Expected behavior for LBE-1420 GPSDO?

[BrianL]

I tripped down into the rabbit hole of time and frequency accuracy, and am hoping the community might help me understand what I am encountering. 

I snagged a Leo Bodnar LBE-1420 GPSDO based on independent research and reading through the testing and commentary on the forum. Seems like a nice bit of kit for this adventure. Using the 1420 as a frequency reference, I have been attempting to calibrate the OCXO in an HP 5334B freq counter. I am not getting it as close as I think is possible.

With the 1420 set to 10MHz on channel 1 of the scope, triggering on channel 1, the OCXO on channel 2, I tune the OCXO until there's little to no drift on channel 2. Seems straight forward and how I've aligned oscillators for ages. Except when I think I get to no drift, suddenly there's drift. Hmm, ok maybe the crystal is settling. Align it again, seems stable and nope, suddenly more drift.

Frustrated, I happened to discover a nice LPRO-101 rubidium source in excellent condition and for an even better price, which is also not calibrated yet. Tried tuning the rubidium source using the 1420 and it's all the same story. After nearly two weeks of WTF, I finally discovered that my copy of the 1420 seems to have a random wander that also seems highly random in speed and direction.

Here's an 8x speed video of what I'm encountering. This is a smaller wander, but others have been even more severe. The LBE-1420 is the square wave, and the sine is the OCXO in the HP and is the trigger.
LBE-1420

Is this expected or known behavior of the PLL in the LBE-1420, have I ended up with a bad copy, or do I not have a clue how a GPSDO is supposed to perform?

Cheers!
Brian

[J-R]

First thought is: are you sure you always have a solid GPS signal?  If it's losing lock for some time, the TCXO might drift a little.  When lock is regained, it will slowly bring it back into alignment.

[KE5FX]

It looks like the wander in your video is on the order of +/- 30 to 50 ns, correct?  That's not too far from typical, especially if you don't have a solid GPS signal and/or the receiver isn't operating with a valid self-survey position. 

I ran a quick plot of one of Leo's GPSDOs not too long ago, and it looked pretty similar:



This was a cold start at an unsurveyed location, without any attempt at optimizing the setup.  So I suspect you're basically OK.

[J-R]

The video is 11 seconds long, so slowed down 8 times would be about 90 seconds.  I have the earlier generation dual-output GPSDO and it was very stable when adjusting my 5334B.  Movement on the scope wasn't discernable unless I placed a marker and let things sit for multiple minutes.

I still think this is a lock issue, but Leo mentioned disabling Galileo to improve performance: https://www.eevblog.com/forum/metrology/leo-bodnar-mini-precision-gps-reference-clock-teardown/msg3221996/#msg3221996

[Leo Bodnar]

Brian,
If you are getting to that level of stability, let the 1420 warm up for an hour and make sure airflow to it is restricted.
1420 is tiny and its thermal mass is very low.
The easy way to restrict airflow is to put a piece of cloth over it, say a towel or few layers of bubble wrap.
This should remove at least some phase movement that you are seeing.
Leo

[BrianL]

Thanks for the insights community!
I did just notice I was on fw 1.06 and there was a 1.07 out there. Here's signal levels shortly after a restart:


I am in a dense urban environment, so full sky visibility is a bit of a challenge. The number of satellites visibile and their levels seem sufficient to my inexperienced eye.

Thanks for commenting Leo. I did try the towel trick, noted from from another thread and although the device got uncomfortably warm for my liking, I was seeing roughly the same performance. All the gear has been up for three solid weeks and should be at full operating temperature, with or without the towel.

I hadn't seen the thread on disabling the Galileo constellation, but did come across a note on time-nuts suggesting not to use multiple constellations simultaneously. I'll try disabling Galileo and see how that goes.

If the phase noise is within expectations for a cold start but the 1420 has been up for a few weeks, is this still suggestive of poor GPS reception?

[Brociall]

thanks, it's halpful

[BrianL]

Bringing up cgps or gpsmon on the 1420 shows that my device is only receiving GPS and GLONASS constellations. Since the 1420 doesn’t natively expose the serial port on the ublox chip and the datasheet for the M10050KB is only available under NDA, I might poke around to see whether I can find the serial port so I can connect it to ucenter and investigate further. Unless someone can save me the time hunting and probing around and knows the pins off the top of their head?

I have an additional GPSDO for the rubidium clock on the way, because one can never have too many references. I’m curious to see how this one behaves. I may also take the gear for a short drive to the park to see what effect an unobstructed full sky view may or may not have on the 1420 overall.