Skyline Broadband WISP

two tall redwood treesThis is our second or third largest tree (the one on the left.) It’s a coast redwood that I think is about 6 foot diameter and nearly 200 feet tall. This tree is special because around 170 feet up its trunk is home to the radio and antenna that have provided our internet service for the last 11+ years.

We’ve relied on Skyline Broadband for our only reliable connection to the outside world (which goes down far less than the land line.) Skyline Broadband was, until recently, a mom and pop wireless ISP (WISP) started by a retired couple living remotely who were looking to bring some bandwidth over the mountain from Silicon Valley for themselves and found a sweet deal on some spectrum and tower placement so they were able to become an ISP and distribute that bandwidth to un-served and under-served folks in my area of the mountains.

Skyline Broadband has been a great ISP. Their uptime has been stellar. Their service has been solid. Their speeds to us are limited because even with the elevation we get from the tree mount we still don’t have line of sight to the access point up the mountain and so must use a lower frequency radio that isn’t capable of high speeds. (I think their top tier plan is 20 Mbps, we’re getting 1.5 Mbps.) We will continue to subscribe to SB as our failover — at least for a while, even after we’ve got Starlink up and running.

(Skyline did recently sell to another local ISP out of Half Moon Bay so while it’s still relatively small and local it’s not the same mom and pop outfit that helped me find the best location and line up a tree climber for the mount. I don’t feel as bad about planning on its obsolescence as I would have if it was still under the original ownership.)

I’m a big fan of local businesses, especially those providing critical services, and I worry that SpaceX, a pretty large corporation, is about to put a strain on many WISPS possibly putting some of them out of business. But it is possible for them to compete, to stay current with infrastructure and deliver competitive speeds and pricing so they’re not necessarily doomed. I certainly hope Skyline Broaband, now, survives and continues to offer valuable connectivity to those who can’t get it elsewhere. We’re certainly thankful for their years of service.

250 Feet Tall?

Large coast redwood in Asa's yardThe tree on the left is our “big tree”. (Click the image to see the full height.) It’s the largest diameter and tallest tree on our property. After measuring some nearby trees of similar height using the NASA Globe Observer app, I now think that the big tree (which is about 7 feet in diameter) is more than 250 feet tall.

(The tree to the right is considerably smaller. It sits a bit higher and it’s only about 4 foot diameter but like most redwoods, it grew tall before wide.)

GLOBE Observer

I found a pretty cool app from NASA called Globe Observer which, among other things, lets you measure the height of a tree. I’d often wondered how tall a few of our bigger trees are and today I used the app to measure one of our biggest Douglas firs. (We’re mostly redwoods but we have about half a dozen big Doug firs that range from a few feet in diameter to over 5 feet.)

The big Doug fir tree that I measured today has a diameter of just over 5 feet at chest height (which I already knew) and today I learned it’s about 230 feet tall. Wild! That’s taller than a 20 story building.

I can’t see the base and top of our big redwood from any easy vantage point, which the app requires, so I can’t easily measure it but it’s a 7 foot DBH tree that’s sits lower and sticks up above this Doug fir so I’m guessing it over 250 feet tall.

These are some pretty big trees — nothing like the old growth redwood giants in the park next door, but substantial nonetheless.

If you’re interested in measuring trees around you, check out the app in the Android and iOS app stores:

Note: I didn’t explain the app. It’s an augmented reality app that uses your camera to overlay a marker you line up with the bottom of the tree and the top of the tree. Presumably the app uses the gyroscope in the phone to calculate the angle you tilted the phone when marking the bottom and top of the tree. Then the app asks you to pace off the distance to the foot of the tree. From your height it estimates your stride length and the distance to the tree. With the distance to the foot of the tree and the angle it can calculate the height of the tree.

Plan for Testing Setup

I’ve determined that I need to run power out to a tree in in the back yard to make Starlink work. But it’s possible that the tree mount itself won’t work — that even 100 feet up in this particular Douglas fir the Starlink antenna won’t have a clear view of the sky. There are much taller redwoods all around that may still obstruct the view enough to make this location impractical.

