Yes, I'm sure. I did check the cabling. :) If I was somehow measuring the Tbolt or 4065A against itself, there wouldn't be any frequency offset. The Oscilloquartz 3210 (which appears to be an OEM'd 4065A) is spec'ed at 3E-13 @ 100K seconds. The 4065C is even better at 8.5E-14 with a noise floor of 5E-14. My data run is continuing. The end of the graph is flopping around as usual, but it's now showing Total Deviation of about 1.4E-13 @ 100K seconds with a total of 226 800 readings at 1 second intervals. Frequency offset error is still 6E-13. Autocorrelation of the results shows diurnal spikes at approximately 86 160 and 172 310 seconds (i.e. one and two sidereal days). Ed On 11/14/2014 11:00 AM, Azelio Boriani <azelio.boriani@gmail.com> wrote: > Are you sure you are measuring the 4065A against the Tbolt? TBolt > feeding the start and 4065A feeding the stop and a counter in time > interval mode? > > On Fri, Nov 14, 2014 at 3:26 PM, Ed Palmer <ed_palmer@sasktel.net> wrote: >> FYI, I found a source for the PLCC44 version of the STEL-1173 that's used in >> some Datum FTS Cesium standards. I got mine from here: >> >> http://www.aliexpress.com/item/Free-shipping-10PCS-STEL-1173-CM-PLCC/2023682701.html >> >> I left positive feedback, not sure why it doesn't show. >> >> Shipping was very slow, but the chips appear to be NOS. There are no >> scratches on the leads which suggests that they were never installed in a >> socket. I was having trouble with one so I decided to sacrifice it to see >> what was inside it. The die is labelled 'ORBIT B-100'. When I chopped out >> the dead chip from my FTS-4065A, the bottom was stamped with the word ORBIT. >> ORBIT Semiconductor works on FPGA to ASIC conversions. It looks like >> Stanford used Orbit as a fab house. >> >> I'm currently measuring my 4065A against my Tbolt. After about 37 hours, >> I'm still not seeing anything that I can attribute to the 4065A. All I'm >> seeing is the Tbolt and the typical response that every GPSDO gives. I'm >> looking at Total Deviation values of around 1E-13 at 100Kseconds. Frequency >> offset error appears to be about 6E-13. >> >> All diagnostic status values are within spec. I had to adjust the signal >> level from the tube - it was too HIGH! >> >> I decided to solder in a socket so that I won't have to go through any pain >> if this chip dies in the future - and I now have spares! >> >> Ed >>

Ed, On 11/15/2014 04:38 PM, Ed Palmer wrote: > Yes, I'm sure. I did check the cabling. :) If I was somehow measuring > the Tbolt or 4065A against itself, there wouldn't be any frequency offset. > > The Oscilloquartz 3210 (which appears to be an OEM'd 4065A) is spec'ed > at 3E-13 @ 100K seconds. The 4065C is even better at 8.5E-14 with a > noise floor of 5E-14. My OSA 3210 is not the same core as 4065A. > My data run is continuing. The end of the graph is flopping around as > usual, but it's now showing Total Deviation of about 1.4E-13 @ 100K > seconds with a total of 226 800 readings at 1 second intervals. > Frequency offset error is still 6E-13. Autocorrelation of the results > shows diurnal spikes at approximately 86 160 and 172 310 seconds (i.e. > one and two sidereal days). Most likely thermal sensitivity. Got myself some STEL 1173 in PLCC 44 from that source. Thanks for sharing link! Cheers, Magnus

Hello! On 18/12/2014 5:37, Magnus Danielson wrote: > Ed, > > On 11/15/2014 04:38 PM, Ed Palmer wrote: >> Yes, I'm sure. I did check the cabling. :) If I was somehow measuring >> the Tbolt or 4065A against itself, there wouldn't be any frequency >> offset. >> >> The Oscilloquartz 3210 (which appears to be an OEM'd 4065A) is spec'ed >> at 3E-13 @ 100K seconds. The 4065C is even better at 8.5E-14 with a >> noise floor of 5E-14. > > My OSA 3210 is not the same core as 4065A. > The Oscilloquartz that has the same core as the 4065A is the 3120 (EUDICS 3120 they call...), not 3210. I'm now doing a comparison of a Lucent KS-24361 against it, and have not yet enough data for 100K seconds although it is showing an ADEV there around 3e-13. Regards, Javier