OCXO's and friends

[Lemonizer]

Hello everybody !!

I'm getting more and more interested by OCXO's, and GPSDO's in particular.

I read multiple threads about a few known brands like Morion, Oscilloquartz, ISO-Temp, etc, and I'm quite interested to know which ones are in the top 3, stability/phase noise/overall quality-wise !
I'm asking the same question for GPSDO's, and GPS modules too !

I know that not all OCXO's even from the same brand/model are equal, some are better than others (its the same for every production product I guess), but  I'm sure this is like wine, some years must be better than others ! And some very knowledgeable and qualified peeps here must know things that aren't in books !

I was interested by purchasing an Oscilloquartz 8663-XS, a Morion MV89A, and a few others cheaper like Bliley, NDK or ISOTEMP to measure them over days/weeks, and see which one is a good performer. Do you have anything to say on those models ? They seem to be highly regarded here ! (the morion seems to have issues !).

I'm soon getting a BG7TBL GPSDO from 2018-01, his distribution amplifier (to connect the GPSDO to my HP 5345A, and Siglent 1032X) and the FA-2 counter (that I will also use with the GPSDO as his reference).

I'm wondering what people think about that FA-2 counter, compared to high-end counters like Agilent, Pendulum's ones, just accuracy wise (once used with a well stabilized GPSDO, with a good signal ofc). I mean, it seems to be pretty nice, it got a huge resolution, and its mostly the external reference that does all the work accuracy wise, right ?
I thought it would be nice to later replace its OCXO with a better one (once I got my hands on a few that were measured during days, prior to that) and power it with a linear supply, instead of a wall-wart. Am I right ?

If it was in a nicer box, with a big brand sticker slapped on it, would people pay much more for it ? (the 1 million dollar question !  )

I'm still pretty new to this, I'm trying to read as much as I can to avoid asking redundant questions, I would greatly appreciate if any veteran time-nut could help me out on those ! 

Have a nice evening/day ! 

Lemon

[KE5FX]

A few random thoughts --

- Right now the OCXOs with the best short-term stability are probably the high-end models from Rakon, intended as replacements for the Oscilloquartz BVA models.  They are about 5x-10x better than most other high-quality oscillators.  You probably don't need one, but if you do, the price will run well into five figures in either Euros or USD.

- In terms of value for money, it's hard to beat a well-aged Morion or C-MAC DOCXO.  These are often sold on eBay out of Hong Kong or Shenzhen, having been removed from decommissioned hardware by someone with a blowtorch.  So you should plan on buying 2 or 3 to get a working one, and a dozen or more to get an outstanding one.  At the prices they sell for, this value proposition is quite reasonable.

- The BG7TBL GPSDO has good overall performance but if you are interested in amateur time and frequency metrology (as opposed to simply wanting a good frequency standard for the ham shack), you will want to make sure he has fixed the bug that was documented earlier. 

- No experience with the FA-2 personally, but I've been hearing good things about it.  Customers would pay more for it if it were in a nicer box with a Keysight logo, but unless you're being paid for your own work or trying to impress visiting time nuts, a fancy brand name adds little value to a counter.  Unlike an OCXO or a GPSDO, a counter maintains the accuracy and stability of its reference, as you've noted.  If it's not working properly, it probably won't work at all.  There is unlikely to be any benefit to messing with the power supply, in any event.

[Lemonizer]

Thanks for your answer, I appreciate that !

Those Rakon's are sure pretty neat, but yeah, out of reach sadly  !

About the C-MAC, those for example ? https://www.ebay.fr/itm/OCXO-Oscillator-C-MAC-STP2187-10-Mhz-Double-Oven-Sine-wave-12V-EFC-Freeshipping/252616346274?hash=item3ad11b8aa2:g:qaAAAOSwSwNbNDIj

And for Morion's, the MV89A is a good option ? Looking at their price, I'd rather directly get a few of the 'best', if I just need to spend a few bucks more.

Thanks for pointing out that bug on the BG7TBL GPSDO, hopefully its fixed by now, we will see !

