10Mhz Reference. Which one to trust: GPSDO vs Rubidium

[4thDoctorWhoFan]

The auction site has delivered to my home two pieces of equipment to be used for a 10Mhz reference.

1) Rubidium standard = Lucent RFG-M-RB
2) GPSDO = One of the newer BG7TBL units with the blue LCD screen

For both of these units, who knows when the last time they were calibrated.
After a few days of burn in, the two units are about 50 millihertz apart.

I guess my question is which of these two units would you trust more to be closer to the exact frequency?

Thanks,
Tony

[nctnico]

The GPSDO is linked to a central time reference through the GPS system. The Rubidium clock is a freerunning clock so not accurate at all.

[KrudyZ]

The GPSDO is linked to a central time reference through the GPS system. The Rubidium clock is a freerunning clock so not accurate at all.

Agreed, assuming that the OP's reference actually locked onto the satellites 

[4thDoctorWhoFan]

The GPSDO is linked to a central time reference through the GPS system. The Rubidium clock is a freerunning clock so not accurate at all.

I was under the impression that rubidium clocks were accurate.  I thought the whole rubidium physics package inside the reference is what makes them accurate.  You know, it's a physics thing.    Just like the Cesium physics package is even more accurate.

So in your opinion, I should go with the GPSDO right?

[Mr. Scram]

As far as I understand rubidium standards need to be adjusted too. They're just very stable once calibrated. Meanwhile, GPSDO is actively maintained and adjusted.

[nctnico]

The GPSDO is linked to a central time reference through the GPS system. The Rubidium clock is a freerunning clock so not accurate at all.

I was under the impression that rubidium clocks were accurate.  I thought the whole rubidium physics package inside the reference is what makes them accurate.  You know, it's a physics thing.    Just like the Cesium physics package is even more accurate.

So in your opinion, I should go with the GPSDO right?

Definitely yes because the GPS is referenced to frequency standards. According to my sources the Rubidium clock is likely to have worse phase noise compared to the GPDSO.

[glarsson]

The GPSDO is linked to a central time reference through the GPS system.

Assuming locked to enough satellites and no software bugs. Many BG7TBL units have software bugs that makes them always generate a 10MHz that is slightly wrong.

[4thDoctorWhoFan]

The GPSDO is linked to a central time reference through the GPS system.

Assuming locked to enough satellites and no software bugs. Many BG7TBL units have software bugs that makes them always generate a 10MHz that is slightly wrong.

Well, that does NOT fill me with confidence.
Any idea which units have this problem?

[glarsson]

Well, that does NOT fill me with confidence.
Any idea which units have this problem?

It was discussed in this monster thread: https://www.eevblog.com/forum/testgear/bg7tbl-gpsdo-master-reference/
I decided to buy a GPSDO based on the Oscilloquartz Star 4 instead and haven't kept up with the BG7TBL discussion.

[nctnico]

The GPSDO is linked to a central time reference through the GPS system.

Assuming locked to enough satellites and no software bugs. Many BG7TBL units have software bugs that makes them always generate a 10MHz that is slightly wrong.

But that is a problem with some of the earlier BG7TBL units and not a GPDSO in general. AFAIK this has been fixed in later units and either way the absolute accuracy is likely to be better than the freerunning Rubidium clock.

[4thDoctorWhoFan]

Well, that does NOT fill me with confidence.
Any idea which units have this problem?

It was discussed in this monster thread: https://www.eevblog.com/forum/testgear/bg7tbl-gpsdo-master-reference/
I decided to buy a GPSDO based on the Oscilloquartz Star 4 instead and haven't kept up with the BG7TBL discussion.

I'll have to fully read that thread.
The BG7 I just purchased has one of the Trimble OCXO's installed.  It seems like a nice OCXO.

[kj7e]

The GPSDO is linked to a central time reference through the GPS system.

Assuming locked to enough satellites and no software bugs. Many BG7TBL units have software bugs that makes them always generate a 10MHz that is slightly wrong.

Well, that does NOT fill me with confidence.
Any idea which units have this problem?

Some of the earlier units, from 2015, maybe 2016.  The error was very slight, IIRC 200uHz from 10MHz low or 2E-11.

[kj7e]

The GPSDO is linked to a central time reference through the GPS system. The Rubidium clock is a freerunning clock so not accurate at all.

I was under the impression that rubidium clocks were accurate.  I thought the whole rubidium physics package inside the reference is what makes them accurate.  You know, it's a physics thing.    Just like the Cesium physics package is even more accurate.

So in your opinion, I should go with the GPSDO right?

Rubidium clocks are very stable, but only accurate if they are calibrated to be so.  GPSDO while locked is essentially locked to U.S. Naval Observatory Atomic clocks, which are kept in sync with NIST.  A good GPSDO will have much better long term accuracy and can be used to calibrate the Rubidium standard.  I would trust your GPSDO and use it as your master reference.  Even if you had one of the BG7TBL GPSDO's with the slight offset, its not enough to fret over.  200uHz is still 250 times closer to NIST than 50mHz.

[glarsson]

But that is a problem with some of the earlier BG7TBL units and not a GPDSO in general. AFAIK this has been fixed in later units and either way the absolute accuracy is likely to be better than the freerunning Rubidium clock.

I never wrote that it was a problem with every GPSDO. I only pointed out that you don't necessarily have an accurate 10MHz just because you have a GPSDO.

Btw, I don't think "freerunning" and "not accurate at all" are correct attributes to give a rubidium standard.

I have tried to compare my sa.22c to my Star 4.
Attempt 1: Frequency difference between A & B using Pendulum CNT90. Noise centered on 0.
Attempt 2: Time from A to B using Pendulum CNT90. Expected slow sawtooth but found noise.
Attempt 3: Display signals on oscilloscope. Observed two static sine waves. Phase shifted but stationary. Got bored after an hour.

The Sa.22c do have a PPS input for automatic adjustment, but I have not used it.

[4thDoctorWhoFan]

Rubidium clocks are very stable, but only accurate if they are calibrated to be so.  GPSDO while locked is essentially locked to U.S. Naval Observatory Atomic clocks, which are kept in sync with NIST.  A good GPSDO will have much better long term accuracy and can be used to calibrate the Rubidium standard.  I would trust your GPSDO and use it as your master reference.

Great, that is good to know.
I'm not sure how good my GPSDO is but I will use it for my reference.

I have one of the newer GSPDO's with the blue LCD screen that are all over the auction site.  Guess I will have to read through that other thread to see if it is any good.  It seems to work well.

[kj7e]

The Dec 2017 (LCD Screen) BG7TBL GPSDO's are good and the Trimble oven is very good.  The time constant could be optimized for even better short term performance but for everything outside a metrology lab they do just fine.  I have one also, very pleased with its performance also compared to my HP 53132A with Option 010.