Hacker News Reader

Comment by walrus01 | Hacker News Reader

Starlink is highly ambitious. What they're trying to do is basically a miniature version of the o3b network MEO architecture, where every CPE terminal needs to track two satellites simultaneously. The difference is that an o3b terminal starts from about $30,000 , they're trying to build something with a CPE cost under $700.

This is literally a "has never been done before" satellite communications challenge. The only thing comparable for LEO based services is the Iridium architecture, which is designed around a very different use case and data rate (about 3600 bps, for the first gen stuff).

View on HN

...and Iridium famously resulted in the largest US bankruptcy ever, nine months after launch:

https://en.wikipedia.org/wiki/Iridium_Communications#History

Initial investment ~$6B, sold to US gov't for $25m. Here's hoping Spacex has better luck.

Yes, and the successor corporation of Iridium which bought it out of bankruptcy, and renamed itself to Iridium, has been wildly successful in the maritime, government, aviation, exploration/expedition, military markets. Which is where having a 2400 to 3600 baud (or voice and SMS) service that functions literally anywhere on the planet, pole to pole, is a highly desirable service.

The original 1998-2000 marketing materials for Iridium are ridiculous, like a traveling executive is going to take a gargantuan expensive satellite phone with them when traveling internationally, rather than relying on GSM roaming agreements.

> where having a 2400 to 3600 baud (or voice and SMS) service that functions literally anywhere on the planet, pole to pole, is a highly desirable service.

I spent a year at the South Pole, where Iridium was, practically speaking, our only link to the outside world for most of the day. The higher bandwidth satellites are only available part of the day, either due to their orbit putting them below the horizon (Iridium uses polar orbits, unlike most communications satellites), or them being booked by other users. As you say, it's priceless for the occasional problem where help is needed from experts up North. Nominally, the station also has 24/7 text email through Iridium, and a couple voice lines, which is mostly a good thing for people's sanity.

Shortwave radio could address some of those uses, and there is a one there too, but it would be a rather different experience without something like Iridium.

one of the other satellite architectures that could address the South Pole high bandwidth access problem, basically the same thing as o3b, but in 90-degree inclined polar orbits. The o3b meo satellites are generally visible anywhere within plus or minus 45 degrees latitude from the equator, because they are in Equatorial orbits. But if there was sufficient market demand there is nothing to say that a highly inclined orbit, similar system could not be deployed.

the existing starlink test satellites are in highly inclined orbits, so if there were a great deal more of them, this would also be a possible solution.

Wildly successful after $6b in debt was written off, and only in a small (but profitable) niche, not the originally envisioned mass market. I know it sounds ridiculous now, but the original plan for Iridium was to replace that traveling exec's GSM phone, with the expectation that handset size & pricing would drop as the service grew.

There are a lot of "has not been done before" going on right now in the satellite industry. Musk is the only one getting press, though. And keep in mind that the cpe cost hasn't been justified anywhere I've seen. Phased array is still far from $700, so it's not clear how they plan to recoup that.

Yes I am also highly skeptical of what the CPE cost will be. Starlink has been over hyped in the non-satellite-industry technology press as if it will be a panacea for residential, rural internet access and for small business. But I'm thinking the rooftop terminal cost might be more like $3000-5000 of hardware, not totally outside the realm of possibility to recoup the cost for on 36 month contracts, but not something that's going to allow for $90/month residential broadband. More like a small-medium business product. Or $250-300/month.

If they can't do it economically and with enough gain, SNR and low noise factor using phased array (questionable), the rooftop CPE is going to have to look like an identical pair of motorized, two axis tracking 50cm compact cassegrain dishes, each with its own BUC+SSPA and Rx RF chain. Basically miniaturized o3b. Again not cheap.

> $3000-5000 of hardware

How do you get to that?

The will mass produce these themselves seems highly unlikely that they would be so expensive.

I'm basing it on my knowledge of all comparable two way satellite antennas, Rx/tx rf chains and modems. But there really is nothing comparable. The cost may end up being opaque and hidden to the end user wrapped up in 24, 36, 60 month service contracts. Sort of like how you can get a $950 phone for "$0" with contract.

Current phased array two way satellite stuff that is small, and intended for geostationary purposes, is manufactured in such small quantities that it is hard to compare.

If it is cost-effective for SpaceX with 0 customers so far, do you think DirecTV, dish, etc, with millions of subscribers for more than a decade wouldn't do it? I would be good money they've spent their fair share of r&d on this, and the good ol ~70cm parabolic dish wins out. And they don't even have to solve the problem of the satellites moving.

DirecTV and others' Rx only geostationary market is 50x larger than small two way vsat, because most of the USA doesn't need vsat service. They're all based on standard cheap geostationary rx only tech. A DirecTV dish and feed + lnb are super simple and cheap to manufacture.

I love jargon heavy posts. I just dropped an hour learning about o3b, MEO, CPE, low earth orbit, data rates. This kind of googling leads to new hobby pursuits and some times new career thinking.

Now study geostationary c, ku, and ka band transponder capacity, earth stations, vsat modems/dishes/rf chains, to have a better understanding of the current two way satellite status quo.

Ka, as in, radar?

Yes, though two way satellite Ka is generally uses to refer to all of the bands above the "traditional" Ku band FDD frequency channels, but below 31GHz.

Not exactly the same bands as ka radar.

Me too, I've learned so many things by spending a few hours going down the rabbit hole.

Reply on news.ycombinator.com