About the GPS part of a GPSDO, as long as it locks correctly and its getting a strong signal, do I benefit from getting a 'better' one, accuracy wise ? Will those lonnnng string of digits on my FA-2 counter (with the GPSDO as a ref) will be more accurate, to put it simply ?

Lemon

[KE5FX]

About the C-MAC, those for example ? https://www.ebay.fr/itm/OCXO-Oscillator-C-MAC-STP2187-10-Mhz-Double-Oven-Sine-wave-12V-EFC-Freeshipping/252616346274?hash=item3ad11b8aa2:g:qaAAAOSwSwNbNDIj

And for Morion's, the MV89A is a good option ? Looking at their price, I'd rather directly get a few of the 'best', if I just need to spend a few bucks more.

Yes, but as others have noted you may have to go through several to get a good one.  I don't know of anyone selling cherry-picked MV89As.  If purchased new from Morion, they will probably cost multiple thousands of dollars, and you still won't have any guarantee of getting an unusually good one.

For C-MAC, I've only bought STP-2055Bs, so can't comment about the  others.  The best OCXO I have in-house is one of those.

About the GPS part of a GPSDO, as long as it locks correctly and its getting a strong signal, do I benefit from getting a 'better' one, accuracy wise ? Will those lonnnng string of digits on my FA-2 counter (with the GPSDO as a ref) will be more accurate, to put it simply ?

Yes, that's the idea.

[Lemonizer]

Ok thanks, I'll be looking for 2055B's, it seems theres none on ebay right now. What is a good price for those ? (non cherry picked, just 'working' )

About GPS modules, what are the best ones around ? How are ranked those Ublox that comes in BG7TBL's GPSDO's ?

[Lemonizer]

Ho and also, as long as you are using a good GPSDO or any other good freq reference with the counter, is there any benefits of getting an older Agilent or other high quality counter, over that FA-2 little boi ? I mean, if you just need the resolution, its fine right ? The accuracy will be the same, as long as you use the same reference ?

[Cerebus]

One thing to watch out for if you're buying 2nd hand OCXOs is not to be fooled by the retrace when they are put back into service. These have been sitting around unpowered for a long time. When you next power them up again they will drift like buggery, but it goes away after a few hours/days. Basically they have been through the ageing process, but while powered off 'forget' this, then when powered up they rapidly retrace the ageing that originally took years over the space of a new hours or days.

This also means that you can't tell whether you've got a 'good un' or not until you've left them to soak test for a few days.

[james_s]

I think it depends on just how good you need it to be too. I have a couple of Trimble double oven oscillators and a Datum single oven, the Trimbles are a bit beat up but all of these seem to work just fine. Even if what you get is not the hand picked cream of the crop it's still likely to be dramatically better than a common non-oven oscillator. Buy a few, play around with them, the surplus ones are cheap enough that you can pick up several and they're kinda fun to play with. One that is a dud as an ultra stable reference might still be a nice upgrade for that old frequency counter you have in the closet.

[Gyro]

I've heard of the occasional case where an OCXO has drifted outside the correction range of the frequency adjust pin. A few of the older OCXOs have a screw on the side, which, when removed, gives access to a trimmer on the oscillator inside the oven. This can maintain the OCXO within adjustment pin range for very long periods, eg.  https://www.ebay.co.uk/itm/PIEZO-2900082-47-10MHz-12V-OCXO-Crystal-Oscillator-replace-Trimble-37265-227-LW/322498394716  (mine is a Datum one).

I don't know how phase noise etc. compare with more modern, small body types, but it is an option. Alternatively you can just buy another OCXO if yours drifts out of range.

[Lemonizer]

Thanks a lot for the replies !

I heard about that adjust screw, and I fear to buy some second hand OCXO's that are out of range and can't be adjusted anymore !
But well, I found some sellers that guarantee them, I will order a Morion MV89A, and some others to test them, and maybe I'll find a better one to swap it with the one in the BG7TBL GPSDO !

[jpb]

The Star4+ GPSDO has a good OCXO and it is a well designed unit (commercial board re-housed I think) and it talks to Lady Heather nicely.

