Having explored Dishy v2 (the round version) in detail, I was excited to get my hands on a v3 – the newer, smaller rectangular dish. However, they’ve been pretty hard to get ahold of, and my “cell” is currently full.

Recently, SpaceX enabled “roaming” functionality, which means the user terminals (uterms – lovingly referred to as “Dishy” by SpaceX) are able to work in any active cell, not just their home cell. This makes it possible to take the uterm just about anywhere and get online – pretty cool. I’ve tested this and it works well with my v2 uterm.

I figured this would be a good opportunity to get a v3 uterm and see how/if roaming works when the uterm has never even been at its home location.

I found an active cell (about 200 miles away from me) and ordered a v3 uterm to ship to my home. The home cell didn’t matter much to me, since I plan on roaming with it anyway.

Surprisingly, it shipped faster than estimated, and about 2 weeks later, it showed up:

Unfortunately for this Dishy, it’s going to undergo some surgery in the name of my insatiable curiosity.

Initial Observations

Compared to v2, the v3 uterm seems more “polished.” It feels more substantial (even though it’s lighter), and the whole exterior seems to be some type of plastic/composite mix rather than the plastic housing on v2. Unlike v2, it’s not clear how the housing is assembled, and how we might start taking it apart.

The v3 uterm combines the PoE injector with the router, which is annoying – there is no ethernet jack anywhere, you only plug the router into the wall, and the uterm into the router (using a proprietary micro-HDMI-looking jack).

Beyond that, it’s much smaller. v2 is circular and about 24″ in diameter, while v3 is rectangular at 12″ by 19.” This will come in handy for one of the upcoming experiments 😃. It’s also significantly lighter, weighing in at about 9lbs, vs. the previous 16lbs, although I suspect some of this weight savings comes from the less substantial metal base.

Booting Up Somewhere Unfamiliar

Based on some previous research, it seemed to me like the uterm has two communication channels – a “control plane” and a “data plane” – the former for configuration/control traffic and the latter for user traffic.

My thought was that the uterm might coordinate data plane stuff over the control plane, and that with the activation of roaming, it might be able to establish a control plane link anywhere – even somewhere new.

So, off we go. Plugged the uterm in and after a few seconds it un-“stowed” itself and went to sky-level.

And…nothing. This is normally where it starts, then shortly after it moves itself to an orientation optimized to keep the best view of satellites as they pass over.

I left it like this for about 30 minutes, and it became clear that it wasn’t going to connect. Nothing at all in the “stats” screen, indicating that it wasn’t even getting a passing connection. The app display was stuck on “Booting” and not even progressing to “Searching.”

So, this means that either the uterm or the network (or both) are expecting it to be in the service area cell, not where it is now, and won’t even try to communicate outside it. This is interesting, and counter to what I expected.

What can we do? Let’s try updating the service address.

Changing the Service Address

Starlink lets you change your service address yourself, online in the customer portal. So, maybe we can find something closer?

I scrolled around a bit and found a cell about 10 miles away that was “open.” This is well within the distance users were reporting success before roaming was even a thing – there is still some “fringe” coverage around the “cell” where connectivity is possible, albeit degraded.

Within seconds after clicking “update address” the app screen transitioned to “Searching” – a good sign. And about 5 minutes later, the uterm rotated to the expected north-ish orientation and…

Sweet!

This implies the network does care where the dish is, at least on initial contact. I’m somewhat confused as to how this is different than when you turn the dish off, move 200 miles away, and turn it back on (which seems to work), but maybe the initial connection needs to happen where expected.

It’s possible this was just total coincidence, but it did come online right after the service address change.

In any case, we’re in business. Let’s see how it performs:

That’s not bad, not bad at all!

Dishy V2 vs Dishy V3

Since I have access to a V2 uterm, I figured it would be interesting to compare. Does the reduced antenna area affect the signal or speed?

I set them up about 10ft apart from each other, so that they had comparable views of the sky, but not so close as to potentially cause interference. I was surprised to find the V3 signal appeared to be much worse during the same time period:

Dishy V3
Dishy V2

This was pretty surprising. The periodicity of the V3 interruptions led me to believe that these weren’t purely obstructions, but rather worse satellite tracking or sensitivity. Bummer.

However, at some point I rebooted V3, and everything changed:

Dishy V3 after reboot

Much better! Pretty much the same as the V2 dish. So, I don’t know what the initial issue was, but things got much better after the reboot. I eventually moved them even closer together to see if there was any interference, with nothing notable:

Cousins?

Initial Conclusions

All in all, V3 seems to be a significant upgrade from V2, at least in terms of form and package. And, it seems to work as well as V2 in my limited testing, although I still suspect V2 will have a slight advantage in fringe connectivity, as theoretically having more antennas in the array means better focus. But in practice, the reduced size probably far outweighs any advantages of the larger terminal.

Unfortunately for Dishy V3, this was just a quick test before going in for surgery.

More to come…