Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I'm curious why we don't just use GPS's UTC time broadcast to set time? I understand that using GPS to track satellites and find location is power hungry, but I can't imagine that simply receiving a time and setting a clock is any more power hungry than this system.


GPS signals are much higher frequency and very low power and thus don’t work indoors. These HF time signals are way more powerful and can be detected inside buildings.


> GPS signals are much higher frequency and very low power and thus don’t work indoors.

With a $20 antenna placed near the window, together with a modern receiver, it is easy to get nanosecond level of precision for GPS systems. Some receivers even have in-door precision specification.

GPS not only work indoors, in terms of timing, it works actually pretty well indoors.


Why would I need that for? nanosecond accuracy? GPS indoors?


e.g. to get yourself a stable frequency standard that can be used in all kinds of electronics projects & equipments.


In the interest of the previous reply, Casio's GBD-H1000 has the ability to time sync with GPS.


Yeah but it needs to be charged often. That would be a compelling use case for the radio based timing synchronization that are in less power hungry watches, and I'm curious, if this watch just had GPS for timing and none of the other features, if it's power would last as long or even half as long as the older models.


The H1000 is marketed as a fitness watch and solar powered too. The GPS time sync is not an exclusive feature and I find that it isn't necessary to have it on constantly.


GPS determines a solution for position and time simultaneously. It requires 4 satellites to solve for 4 unknowns (X, Y, Z, t). If you know your position you can solve for time using only 1 satellite, but a watch is always moving so it's position won't be known.


You might want to rethink that.

There's no need to 'solve' for time - it's included in the raw data packet from any single satellite.

As any 'solution' carries an inherent error margin, the desire for an excess of sats doesn't stem from solving for time but the reduction of error.


More specifically, you need to solve for the receiver clock bias.

https://www.e-education.psu.edu/geog862/node/1724

Edit: I would welcome corrections though if I am misunderstanding. Perhaps if you're not interested in precise timing you can read the satellite time from the broadcast signal?

Edit 2: OK, I think I understand your point now. The satellites broadcast their timestamps directly. To find a precise position / time you need to find a correction to the receiver clock. But if you don't need a very precise time, you can just use the satellite timestamp.


you understand the reason, GPS units are expensive and take a lot of power ;)

the GPS module itself is probably bigger than most of these watches, and the module itself probably costs more than most of the watches too! (probably something like $100). even if you can power it, which, the power consumed is probably 3-4 orders of magnitude more than the watch.

on the other hand, you have a point about receiving a signal and setting the clock, and I wonder if you couldn't parasitically drive this off the cell network... is there a local reference clock signal in the cell system that you could pull timestamps from and set the clock? cell signals are much much stronger than GPS and don't require constantly tracking satellites to maintain a fix. and while you still need some RF components (as does this casio watch, it's still acting as a RF demodulator!) they can presumably be much simpler and cheaper and lower-power.

(edit: actually I see some people here commenting some watches apparently do exactly this, so your "use the time reference from a single satellite and accept the skew from lightspeed" idea probably does work, although it's a smartwatch and runs its battery quickly.)

really I think most people do not even need continuous updates though. This is something you could build into a "charge cradle" for the watch and have the time reference sent via a data carrier on the inductive charge signal. which does open GPS sync back up a little bit (although wifi/bluetooth would be cheaper and just as good for most people) since the GPS can live in the cradle too.

people are talking about solar watches etc and to be honest if you've got 1-2 years of battery life who cares about solar? Pop it on the charge cradle once a year and charge it inductively. Inductive charging might actually be simpler/cheaper than even those micro solar panels watches and calculators use, and everyone has gotten into the habit of charging their smart watches overnight anyway. And if you do that, it allows you to do the cosmetics of the watch however you want, since you're not tied to having a big solar panel in part of it.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: