[time-nuts] GPSDO that isn't GPS locked for large time scales?

[AC0XU (Jim)]

Time Nuts-

I recently purchased (Ebay) a GPSDO from an overseas manufacturer. This is a new item and the Ebay description seemed quite encouraging. Having had one in the lab for several weeks what I observe is that:

1) ADEV for short to medium durations (Tau <1,000 sec) is comparable to or better than other GPSDOs in my lab. Not as good as the best undisciplined oscillators in my collection, as expected, but no problem here.

2) ADEV for longer durations (Tau>10,000 sec) is considerably worse than other GPSDOs and Cesiums in my lab. I measure  well over 1E-11 at tau=100,000 sec for the new unit. This is my problem with it. As far as I can see, it  behaves like it is not actually locked to GPS.

I queried the vendor about the apparent lack of synchronization with GPS and this was his response.

"I understand everything. When constructing this device, I was guided by the need to get a good 100 MHz and 10 MHz signal, which will be maintained with an accuracy better than 0.1 ppb throughout its life. I used the GPS signal to correct the frequency of the main OCXO as the elements ages. I tried to ensure that the frequent loss of GPS signal and its quality did not affect the parameters of OCXO itself. For this reason, the adaptive algorithm is very delicate in the final phase.
The frequency correction program causes it to stop regulating OCXO in some cases.
This happens when:

1. OCXO will achieve accuracy of 0.01 ppb.
2. No GPS signal.

The control system is temporarily frozen when the frequency accuracy of 0.01 ppb is reached. Therefore, ADEV for T> 1000 seconds resembles the typical behavior of OCXO during free operation. These parameters do not change when the GPS signal appears suddenly after a long absence. When constructing the device, it was assumed that all stability will be dominated by OCXO, and a value better than 0.01 ppb is not needed in telecommunications applications. I would add that the value of 1E-13 has never been a goal to achieve, especially due to the additional costs. Maybe if I gave up the 100 MHz section, I could do it for the price."

It surprised and dismayed me that large-tau behavior is so poor in a GPSDO. The several other GPSDOs in my lab all have ADEV around 1E-13 at 100,000 secs (based on a mass assessment of the GPSDOs and cesium oscillators).  It makes me wonder what the point was in adding the GPS receiver to this device. Is there something that I am not understanding?

Thanks in advance for your comments - Jim