Mastering Precision 2: 10MHz Reference Testing & Analysis (BG7TBL GPSDO)

EEVblog forum link to the post containing the Excel Document - www.eevblog.com/forum/metrolo...
A huge thank you again for all that continue to Join us in the lab for repairs and alignments. The channel would not exist with out all of you so thank you again for dropping by.
Ways to support the channel:
Patreon - / zenwizard_studios
Paypal - [e-mail]
In this video we take a look at an offering from the great auction site in the sky for a GPSDO. I have been asked a few time what I think about GPS units and given that we have a lab available to us here and gear that can DO the testing It was time to get it going. So here is the first GPSDO that we are testing! I am starting to build a table for some stability references to aid in picking the GPSDO that is right for you!
** Repair exposes the Tech to mains electric and should only be done by trained personnel. Errors can be fatal. **

[@lmamakos]

Battery could be for the GPS receiver module so it can preserve the GPS/GNSS satellite almanac or ephemeris data to improve time-to-first-fix times.

[@foxale08]

This is correct. If the battery is removed the gps takes much longer to start up.

[@delray-qj3qb]

very interesting video, thank you for making it. do you have some explanations for dummies on what Allen deviation is and how to measure it properly ? in your examples, I see you are changing gate time. what is that about changing the gate time for Allen deviation?

[@ZenwizardStudios]

Allen Deviation is a really good way to sum together all the “Noise sources” and roll them in to a number for easy comparison. The lower the Allen Deviation the better the oscillator “typically”
The gate time is a way to control the sample window of interest. Some equipment has a long measurement time and the more the oscillator moves around the less stable the measurement.
Changing the gate time longer also gets some of the slowing drifting from peak to peak. A gate time of 1ms will be far less sensitive to the oscillator walking over the course of several minutes then one of 100 to 200 seconds. At the longest gate time the measurements took 2.5 hours to complete
Where as a longer gate time will average out the high frequency noise and make an oscillator look more stable then it really is.
Hence why I built the Excel sheet with several gate times, but ALL gate times were sampled 100 times. So there is a sampling of high speed as well as low speed stability. This comes in to play if measurements of interest are 1 to 10 seconds or is something is being monitored for weeks, months, years.

[@AndrewandDebbie]

Do you also have the Leo Bodnar GPS reference? Wondering how they compare. I'm in the UK and should probalby just get LBE-1420 . . .

[@ZenwizardStudios]

I have a dual port leo that I will be putting through the same testing as a reference.

[@andrewfischer1356]

@@ZenwizardStudios Look forward to the video.

[@BobDarlington]

Check the time-nut archives. The Leo Bodnar units test pretty poorly compared to something like a Trimble Thunderbolt. It may meet your needs tho.

[@ZenwizardStudios]

My testing agrees to the Time-nuts results (video will be out shortly) I was kind of curious if there was an abnormality but this does tend to confirm the results. One critical note on the leo is is NOT a fixed output. My is adjustable from 10 Hz to 400 Mhz with 2 ports. With that range of adjustment I would expect there to be some stability degradation when tested with a fixed and tuned for stability 10Mhz.

[@raymiller5738]

I have the older Leo Bodnar single output unit and added some heat sinks to the unit, both internal and external to lower the localised hot spots.
If the Leo unit stabalizes over 60seconds and the GPS signal needs to be averaged over 60minutes to gain what the real accuracty is as it cycles to reduce its precission. What precession does the Leo achieve in the 60sec stabalisation time? I'd be interested in the non GPS locked stability figures at 10Mhz. thanks for the testing videos. cheers from Australia

[@CraigPetersen12f36b]

I have one of these units as well with date code 2020-06-10. I have compared mine against my Rubidium standard and I'm getting similar results to yours using a HP 5371A Frequency and Time Interval Analyzer. I have also characterized the unit using Time Lab, and for the money, it's certainly a very cost effective option than using a Cesium standard. I have had no issues monitoring the unit with Lady Heather.

[@BobDarlington]

Everything is a cost effective option vs cesium standard. ;-)

[@cashfdy2629]

Measuring GPSDO requires the Allan variance