes me wonder if I should downgrade to a secondhand REV3.
Can anyone suss out from the text, if REV4 would work with what kind of PoE?
I have what must be a REV3 dish, powered by a custom power source due to incompatibility of the "PoE" used by REV3. I'm doing that to avoid using the starlink router/AP piece. This is a standard, well-blogged about solution.
I thought PoE was good for up to (or exactly?) 56V, so I guess the incompatibility in REV3 is more about the wattage delivered, not the voltage.
I would be eager to get a REV4 just for the jitter improvement, which is a big issue for me. If it also can use standard PoE, that's another huge win. Or at least, if it can use the same kind of power supply I'm using for REV3, that's fine as well.
if you don't have terrestrial internet, i'm doubtful there's another solution that goes so cold. are you currently using such a solution? please share if so.
I'm looking for alternatives because I am limited to 5Mbps down and 1Mbps up and pay extra per month after 100G of data. There are plans that can at most double those data rates but it starts getting expensive.
You can hang a fiber in the trees or trow in water (irrigation). Can be decades before it is detected.
Then the fiber part I didn't understand at all. But I didn't take away he was suggesting you run fiber except maybe from the dish to your MPOE. I could have misunderstood.
Of course this is useless because your ISP isn't going to support it on their end. And GP completely ignored the -40C issue. I think he ignored all context and was only addressing your bandwidth remark.
Lo and behold, the ubiquitu 60Ghz antenna do go down that cold. That might be something your WISP would support, but still weather might be a problem with signal quality.
It feels to me like it would be worth it to not have internet during those extra cold days, if 95+% of the time you have much more bandwidth. Or rig up an external heater -- you only need to get +10C out of it.
I have a similar connection setup to you, but I'm not in cold weather. I run starlink plus my WISP connection. I could reduce it to just starlink if I could get low jitter. Well, not quite. I have to backhaul my cellular signal over the WISP connection. I like how that works vs a booster antenna. I'm not sure if that's going to work well over starlink, jitter or not.