I wish I could give some numbers. Sadly, I don't know how to calculate them nor do I have the equipment to do empirical measurements of that level of accuracy.
But I could do an arm-waving estimate. Suppose that the delayed signal component is in phase with the direct signal, but 20 dB down, and is delayed by 100 nsec. Then I could see how the apparent centroid of a given chip could be displaced by about 10 nsec, which would be enough to cause complaint from many time-nuts. Personally I'm not into time keeping of that order of magnitude, but I'd like my GPS- disciplined oscillator to be phase stable to within 10's of ps over about a 1 minute time frame. At present I cannot seem to do this well unless I let my Rb free-run. Dana On Sat, Jun 6, 2020 at 8:43 AM Tom Holmes <thol...@woh.rr.com> wrote: > No caveats required. > > Give some numbers on how loud that MP signal has to be to cause a problem. > My original example of line losses for a relatively short cable still > suggests to me that it is a minimal problem. If the S/N of a satellite is > poor, the receiver algorithm is likely to not use it if there are better > choices, and there often will be. > > Tom Holmes, N8ZM > > -----Original Message----- > From: time-nuts <time-nuts-boun...@lists.febo.com> On Behalf Of Dana > Whitlow > Sent: Saturday, June 06, 2020 8:52 AM > To: Discussion of precise time and frequency measurement < > time-nuts@lists.febo.com> > Subject: Re: [time-nuts] ThunderBolt question > > Basically true, Tom, but with some caveats: > > The GPS system was originally designed, as you say, to provide > positions and velocity > information to military users in the field. Errors of a few meters were of > little consequence. > > But now, we have the time-nuts, surveyors, etc, who are decidedly pushing > the envelope > of GPS accuracy. Some users want sub-nanosecond timing accuracy, surveyors > want > mm-level position accuracy, etc, and doing these things indeed requires > extreme attention > to detail. > > As I understand it, multipath components that are delayed by about a few > chips or more > basically appear as a wee bit of added noise and have negligible > consequence unless > they are comparable to the direct signal in power. However, MP components > whose > delay is less than about a chip time have the effect of tugging the > apparent temporal > location of the direct signal to some new value. Note that this is in the > range of practical > round trip cable delays in at least some installations. At Arecibo, for > example, the one-way > cable lengths from GPS antennas on the roof to their respective receivers > was about > 200 ft. Round trip was thus 400 ft physical, hence up to about 600 ft > depending on > what the cables were using for the dielectric. That's about 0.6 chip > length for the C/A > code GPS signal. > > Another factor which will influence the error introduced will be the RF > phase relationship > between the direct and delayed signal component. > > Gee, this stuff gets complicated ... > > Dana > > > On Sat, Jun 6, 2020 at 7:30 AM Tom Holmes <thol...@woh.rr.com> wrote: > > > Dana... > > The question that comes to mind is just how much effect a weak a long > > delayed reflection will have on overall system performance since it will > > only matter to SV’s with poor S/N. The modulation scene which allows all > > the SV's to transmit on the same frequency has to be pretty robust in the > > face of both widely varying signal strengths and multiple signals > arriving > > at different times. It’s a similar scheme to CDMA cell phones, which > > operate in a much more difficult environment with regard to signal > > strengths, multi-path, and number of on channel signals. And those work > > amazingly well. > > > > Further, I am led to believe that once you have enough SV’s in view to > > get a good set of ‘readings’, ionospheric effects are the limiting factor > > until you go to a multi-band receiver. > > > > Yes, to wring the last ounce of performance out of GPS takes attention to > > the details, but don’t lose sight of how it was designed to work for > users > > in less than optimum (military field operations) in the first place. > > From Tom Holmes, N8ZM > > > > > On Jun 6, 2020, at 7:14 AM, Dana Whitlow <k8yumdoo...@gmail.com> > wrote: > > > > > > It's one thing to maintain lock in a multipath environment, quite > > another > > > thing > > > to get "full" accuracy of GPS measurements of PVT. > > > > > > An interesting difference between my scenario of poorly matched > > impedances > > > and "ordinary" multipath is this: In the poor matching scenario, all > the > > > received > > > signals will be impaired identically, while in the ordinary multipath > > > scenario, > > > signals from different satellites will suffer different (and > > time-varying) > > > multipath > > > impairments. I'm not at all sure what effect this difference will have > > on > > > final > > > outcome, but my gut feel is that the case where all signals are > impaired > > > identically > > > could lead to worse effects. > > > > > > Dana > > > > > > > > >> On Fri, Jun 5, 2020 at 8:43 PM Tom Holmes <thol...@woh.rr.com> wrote: > > >> > > >> Dana... > > >> > > >> I think that you are neglecting two important mitigating factors. > > >> > > >> 1. the cable loss at 1575MHz, even for a 25' run of RG-6, reduces > those > > >> reflections quite a lot from one end to the other. It amounts to 2 > > - 3 > > >> dB in 25', depending on cable quality. > > >> > > >> 2. a 1.5:1 SWR is not a very big reflection to begin with, on the > order > > of > > >> 20% of the incident power, about 7 dB. I am rounding a lot here just > to > > >> keep the math easy...for me. > > >> > > >> By the time a reflection has made the round trip from the receiver > back > > to > > >> the antenna and them back to the receiver, which is how the delay > would > > >> have > > >> to manifest itself, it will be down at least 15 dB from its original > > self, > > >> and probably more. Given the coding of GPS signals which allows > several > > >> satellites to share a common frequency band, that is not going to be > > much > > >> of > > >> a problem. And if only one end of the path actually is 75 ohms, then > > there > > >> won't be a delayed signal. > > >> > > >> Tom Holmes, N8ZM > > >> > > >> -----Original Message----- > > >> From: time-nuts <time-nuts-boun...