Starlink definitely relies on groundstations. There is one a few towns over from me. They are all over the USA. The satelites get the internet from the groundstations scattered all over.
1 Like
ah - based on the wording of the reply I replied to, I read it as saying that the terminal was using a base station, which it does not.
I do not think your terminal is locked to a base station. it is just that the constellation above you is always using that base station. If the temrinal moved enough for the sats to need to use a different base station it would…once starlink allows you to move at all. Right now they are limiting movement due to having only a small percentage of coverage in place.
what kind of speeds are you seeing on Starlink?
Up to 150Mbits down, 30up. Reliably in the 100Mbit range. Intermittent 20 second outages (I’m tuning the health checks to see if they are really smaller than that).
The statistics tell me 1 min of no sats… 3 mins of “other” outages. No obstructions.
So, fine for streaming and downloading… but you need another WAN for full stability and a SFtunnel for critical applications.
2 Likes
The groundstation you are referring to is a “Teleport” or at least that is what my VSAT provider calls their sites. It is basically where the trunk of information from the satellite is passed back down for real-world connectivity. Our large VSAT provider has five teleports across the globe as they provide world-wide coverage. My beams typically connect through three locations in the USA (Houston, California, and Maine). The StarLink is a two-way satellite terminal where it downloads and uploads directly to the spacecraft within a spotbeam of service. StarLink doesn’t currently allow you to travel outside your spotbeam but I see that capability coming once they have the network built out more. I am very interested in it for disaster response as it is far more powerful and less expensive than my 1.2M Driveaway dish. My current package costs in the neighborhood of $58,000 and around $12,000 annually for unlimited service for 10 days a month. Each additional day is another $1200 (30 minutes counts as a whole day).
I am waiting on my Beta approval too!
I’m assuming these “outages” are while the dish re-points to the next satellite and the handover doesn’t work flawlessly…
How often are these outages?
The outages are not during regular satellite handover, since that happens at most every 10 mins. They are on the order of 45mins to an hour apart and for only 5-10 seconds. I need to gather some more data from the sat and run some smoke pings. One theory is that the earth rotates under the shells on the order of about 45mins, Therefore each ring of sats will go from directly overhead vs max oblique angle. Some of the outages correlate to a 45min cycle… others don’t. I should be able to correlate the times of “no sat” to when the 36shell areas are overheard vs an already populated 72 shell area.
Starlink categorizes outages as either “obstructed” which is loss of signal, but the terminal expected to have a sat in sight. “No Sat” where there is a signal loss but the system calculates that no sats were in the correct geometry. and “Beta downtime” Which is just a catch all of ping loss and not the other two categories. So it could be rebooting routers, packet loss on the backhaul, other things like that. I have zero obstructions, so all my loss is No sat or Beta. About 75% Beta, 25% no sat. The maximum has been 4mins/day total.
|Feb 20 10:47:23|WAN: Starlink connected (100.72.190.228)||—|—|
|Feb 20 10:47:21|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 09:46:23|WAN: Starlink connected (100.72.190.228)|
|Feb 20 09:46:15|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 09:32:23|WAN: Starlink connected (100.72.190.228)|
|Feb 20 09:32:15|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 09:26:06|System: Changes applied|
|Feb 20 09:24:40|System: Changes applied|
|Feb 20 06:05:13|WAN: Starlink connected (100.72.190.228)|
|Feb 20 06:05:10|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 06:04:43|WAN: Starlink connected (100.72.190.228)|
|Feb 20 06:04:39|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 05:19:14|WAN: Starlink connected (100.72.190.228)|
|Feb 20 05:19:09|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 04:33:33|WAN: Starlink connected (100.72.190.228)|
|Feb 20 04:32:26|WAN: Starlink disconnected (WAN failed HTTP test)|
|Feb 20 03:57:49|WAN: Starlink connected (100.72.190.228)|
|Feb 20 03:57:41|WAN: Starlink disconnected (WAN failed HTTP test)|
1 Like
I think this link will largely explain the situation. The down-times can be mostly be predicted based on constellation density and geometries. Coverage is far better (even virtually 100%) at higher latitudes but gets rather rough as one proceeds farther south – as expected and acknowledged by Starlink. Of course, one would add to that any outages caused by obstructions due to the user’s siting of “Dishy.”
That simulation is over six months old and used the satellite data from July. It does not update in real time.
I have taken that code and run both the 35° and 25° calculations with data from Feb 4 2021 (there hasn’t been any new complete shells since then) It indicates 25° coverage all day down to about 31° latitude. and 1 min of outage for 35° at my location on the 4th…
This one is an updated visualizer for current data.