So, before I spend a bunch of effort (digging the trench, laying the conduit, pulling the wires, hooking it all up) and a bunch of money (the NEMA enclosure, the conduit and tools, the wire and tools, the equipment rental) I’m going to verify that the tree mount actually works and I’m going to do that with an experimental setup.

The experiment needs to be low cost and low effort but it also needs to work for a while, weeks possibly months while I determine if this Starlink mount location is going to work. We could still get some light showers so the experiment setup needs to be water resistant if not water proof.

I think for the final job, I’d like to go with a $300 Altelix 14x12x8 fiberglass weatherproof vented enclosure with a thermostat controlled cooling fan. For the experiment I’ll go with something like a $15 Plano plastic toolbox and a $10 generic muffin fan. For the final job, it’ll be a couple hundred bucks for the wire plus (maybe) conduit and the trencher rental but for the experiment I’ll go with a $25 generic outdoor multi-outlet extension cord.

So, my ultimate Starlink power and enclosure setup will cost a bit and take some effort but if Starlink works, it will be a small price to pay. The $50 experiment will let me run with the Starlink service for long enough to determine if the location is going to work. That’ll save me the cost and effort of running power to where I don’t need in case that the location doesn’t work.

(If this tree mount doesn’t work, I suspect I’ll be investing a lot more money and effort in a very tall tower.)

Extend the Antenna’s Cable or House the Power Supply Outside

In order to have even a chance of a clear view of the sky, the Starlink antenna must live more than 100 feet from my house, probably something more like twice that distance. The antenna comes with a hard-wired 100 foot cable.

The design of the system is this. The PoE antenna connects to the power supply and the PoE router connects to the power supply. If more distance than the supplied cables is required, SpaceX recommends extending the router side because it uses standard PoE and can be as long as 100 meters while the antenna side uses non-standard PoE and is intentionally limited to about 30 meters.

I’ve determined though testing that extending antenna’s cable is probably not going to work well for me. Whether it’s my particular Starlink hardware kit, obstructions, or something else specific to my environment, or whether it’s a more widespread phenomenon, I’m in the unlucky camp that sees connection failures when extending the antenna’s cable (low voltage reboots even with very high quality 23AWG cable.) This is probably at least some of why SpaceX doesn’t recommend it. It certainly does work for some but may not for others.

I’ve also heard loud and clear from people on the Starlink forums that I don’t really know but seem knowledgeable and from people I know and trust here that the better solution is to just do the work to bring power to where I need it and get a nice weatherproof enclosure for the power supply. It’s not that difficult. It could be useful for other projects. And most importantly it’s most likely to give me a better Starlink service experience.

So, I’m pretty sure now that what I’m going to do is run underground conduit and power to the base of the tree where the antenna’s going to be mounted about 100 feet up. (What about direct burial of AC wires? Easier? Harder?) I’ll screw or strap a vented weatherproof enclosure to the tree which will house the Starlink power supply. Then I will, in a separate existing conduit, run a high quality ethernet cable from the router, which will live inside my house, out to the power supply enclosure on the tree.

By extending on the router side of the power supply rather than the antenna side, I won’t be adding any additional downtime to the system because of an unadvised hack. Instead, I’ll have power to an interesting location in the yard (with an enclosure that could also house an AP that would be well positioned to bring wireless connectivity to more of our property, including our little office/bedroom cabin on the creek which is currently not connected.)

But before I do the hard work of laying the conduit — where I’d have to tear up some landscaping (including moving some very large and heavy flagstone slabs) I’m thinking about prototyping the system to make sure it actually works by running an extension cord from my house to the foot of the tree and housing the power supply there in a piece of tupperware or something like that. We won’t get any significant rain again until October so I’d have time to do the work of building a more permanent solution while in parallel testing out that rest of my mount plans are even working. (Still not sure what obstructions will be like 100 feet up this Doug fir. There’s no good place to get a view until we’re at the top of the tree and we may find other taller trees are still obstructing too much for that location to work.)

Thanks to everyone who helped me explore extending the antenna’s cable. It seems to work so well for some on the Starlink forum that it was definitely worth the experimentation. Thanks also to everyone who helped me decide on running power to the tree.

