For clarification, from reading the Symmetricom manual, the TS-2500 was designed as a reliable Stratum 1, 10 MHz source for communications systems. In addition to GPS, the unit monitors the internal SRI PRS-10, and other 10 MHz frequency references that are presented to it. It tracks and compares these with GPS and keeps long-term records of their behavior. In the event the GPS signal is lost, the TS-2500 generates a 10 MHz output from its internal oscillator, using the PRS-10 and any other monitored reference inputs as references, after first applying weighting factors developed from experience to determine the appropriate frequency offsets. In other words, the PRS-10 output is not utilized directly, but only as a reference. An offset is applied, based upon the retained past monitoring data of the PRS-10 frequency versus GPS. Should the unit suddenly see the GPS departing from its expected frequency, as compared with multiple other references, it might switch to the holdover-mode, after computing the required offsets from past experience. A key point is that no provision is made to control the frequency of the PRS-10; it is simply monitored and the offset from GPS continually recorded. In other words, the PRS-10 does not influence the accuracy or noise of the 10 MHz output unless the GPS reference is lost. Bruce Hunter, KG6OJI

		Attila Kinali

On Tue, 23 Jan 2018 12:28:45 -0500 Bruce Hunter via time-nuts <time-nuts@febo.com> wrote: > A key point is that no provision is made to control the frequency of the > PRS-10; it is simply monitored and the offset from GPS continually > recorded. In other words, the PRS-10 does not influence the accuracy or > noise of the 10 MHz output unless the GPS reference is lost. And it's exactly that hold-over performance that you need for a communication system. UMTS demands IIRC to keep sync between adjacent base-stations better than 10µs at any time. LTE can be as low as 1.5µs. The lawmaker used to demand 24h hold-over and most countries seem to have switched to 72h in recent years. 1.5µs over 72h requires frequency stability at the level of 5e-12. Even if you have a well characterized high-quality OCXO getting bellow 1e-10 at these time scales needs a bit of fiddling and going below 1e-11 is not easy. (unless you apply techniques that can cancel out flicker noise, but as far as I am aware of, nobody does that and I have no experimental numbers on that, just simulations) As for GPS operation, the advantage of an Rb vs a good OCXO is not that big. There is a range between a few 100s and 10-20ks where a GPSDO with an Rb reference can perform better than one with just an OCXO. But it's not going to be more than a factor of 10. Beyond..uhmm.. let me guestimate 50ks the GPS will dominate the frequency stability. I hope that clarifies the need for an Rb in such a unit. Attila Kinali -- <JaberWorky> The bad part of Zurich is where the degenerates throw DARK chocolate at you.