The issue is that all of those “draw circles” around the satellite models have a flaw, the terminal is not oriented straight up, and it does not have a 130° field of view, only 110°. The next time I am outside I will measure the terminal’s angle at my Latitude.
I guess we should write some test code to simulate the # of sats in view at a particualar location given the known limitations. 25° minimum elevation angle, 110° view path, ##° tilt. calculated second by second for a 24 hours span.
Operationally my observed No Sat downtime matches the 35° simulation fairly closely, and that is a simulated 110° view angle pointed straight up.
That address is pretty common with cable modems and we generally write simple outbound policies to get there. That does not work for you?
No it doesn’t, as detailed in this thread:
Peplink will need to add WAN static routes to properly support Starlink terminals.
An Outbound policy will send the packet to the MAC address of the default WAN router. If the out of band management interface uses a different MAC address, then it will not see the traffic. Starlnk and other DSL modems use seperate MACs and therefore need a static route. DOCSIS cable modems seem to use the same MAC address and therefore an outbound policy is enough.
I wanted to revive this thread.
Has Peplink added WAN static routes?
Also, is it possible to bond two Starlinks together? The Starlinks both have a default 192.168.0.1 that can’t be altered for their routers, as I understand it.
yes: Starlink / App (Dish Management) Access / Peplink
Yes of course, although using the starlink app will be challenging in that configuration.
And welcome to the forum!
Thank you!!!
Figured it would be challenging to access via the app . Currently using OpenMPTCP but thinking of switching back to Peplink (have used in the past).
Anyone have any recommendations for Peplink gear that would allow bonding of two Starlinks and two LTE connections all together (ATT and Verizon)? No need for the LTE modules, just need to plug into WAN ports with Ethernet.
I’m looking to do the same at my house , I think I may upgrade to a balance580 or do three devices with the balance two and a usb modem or ethernet dongle.
1 Like
Are you able to share your speedfusion settings for the starlink? Did you do Wan Smoothing, FEC and Bonding? and Can you share priority in each case?
Finally can you share your outbound policy priority list?
I’m in Starlink beta and keeping my steady but slow DSL as backup.
Thank you in advance
Actually mine is currently out of service as I’m moving it to a different roof. I’ll do a write up when I sort it out.
In the meantime - check this thread.
Is everyone generally getting ~75% of the starlink speed with a pepvpn or speedfusion connection?
Got my starlink setup at last. No ethernet adapter yet so using a max transit wifi as wan.
Getting 80-100 mbps connected to the starlink or connected to the max transit for a non-pepvpn speedtest.
But for some reason I’m maxing out at about 12 mbps with speedfusion with balance 210 (speedfusion)->ethernet → max transit wifiwan 5ghz->starlink when the same balance is doing 20+ mbps speedfusion with verizion as wan.
(Prior to October 15 2021 the same balance and fusionhub was doing 80mbps speedfusion speed at a different location with no settings changes, other than disconnecting dsl wan and connecting here to cellular wan and now starlink wan)
Need to troubleshoot to figure out why it’s not performing on starlink, but thought I’d ask if anyone else is having any issues with pepvpn or speefusion traffic over starlink right now.
Upgraded firmware of fusionhub and balance from 4.1.3 to 4.2.0.
Changed from FEC off to FEC=LOW
Only 1 WAN connected (Balance 210-ethernet-max transit cat 18-wifi as wan 5ghz-starlink.)
Now getting
61 mbps download speed on speedtest.net to cloudflare over speedfusion
vs.
108 mbps download speed on speedtest.net to cloudflare without speedfusion
(I was previously getting between 80 and 89 on a 100mbps dsl wan when I had the balance at a different location, so the 61 over starlink isn’t great yet, but it’s a lot better than 12!)
Lots of variability - for some reason the speed test site seemed to play a much larger role than usual (all in the same city as my cloud instance) so maybe something with my cloud provider tonight. Or maybe variability of the connections.
I ended up with the best speeds by setting:
Method: bonding
FEC: off
WAN 1: upload only (verizon)
WAN 2: download only (starlink)
Ended up with 66.40 mbps download speed and 44.01 mbps upload speed as my best result testing to frontier.
For some reason doing the same test to cloudflare in the same city ended up repeatedly at around 10 mbps upload but 50-60 download; choosing servercentral or hivelocity ended up with around 30 download repeatedly. So maybe something going on the internet on the cloud side tonight.
Will give it more testing tomorrow.
WAN Quality analysis is showing 110-150 mbps on the wan side if I read it correctly, but 66 is not half bad!