Hacker News Reader

Comment by nickysielicki | Hacker News Reader

It's not trivial to do things at tens of gigahertz at a wide range of temperatures. They didn't engineer it this way because they didn't consider the existence of Arizona, Texas, and Southern California. The thermal limits exist because RF is absurdly difficult. Maybe in 30 years we'll have TCXOs and other RFICs that make it trivial. Right now it's not.

tl;dr: Your phone doesn't have radios capable of talking on bands like 10.7–12.7 GHz (X/Ku band) and 37.5–42.5 GHz (Ka/V band)

View on HN

It's not trivial, of course. But it's also not okay to pretend that you can sell a device for $500 (which in reality costs $2,300), and skimp on the fundamental things that everyone else spends money on to ensure these things don't happen. This could have been fixed. It didn't as a cost cutting measure. I don't know why you were talking about V-Band. SpaceX is not using V-Band. They are mostly using low frequency kaband, and we'll be using high frequency soon. This is exactly the same as all other satellite internet providers out there that have solved this problem. They are not special in disregard. They just chose to ignore it.

Existing phased-array antennas cost, at a minimum, tens of thousands of dollars.

Getting Starlink Dishy to around $2,000, and I understand it's closer to $1,000 by now, is an incredible engineering achievement. SpaceX didn't accomplish that by overengineering the shit out of it.

I do not doubt that now that this problem has cropped up SpaceX will address it. If they don't, if in a couple of years this is still a problem with the system in full operation, then criticism would be justified.

There is no evidence I've seen showing that it's less than $2000. Please send a link to the source.

> According to SpaceX President Gwynne Shotwell, the company has “made great progress on reducing the cost.” Initially, Each dish would cost SpaceX $3,000 USD to manufacture, Shotwell said they reduced the manufacturing cost to around $1,500 USD.

https://www.tesmanian.com/blogs/tesmanian-blog/starlink-dish

You mentioned $1000. $1500 is impressive, but is there a follow-up?

SpaceX is literally based in Hawthorne, CA. These dishes are made in a place where it gets stupidly hot.

I get your point, but specifically Hawthorne is a coastal city, so it doesn’t get “stupidly hot”. Average max temp is 85F in August.

https://en.wikipedia.org/wiki/Hawthorne,_California

I don't see why they don't just leave the antennas in the dish then run waveguide to a box in the shade.

Edit: after looking at a teardown I suppose it's because there are ~3000 channels. Woof.

It's not impossible either.

Cheap brute force hack: Have a chip burn power to heat the chip up to a constant temperature. This was done in one of the early precision band-gap references.

More expensive brute force hack: Stick a Peltier cooler on the thing, and servo to a well-controlled temperature.

Communication equipment is critical to work for emergency use, and coincidentally, emergency use is when you're most likely to see high temperatures.

tl;dr: Military equipment is required to operate like this, and has insanely high-performance RF equipment.

Military BGAN doesn’t get anywhere near this throughput in this package, so the size-performance is quite different there IMO, and isn’t a fair comparison to what Starlink is doing with their terminal.

The military is not a business with business constraints. If cost and power draw is no concern, then sure, it's possible. Hell, put a little AC unit beside the enclosure.

These are CPEs and they're sold at a tremendous loss already (a quick google says a whopping $2500 loss on a $500 price tag!). They won't stay in business long if they burn an extra $500 per device to make it suitable for a couple thousand extra customers.

Every one of their dev teams likely costs more in a year than that $1M increment in BOM costs.

I’m not saying it’s a good idea, but it’s probably not going to be a kill the company expense.

I design RF/Microwave hardware. The issue is the long term reliability, specifically the semiconductor lifetime drops drastically with temperature, not to mention the passive components. The temp cycling may be more of an issue, as I have seen antenna arrays pull themselves apart during accelerated lifetime testing. I shutdown all of my stuff at 80C.

While you are 100% right, starlink is not marketed for the military.

Starlink, is like Tesla a decade ago. This is their roadster.

Will I get it, yes. Will I complain about early adpotee stuff, yes.

Is my life better just having it along? 1000% yes.

Starlink is marketed to the military, but so far they've been testing it with the antennae they normally use.

Reply on news.ycombinator.com