It’s going to take some luck for Starlink to work at all for me given my location under so many tall redwoods. I shouldn’t push that luck by deviating from SpaceX’s recommended installation path. And that means leaving the antenna cable at 100 feet and instead extending on the router side of the power supply.


SpaceX is primarily a rocket launch company. They design, build, and launch rockets and spacecraft that ferry satellites to orbit and people to and from the International Space Station. Their workhorse launch vehicle is called Falcon 9 Full Thrust and the latest generation of that rocket, the Block 5, has been in service since 2018. (Earlier versions of the Falcon 9 Full Thrust date back to 2015 and the first version of Falcon 9 came out in 2010.) Falcon 9 will continue to be the workhorse for the company for a few years but they’re deep into development of their next generation rocket.

Where Falcon 9 is partially re-usable, the booster does a propulsive landing to be refurbished and relaunched, the second stage of the rocket is disposable. With SpaceX’s next generation rocket, dubbed Starship, the whole stack, booster and second stage, will be fully and rapidly reusable. Both will perform propulsive landings, and if all goes according to plan will be refueled and relaunched with nothing but light maintenance like we see in the aircraft industry.

Not only will Starship be fully reusable, it will be much larger than Falcon 9. Falcon 9 is 230 feet tall and 12 feet in diameter (pretty skinny for a rocket) and has a payload fairing that can carry 35,000 lbs (volume of about 5,000 cubic feet) to low Earth orbit. Starship will be 390 feet tall and 30 feet in diameter and will be able to ferry over 200,000 lbs (volume of almost 40,000 cubic feet) to low Earth orbit.

As soon as it’s ready, Starship will become SpaceX’s primary orbital vehicle. Today it is in the prototype phase and the upper stage has completed test flights up to 10 km with one successful landing.

One of the things that’s really exciting to me about Starship is that it will be able to launch about 400 Starlink satellites at a time (compared to the 60 they can launch on Falcon 9.) With Starship, SpaceX will be able to take the Starlink constellation to the next level, perhaps increasing its numbers by more than ten thousand satellites per year.

Power Over Ethernet

I’ve got a challenge with my Starlink setup. I need to extend the cable between the dish and its power supply. This is because the dish comes with a hardwired 100 feet of cable and I intend to go up a tree about 100 feet with the dish. I don’t want to have to locate the power supply at the base of the tree so I need to extend the cable to reach inside my house. That’s at least 50 more feet and closer to 100 feet if I want to run it through an existing underground conduit to a far corner of the house.

The dish uses power over ethernet (PoE) so both the internet data and the power travel across the same ethernet cable. Now here’s where the problem comes in. Starlink’s dish is very power hungry. It normally draws 100 watts! and can spike up to 180 watts!! That’s far more than typical PoE and probably why SpaceX limited the length of the cable to 100 feet. So, everything works fine with the 100 feet of cable that Starlink ships with but because voltage drops with the length of the cable, when I extend the cable by another 100 feet, the voltage drops enough that the dish is under-powered and fails to boot, goes into a reboot loop, or reboots when power draw spikes.

So, there are a few possible solutions I can think of (and I’m interested if any of you all have other ideas.)

The first solution is to find an ethernet cable with a thicker gauge wire. The thicker the wire, the less the voltage will drop. I’ve tested with 24 gauge and get the constant reboot loop. I’ve tested with thicker 23 gauge and don’t get the reboot loop but do get reboots when the dish draws more than normal power which seems to happen every couple of hours. If I could find some 22 gauge wire, it might be sufficiently hefty to keep enough voltage to prevent those semi-regular reboots. I can’t find anyone selling pre-terminated 22 gauge ethernet cable though so if I go this rout I’ll have to buy some bulk cable and do the RJ45 jacks myself. I guess that’s not a huge deal but it’s something I was hoping to avoid.

The second solution is to go with a shorter extension. The shorter the run, the less voltage drop and so if I drop down from 100 feet of extension to 75 feet of extension, I might not see the low voltage situation and the reboots. Dropping down to 50 feet would probably be even better but that would just barely reach my house with a more direct aerial run rather than the longer more circuitous underground conduit run.