@lists.febo.com> On Behalf Of Dana > > >> Whitlow > > >> Sent: Friday, June 05, 2020 9:01 PM > > >> To: Taka Kamiya <tkami...@yahoo.com>; Discussion of precise time and > > >> frequency measurement <time-nuts@lists.febo.com> > > >> Subject: Re: [time-nuts] ThunderBolt question > > >> > > >> I'd like to point out that mismatches at the ends of an antenna cable > > *can* > > >> cause trouble. When both ends are mismatched, each bit of detail in > the > > >> signal > > >> gets partially reflected back and forth, each time delayed by the > round > > >> trip propagation > > >> delay in the cable, and so you have something like multipath going on. > > >> Fortunately the > > >> successive reflections get weaker with time, generally quite rapidly. > > >> Since many > > >> GPS users seem very concerned about multipath resulting from poor > > antenna > > >> placement, > > >> I think this factor should be considered as well and not just get > swept > > >> under the rug. > > >> > > >> The amplitude of the "multipath" resulting from cable mismatches > > depends on > > >> the product > > >> of the voltage reflection coefficients at the two ends of the cable. > If > > >> either end is perfectly > > >> matched, then the quality of the match at the other end is not > > significant > > >> vis-a-vis apparent > > >> multipath problems and only affects transmission loss. > > >> > > >> But when there is a mismatch on both ends, then the length of the > cable > > >> comes into play > > >> as well. A longer cable means more delay between successive > > reflections, > > >> which is just > > >> like multipath involving longer delays between the direct and the > > reflected > > >> signals. > > >> > > >> Cheers, > > >> > > >> Dana (K8YUM) > > >> > > >> On Fri, Jun 5, 2020 at 7:13 PM Taka Kamiya via time-nuts < > > >> time-nuts@lists.febo.com> wrote: > > >> > > >>> 50 ohm / 75 ohm question is really irrelevant in this kind of thing. > > >>> Trmble itself says in manual, not to be concerned with this apparent > > >>> mismatch. > > >>> In my particular case, I have a home lab standard and existing > > system. I > > >>> have an antenna and network of distribution amplifiers. They are all > > 50 > > >>> ohms and N connectors. Some ports have BNC adapters attached. I > have > > >>> pretty much standardized everything to SMA, N, or BNC. > > >>> > > >>> I boxed a power supply, T-bolt, and buffer amp in a metal case. I > > bought > > >>> a short cable (RG58) that goes from F to BNC. On back of the case, I > > >> have > > >>> BNC to N adapter. I also have a few adapters that goes from F to BNC > > for > > >>> the test bench. It really doesn't matter what you use, as long as it > > >> makes > > >>> a solid connection. > > >>> > > >>> Advantage of F connectors and RG6 are, cheap, abundant, and low loss > > for > > >>> the size. Advantage of having house standard is, less adapters and > > less > > >>> headache..... > > >>> > > >>> --------------------------------------- > > >>> (Mr.) Taka Kamiya > > >>> KB4EMF / ex JF2DKG > > >>> > > >>> > > >>> On Friday, June 5, 2020, 7:22:33 PM EDT, Robert DiRosario < > > >>> ka3...@comcast.net> wrote: > > >>> > > >>> I have a Trimble ThunderBolt GPSDO that I just received. It has an F > > >>> connector for the antenna input, and BNC connectors for the 1 pps and > > 10 > > >>> MHz outputs. Is the receiver input impedance really 75 Ohms, or is it > > 50 > > >>> Ohms and they just used the F connector to distinguish it from the > > >>> others? What do people do, just use a 50 Ohm antenna? > > >>> > > >>> Thanks > > >>> > > >>> Robert DiRosario > > >>> > > >>> KA3ZYX > > >>> > > >>> _______________________________________________ > > >>> time-nuts mailing list -- time-nuts@lists.febo.com > > >>> To unsubscribe, go to > > >>> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > >>> and follow the instructions there. > > >>> > > >>> _______________________________________________ > > >>> time-nuts mailing list -- time-nuts@lists.febo.com > > >>> To unsubscribe, go to > > >>> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > >>> and follow the instructions there. > > >>> > > >> _______________________________________________ > > >> time-nuts mailing list -- time-nuts@lists.febo.com > > >> To unsubscribe, go to > > >> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > >> and follow the instructions there. > > >> > > >> > > >> > > >> _______________________________________________ > > >> time-nuts mailing list -- time-nuts@lists.febo.com > > >> To unsubscribe, go to > > >> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > >> and follow the instructions there. > > >> > > > _______________________________________________ > > > time-nuts mailing list -- time-nuts@lists.febo.com > > > To unsubscribe, go to > > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > > and follow the instructions there. > > > > > > _______________________________________________ > > time-nuts mailing list -- time-nuts@lists.febo.com > > To unsubscribe, go to > > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > > and follow the instructions there. > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.