Cor blimey guv! Really?

Can you elaborate on the ad please. How long ago was this? I guess a photo is out of the question?

The n810 would have an extremely high return rate if mass-marketed as a "gps device that let's you browse the web at your favorite coffee shop".

No doubt, it provides a great web experience, but is the world willing to wait 4 min. for a "gps" device to initialize?

Agreed, however can I ask your source for the 4 minutes to lock on?

The N800 is in this weeks Best Buy ad for $299 FYI (kinda high)

Most GPS devices do tend to take longer than usual to lock onto 4+ satellites under any of the following scenarios: a) they are switched back on at a significant distance from their previous locked location, b) it has been left off for some time or c) the unit is not completely stationary while it's trying to establish a lock. Of course any combination of these will make things even worse.

In all the above cases a significant number of more calculations are required to re-establish the lock. I wonder if any of these were a factor with previous tests? If not, then I really hope this is a known issue and it will be ironed out by the final release.

I pre-ordered the N810 yesterday, largely because of the GPS. Should this not function as "reasonably" expected I will promptly return it for a full refund as it will not be "fit for purpose" and hold on to my N800.

I assume he's referring to the power-on time. I imagine most people using these primarily for gps would fully power them off, rather than leaving them on as we would.

For a cold-start (that is, a GPS without any existing, accurate satellite orbital parameters in its little brain) it will _have_ to wait a bit until it can start locking onto the satellites. This is because the initial data it needs are only transmitted every 30 seconds, so after first searching for, and finding a satellite to lock on to, it has to wait a while (depending on where in the cycle it locks on). Only after that can it start making better guesses and connect to multiple satellites, then download the accurate per-satellite parameters, and _then_ you're up and running!

The next time it's easier (unless you're suffering for the cases Faz describes) because it's got enough data to be ahead of the game. A fix for this would be if the GPS could get pre-loaded with orbital data by a quick hookup to the Internet via the N810's BT or wi-fi connection.

I guess this would be an ideal use for something like Plazes or geoclue - if the GPS coordinates of the WiFi access point are known it should help improve the GPS lock time (similar to AGPS?) If I'm at home and connected to my WiFi, when I leave home I bet the GPS will take a while to lock on as it's been out of sight of a satellite for many hours while indoors, but what it should do is use the GPS coordinates of the last known WiFi connection as a starting point (ie. home).

It should actually be able to connect almost immediately in the situation you describe, unless you moved a fair distance since the last time you had lock. The orbital parameters it collected the last time are still valid after several hours, or even days unless the satellites drift too much out of the orbits that can be calculated from those parameters. As long as you're not moving away too far. Or you're in motion.

How good the GPS receiver works in that situation seems to be depending on the device itself, there's a lot of variation. I'm not sure why, but if the unit's internal clock is poor then that would have a negative effect. For these troubled devices I would imagine all external help (e.g. wi-fi location, accurate time) would be of help for a quicker lock.

Something occured to me.. does units with built-in GPS chips still have an internal clock in the GPS receiver itself? Or is it using the unit's clock? (A GPS will synch its clock to the GPS system when it gets the chance, but it needs a fairly accurate clock for the in-between lock situations.) While experimenting with rdate I found that the N800 system clock seems to be extremely precise, it hardly drifts at all compared to a desktop PC's clock. I mean, half of a tenth of a second after a week..

GPS works by triangulation: each satellite transmits its position and timestamp, and the unit works out by the difference in timing how far it is from each satellite, and thus its position. Having a skewed clock makes things more difficult because you need to calculate the amount of clock drift that would give you a consistent reading.

Speaking of GPS, would the N810 be able to use the timing info to automatically adjust its internal clock? From experience, the 770 and N800 do not have NTP clients, and it's impossible out-of-the-box to keep the clock accurate to less than a minute.