The third solution is to not extend the cable and instead bring power out to the foot of the tree. This is significantly more work, digging a trench, laying conduit, and running wires from the house to the tree. And then I’d have to find a way to weatherproof the power supply (and the UPS I want to use with it.) That would mean finding some kind of case or building something like a small doghouse. The power supply gets quite warm and so what ever I buy or build to protect it from the weather would also need to be vented to allow it to stay cool. Maybe something like a vented battery box would work.

Starlink’s Path to Success Part 3

The third area where SpaceX needs to make progress in order to be successful with Starlink is satellite to satellite communication. Let me see if I can explain.

Right now, a Starlink customer has an antenna that connects to a satellite, and that satellite connects to a nearby ground station where SpaceX has a high capacity link to the terrestrial internet. The satellites are acting as a one hop relay to the regular old Earth-based internet network and SpaceX must maintain not only the satellite constellation but a whole bunch of ground infrastructure spread out across any geography where they want to do business.

To support U.S. customers alone, SpaceX already maintains almost 50 of these ground stations where their satellite network meets the terrestrial network. Any time they want to add service in a new geography, they have to add more of these ground stations.

To save on this cost, SpaceX wants to enable satellite to satellite communication so that your signal can go up to a Starlink satellite, then hop from one to the next until it eventually reaches a satellite that’s over one of SpaceX’s high capacity links to the internet backbone where it drops down to rejoin the internet. In essence Starlink becomes just another part of the internet backbone.

This will mean that Starlink can deploy to any geography in the world (where it’s legal to do so) without needing ground infrastructure there (unless it’s legally required to do so). Rather than having hundreds, maybe thousands of ground stations around the world to serve all of the unserved and underserved, SpaceX could have just a few large ground stations in strategic locations (where politics are amenable and there’s an affordable hookup to the internet.)

Not only will this ultimately save time and money, it will also allow SpaceX to reach new customers with Starlink — people who don’t live near any ground infrastructure at all. The laboratories at the poles come to mind, and ships on the oceans, and airplanes flying over the poles and oceans. Probably plenty of other remote places as well.

So, where is SpaceX with satellite to satellite communication? They have developed a laser interlink system and have deployed the first batch of satellites that include this system. That batch is in a polar orbit, and when they reach their final orbit (a month’s long process of satellite maneuvering) they will be covering a geogrpahy that doesn’t have any significant ground infractructure so it’s a great test case. Starting next year, all of the Starlink satellites launched will have laser interlinks and in about 4 years, the whole constellation will have the capability.

Oh, and one more interesting thing about laser interlinks. Light travels faster in space than it does in glass so it could actually be quicker to move some kinds of internet traffic across the space laser network rather than the glass fiber terrestrial network. In a race, a packet of information tavelling from New York to Singapore over Starlink could beat a packet traveling over the terrestrial backbone. For organizations who care about latency, whether it’s a CDN company or a high frequency trading firm, Starlink could be a faster network for some traffic and that could be rather smaller but very lucrative business for SpaceX.

Starlink’s Path to Success Part 2

SpaceX innovations have made Starlink *possible* but what will it take to make Starlink successful? In a previous post I discussed the need for SpaceX to increase the pace of satellite launches. In this post I’m going to talk about reducing the cost of the equipment.

Because Starlink satellites are in very low orbits, they move across the sky quickly and that means a Starlink customer needs an antenna that can track the satellites. There are a few ways to accomplish this tracking, including motors that mechanically steer the antenna(s) to follow the satellites. But that’s not the rout SpaceX took for Starlink. Instead, SpaceX designed a solid state phased array antenna which can digitally track the satellites without physically moving the antenna. (Note: the Starlink antenna does have two motors for aiming but those are only used for the initial pointing of the antenna so that customers don’t have to worry about how to aim, or hiring a professional installer. Once it’s pointed in the optimal direction, the motors shut down.)

The challeng SpaceX faces with this approach is cost. A phased array that talks to satellites hundreds of miles away is not a cheap endeavor. Estimates are that SpaceX’s costs for the first generation of this antenna are around $2,500 per unit. That’s a pretty steep price for a consumer to pay for equipment (though, no doubt, there are some people desperate enough for better internet service to happily pay that price.) SpaceX has chosen an equipment price of $500 and so each customer that joins Starlink is costing SpaceX about $2,000 in up front subsidies.

I think SpaceX has a short term plan to drive the costs of the phased array antenna down pretty dramatically simply by producing them in higher volume — not changing the design significantly. A next generation antenna might garner further savings through better miniaturization and design, but I think it will be cheaper suppliers, and assembly efficiencies that come with high volume that will make the most impact in the near term.

Because SpaceX chose to build a computer for an antenna (it even has an ARM CPU, along with a GPS chip, power management, and the phased array which is a bunch of RF ICs on a large PCB) we can expect the price to come down with time and new technology generations — just like laptops and phones become more capable and more affordable every year. In 5 years or so I think the antenna cost problem will be well in hand.

So, today every new Starlink customer costs SpaceX about $2,000 and with a monthly service fee of $99 that customer will not become profitable for almost two years. That’s a long time to wait for profits to start rolling in but SpaceX is in a race for customers with several other low Earth orbit internet constellations — mostly OneWeb and Amazon’s Kuiper, so I think they’re going to have to deal with this negative for the immediate future while production ramps up from tens of thousands of units to hundreds of thousands and then millions.

(It’s also possible that SpaceX will open Starlink to businesses and governments where they may not need to subsidize the antenna and in fact the price to large organizations could actually end up subsidizing residential consumers’ equipment.)

The Plan

Here’s a rundown of what I’ve got so far for mounting the Starlink “dish” in the tree and extending its cable to my house. Prices are as used but I had to buy some of the items in bulk so my actual costs are a bit higher. Also I bought online rather than going into a hardware store during the pandemic so prices aren’t necessarily the best I could get if I was willing to risk COVID-19.

Mounting the Antenna: Using a J pole mount with a couple of extra support brackets attached to the tree with super long structural screws. Cable management is zip ties and cable clips with long wood screws.

Main mast: 1 of Channel Master CM-3090 Telescoping Universal Antenna Mast ($54)
Antenna connection to mast: 1 of 1/2 x 3″ Grade 8 Hex Cap Bolt, Nut, Flat & Lock Washers ($6)
Stabilize mast: 2 of 8″ Heavy-Duty Antenna Wall Mounts ($24)
Secure mast and mounts: 10 of 1/4 X 12″ Hex Head XERATH Timber Framing Wood Screws ($10)
Secure cable to mast: 4 of TR Industrial Ultra Heavy Duty Multi-Purpose UV Cable Ties ($1.50)
Secure cable to tree: 10 of THE CIMPLE CO Cable Clips ($7)
Secure cable to tree: 10 of #8 X 4″ Stainless Oval Head Phillips Wood Screw ($3)

Extending the Cable: Using an outdoor rated coupler to add 100′ of Cat5e cable length. The coupler is housed in a second waterproof cover.

Ethernet Extension: 1 of 100′ American Teledata Outdoor CAT 5E Shielded UV Rated Patch Cord ($40)
Ethernet Coupler: 1 of VCE IP67 Waterproof RJ45 Shielded Female to Female Coupler ($10)
Weather resistance: 1 of CordSafe Electrical Extension Cord Protective Safety Cover ($7)
Weather resistance: Henkel Corporation Loctite Clear Silicone Waterproof Sealant ($5)

Tools: Impact driver and bits for fasteners. A drill guide, clamps and bit for drilling the J pole to couple to the dish’s mast.

Drilling pilots and driving fasteners: Milwaukee M12 Fuel 2nd Gen Impact Driver and XC 6.0 battery ($200)
Drilling pilot holes: Bosch 3/16″ Impact MultiConstruction Drill Bit ($6)
Driving various fasteners: DEWALT Nut Driver Set, Impact Ready, Magnetic, 5-Piece ($19)
Drilling main mast for coupling to antenna mast: Milescraft 1312 DrillBlock- Handheld Drill Guide ($14)
Drilling main mast for coupling to antenna mast: Irwin QUICK-GRIP Bar Clamp, 6-Inch, 2-Pack ($20)
Drilling main mast for coupling to antenna mast: DEWALT Drill Bit, Impact Ready, Titanium, 1/2″ ($18)

Diagram showing Starlink tree mount