"Didier Juges" <[EMAIL PROTECTED]> wrote: > That's what I meant I suppose. The interesting part is that the timing GPS > receivers don't continualy try to select the best satellites from all those > it can see, so other than eliminating outliers, there may still be some that > are not far out enough to be eliminated, but will negatively affect the > average. What is the criterion to determine an SV is an outlier? How far off > should it be? > > An algorithm could select the best 3 at all times for instance, even if 5 > are visible and unmasked. I am not sure that would be better, but it's a > possibility. On the other hand, I suppose that's what the mask if for. Maybe > we need a dynamic mask so that we could force the receiver to never look at > more than the best 3 or 4?
Didier - I think this is what many are doing when they set their timing receiver to only use satellites above a given elevation. > > > If so, > > > what's the point of a self survey? > > > > ??? If you're wrong by 1000 feet about where you are, then > > that could be hundreds of nanoseconds in time error. And that > > hundreds of nanoseconds would jump around like the dickens as > > you chose different satllites or the chosen satellite pans > > across the sky. > > > > I agree, but to do a self-survey, you need to look at more than one > satelite, which was my point. Oh, all the timing receivers I know of refuse to self-survey without at least 6 satellites in view. By definition it's impossible to do without at least 4. Presumably those who advocate very picky elevation angles dial the angle up only after getting a decent survey. I think the Motorola timing receivers specifically disable any custom elevation angle mask before doing a self-survey; at least that's the way the one in my Z3801A works (although to be honest it might be the firmware in the Z3801A and not on the GPS board). > > > I talk about WWV because most ham operators have a receiver that > > > receives WWV, so it is an inexpensive and convenient way to get an > > > absolute reference, even though it is not as good as GPS. > > > > WWV is great. But again, you gotta know where you are to have > > the right time. Fluctuations in propogation will change path > > length and if you don't know the path length, it is NOT an > > absolute reference in the nanosecond or even microsecond > > range. For typical conditions the propogation delay for WWV > > cannot be estimated much better than the tens of > > microseconds; combining WWV with a better "absolute" > > clock in fact lets you study propogation by studying the > > shifting of the various ionospheric layers. (There have been > > ham operators who do this, including me!) > > > > Tim. (N3QE) > > > > OK, once again I confused "frequency standard" and "time standard". I > promise that's the last time :-) > > Yes, it's interesting to look at the WWV signal wander on the HP 3586, > compared to it's OCXO, and while listening to it on the speaker or on a > receiver. I need to plot it over a 24 hour period or so. It depends on where you are and how long you average over, but by eye I see WWV on 10 and 15 MHz going up and down by tenths of a Hz in the several minute timespan from here on the East Coast. Dave Mills has done a lot of DSP-type work on extracting the leading edge of the second and minute pulse tones from WWV in autocorrelator fashion. I think that overall, you and I need to better quantify needs before deciding on a solution as suitable :-). It's easy to overkill in time and frequency, and I think the point of this list is to do even more overkill! Personally I am so delighted with the abilities of any but the cruddiest GPS timing receivers that I see no need to tweak elevation angles. (I think I already said that I think that several consumer non-timing GPS receivers common out there are not worth crap for timing applications. Remember the SV6, it's PPS will go in and out by six milliseconds as satellites go in and out of view!! Most new stuff is better...) Tim. _______________________________________________ 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.
