Looking for a silver standard in GPSDO modules

[tkamiya]

I have several GPSDOs but they are kind of large.  I also have smaller board but they are awfully undocumented and doesn't keep a good lock.  So, here's the question:

I know Trimble Thunderbolt is the GOLD standard.  But they are becoming rare in used market, at least for reasonable prices.  New one is $1500.

What's the SILVER standard currently?  By that, I mean
. suitable as lab standard
. WELL documented
. locks well and stays locked
. have RS232 and 1 pps
. 10MHz sine output
. reasonably priced in used market
. plentiful  (I'd like to have a few)
. relatively small and modularized
. supported by Lady Heather

By the way, my antenna has a good view of entire sky with 26db gain.  Cable is of high quality and short.

There are SO MANY in market and I don't deal with too many choices well.  Can someone help me out?  What did you have good luck with?  Or what do you recommend I AVOID?

[tkamiya]

I had someone ask me what I mean by "silver standard".  Then I realized this forum is quite international.

In US, when we say "gold standard", that means it's the best there is.  An example will be, "Tektronics is a gold standard in oscilloscope".

"Silver standard" isn't really a standard idiom but I mean by it, the next best thing.  Similar idiom would be "second fiddle".  So I'm really asking for what's the next best thing to Thunderbolt GPSDO, since it is becoming rare.

[jpb]

I think the Trimble became a go-to GPSDO because it was the only one around for ages.
It also is very programmable, I understand (I don't have one), which is a big plus for people wanting to experiment.
As a GPSDO though, I think it is a little variable depending on what OCXO it had - leapsecond has lots of results.

I've been pleased with my oscilloquartz Star4+ but I have hesitated to put it forward yet as I've not managed to measure its ADEV as accurately as I'd like but I think it is pretty good.
It also allows the time constant to be changed up to 5000 seconds though it doesn't save the value (you have to reset it goes through a power cycle). The option to extend the time constant is important, I think, as many of these GPSDOs were designed more for timing than having a smooth 10MHz reference output i.e. the frequency needed to be within a range but this may mean that you get corrections which are relatively large happening relatively frequently. That is the correcting factor, the GPS 1PPS, has a large ADEV at 100 seconds and a much smaller one at 5000 seconds so corrections made every 100 seconds may be relatively noisy.

[tkamiya]

Speaking of OCXO, (I don't know what type)  I was recently trying to calibrate Rb module that contains OCXO.  I was quite surprised how fast frequency drifted.  Say I have the whole unit in an enclosure with all sides closed.  Then open the top to access the adjustment pot.  I can see in my frequency counter (which was disciplined to a third GPSDO) frequency drift almost immediately.  Reverse happens when the top is closed.

Also, just yesterday, I was playing with another GPSDO board which has single layer oven.  With case open, it took forever to heat up.  After few hours, it still hasn't reached equilibrium.  Along the way, the unit let to go of the lock few times.  With case closed, it warmed up quickly.  It is now running stably and temperature is steady.

[testpoint1]

I will test a latest Trimble board soon, chip is 96975, to discipline Pendulum GPS-12R clock.

[KE5FX]

The Thunderbolt is a good choice for most amateur users but it isn't at the top of the GPSDO pantheon.  If you had to name a "gold standard" among units commonly seen on the surplus market, it would probably be one of the HP Z3801s, tested a few years ago by Tom at leapsecond.com:



For current-production models I'd say the Jackson Labs low-noise rubidium GPSDO would give the Z3801A a run for its money as the "gold standard," if given a good antenna.  Tom's results above are quite a bit better than what I tend to see here, including when I tested the LN Rb, but I've never seen anything outperform the LN Rb.  (Disclaimer: I work with JLT but have no involvement with that product.)

The "plutonium standard" would be Ole Petter Ronningen's recent work on real-time kinematic correction for GPSDOs, which has the potential to blow away everything else by a decisive margin but is extremely complex to set up.

So, um, yeah, Thunderbolts are usually fine.  If you want something smaller with similar performance, the Jackson Labs catalog is full of excellent candidates (same disclaimer applies).

[texaspyro]

The "plutonium standard" would be Ole Petter Ronningen's recent work on real-time kinematic correction for GPSDOs, which has the potential to blow away everything else by a decisive margin but is extremely complex to set up.

