Hal Murray wrote:
Providing an RTC has benefits, especially when considering
week-rollover  issues, since when the receiver wakes up it has no idea
of date at all,  pulling in the RTC time and date is a sufficient
hint, and adjusting  with detailed info from the GPS is a trivial
extention. Then adjusting  the RTC is not a hard thing to do every
once in a while. The same  problem could also be solved using EEPROM
space. A byte would suffice.

My comment about batteries was thinking of non-rechargable ones. There is no reasonable way to open the unit to replace a battery. What's the lifetime of a rechargeable battery? I guess it doesn't matter much if the capacity decreases a lot as long as the unit can recover when the battery is dead.

My model of a GPS receiver is that it uses the time and ephemeris data to predict the frequencies to listen on. It needs to correct for Doppler. That's assuming the position hasn't changed (much).

How long is an ephemeris good for? If it's too stale or the RTC has drifted too much then it might as well start cold.

The ephimeris data could get you pretty good cold start even after a year or two, even if it is probably rated as dated anyway. If you haven't moved significantly since last fix (like different continent) it is enought to start tracking the right set of sats to get going. It takes 15 min to get a full ephimeris for the full constellation, and that only requires one single fix anyway. As soon as you get a fix you get the precission ephimeris anyway, and that is what you use for positioning, so the almenac is just for getting an initial fix.

Storing ephimeris and last fix on CMOS SRAM with battery backup gives a head-start, which is more important for positioning GPSes than timing GPSes.

Battery isn't meaningless, but maybe just not worth the trouble for some cases.

Several GPSes have battery input from external, you could put a lead-cell or whatever there if you want.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Reply via email to