The latest version of Lady Heather has a lot of the features needed to implement this.  It calculates solid earth tides.  it gets L1/L2/etc obervations from the GPS receiver. It calculates sawtooth corrected PPS measurements.  It can drive an external DAC and read in data from an "ADC" or external processing program.  Alas, it does not currently take in IGS corrections,  but it could output whatever data it has to an external program.

[Theboel]

The "plutonium standard" would be Ole Petter Ronningen's recent work on real-time kinematic correction for GPSDOs, which has the potential to blow away everything else by a decisive margin but is extremely complex to set up.

The problem for Ole plutonium is he use his AHM and geodetik grade GPS receiver for reference so I do not know is RTK really working for time / freq based or AHM made the result go to plutonium level 

[jpb]

The "plutonium standard" would be Ole Petter Ronningen's recent work on real-time kinematic correction for GPSDOs, which has the potential to blow away everything else by a decisive margin but is extremely complex to set up.

Thank you for linking this, very interesting reading.
The thought that struck me though, is in order to need to do such careful correction you need a very good OCXO in the first place. If you just have an averagely good OCXO with ADEV around the 1 or 2 x 10^-12 then you can't take advantage of the sophistication anyway as error is dominated by the OCXO.

What I've thought about is combining multiple GPS receivers with different antenna so they are looking at different satellites and making the assumption that any error will be in the form of an extra delay so take the one with the least delay. You'd still need to correct for sawtooth etc. At the very least you could get an error indication by checking the apparent difference in position between the antenna, if it matched the known value then signal is good if there is a significant difference then that measurement would be discarded.

But much of it comes down to how good the oscillator being disciplined is - presumably if you could afford a hydrogen maser then the GPS side of things could be very simple as averaging could be over a few days.

[Theboel]

I think the problem is more complicated than just use external correction to have better stability in a GPSDO.
First, Ole use geodetik gps receiver TRIMBLE NETRS to be precise and this receiver has an external clock reference port and Ole use his active hydrogen maser for this reference.
second, The Trimble netRS use dual freq and as far I understand dual freq GPS receiver has better accuracy in term of position because its can minimize any disturbance in atmosphere.

the fact that Ole use better reference in the GPS receiver compare than most of us (especially me) and use better GPS receiver (because I do not have yet dual freq GPS receiver). made me wondering if any body can tell me :
"is there any correlation between better accuracy in position with more stable 1PPS output ? what I mean a real measurement not just assumption"   
 another thing bother me is Ole do not show what his GPSDO compare to is he have second AHM or maybe a BVA as it short enough tau

with two HP 58503A I can get 10E-13 but in very long tau about 10E5 second for 10E-14 I am not sure
sorry for my broken English 

[jpb]

I think the problem is more complicated than just use external correction to have better stability in a GPSDO.
First, Ole use geodetik gps receiver TRIMBLE NETRS to be precise and this receiver has an external clock reference port and Ole use his active hydrogen maser for this reference.
second, The Trimble netRS use dual freq and as far I understand dual freq GPS receiver has better accuracy in term of position because its can minimize any disturbance in atmosphere.

the fact that Ole use better reference in the GPS receiver compare than most of us (especially me) and use better GPS receiver (because I do not have yet dual freq GPS receiver). made me wondering if any body can tell me :
"is there any correlation between better accuracy in position with more stable 1PPS output ? what I mean a real measurement not just assumption"   
 another thing bother me is Ole do not show what his GPSDO compare to is he have second AHM or maybe a BVA as it short enough tau

with two HP 58503A I can get 10E-13 but in very long tau about 10E5 second for 10E-14 I am not sure
sorry for my broken English

I'm still trying to get my ADEV measurements better at low taus, but the attached plot shows a samsung GPSDO measured against a Star4+ GPSDO. The Samsung is at whatever the default is, I'd guess around 200 seconds but the Star4 is set to 5000 seconds. I did a very long run of over a million seconds (i.e. nearly 2 weeks) and you can see that it gets down to 10^-13 in 10^5 secs which is in agreement with your results and is probably determined by the GPS system so is probably true for any GPSDO i.e. if tau > 10^5 then all GPSDOs probably look similar - they just differ at lower taus.

NB the plot is only accurate above taus of around 100 seconds - possibly down to 10 seconds or so but definitely not below that.

[Theboel]

So You nedd to talk to Mr KE5FX aka Mr John miles for help
this is one of his creation with his magic wand

http://www.ke5fx.com/tbolt.htm