I have a couple of them so I measure one against the other for ADEV. If you set the time constant to a longer value (I've changed mine to 3600 secs) using Lady Heather then I found that you can flatten the peak in the ADEV curve to keep it below 5 x 10^-12.

They are a little bit more expensive than BG7TBL at around $150 e.g.:
https://www.ebay.com/itm/10Mhz-GPSDO-OSCILLOQUARTZ-OSA-OEM-GPS-GPSDO-10MHz-1PPS-STAR-GPS-Clock/262861459973

Occasionally they are auctioned rather than buy-it-now, I got my second one this way and it was less than $100.

I've screwed mine to aluminium blocks to even out temperature variation - I don't know if this does a lot but I feel that it does.

[Lemonizer]

Hoo I think I saw those yes ! I was tempted at first but ended up choosing the BG7TBL, as it was closer (germany, I live in france).

Do you think, accuracy-wise, that its much better than the BG7TBL ? If I let it run 24/7, and let's say they both get the same strong signal.

[Lemonizer]

What do you think of this one ? https://www.ebay.fr/itm/Oscilloquartz-Star-4-GPS-Empfanger-10MHz-1PPS-GPS-disziplinierte-Uhr-GPSDO/292905940612?hash=item44328df684:g:vfEAAOSwfsdcMdg3

[jpb]

What do you think of this one ? https://www.ebay.fr/itm/Oscilloquartz-Star-4-GPS-Empfanger-10MHz-1PPS-GPS-disziplinierte-Uhr-GPSDO/292905940612?hash=item44328df684:g:vfEAAOSwfsdcMdg3

That looks to be the same one.
I think they are all based around the same Oscilloquartz board:
http://www.unitest.com/pdf/osa-star3-4.pdf

[jpb]

Hoo I think I saw those yes ! I was tempted at first but ended up choosing the BG7TBL, as it was closer (germany, I live in france).

Do you think, accuracy-wise, that its much better than the BG7TBL ? If I let it run 24/7, and let's say they both get the same strong signal.

I don't have the BG7TBL to compare. Initially the BG7TBL had a slight software bug by which its frequency was a little offset but I think that is now fixed.
The advantage of the Star4+ is you can change the timing constant - the default is 200 secs which I think is far too short.
I showed some results for this in another thread:
https://www.eevblog.com/forum/metrology/some-adev-measurements-for-star4-gpsdo-at-different-time-constants/
Since making those measurements I've improved my measurement system a bit but am still doing measurements so don't have much further to report.
Here are some early measurements on BG7TBL (by ke5fx not by me):
http://www.ke5fx.com/gpscomp.htm

[Lemonizer]

Thanks for the links ! I'm trying to understand how to read them, but there's something I don't get :

It seems that the longer you let the GPSDO stabilize, the more accurate it is, and at approx 50ks or so, it seems to be in the 1E-12 terittory, if I leave it even longer (for weeks) can it hit 1E-13 or better ? Or there's diminishing returns (I bet there is, but when approximatively ?) ?

[Lemonizer]

I'm also thinking at another thing, is it an issue if I use one of those extension cables https://www.amazon.fr/femelle-C%C3%A2ble-coaxial-dextension-32-80ft/dp/B079YT7NM3/ref=sr_1_6?__mk_fr_FR=%C3%85M%C3%85%C5%BD%C3%95%C3%91&keywords=cable+sma+extension&qid=1585583101&s=electronics&sr=1-6 with the puck antenna that comes with the gpsdo ? Will it decrease accuracy ? I have a clear sky view but 5 meters of cable might be a tad short. Also, I'm looking around to get a bigger antenna, but I have trouble to choose one, I don't really know what to look for (I read that some antenna can get fried by the GPSDO, as there's a voltage mismatch, or something like this) and where, on ebay it seems theres not a lot of options !

[jpb]

Thanks for the links ! I'm trying to understand how to read them, but there's something I don't get :

It seems that the longer you let the GPSDO stabilize, the more accurate it is, and at approx 50ks or so, it seems to be in the 1E-12 terittory, if I leave it even longer (for weeks) can it hit 1E-13 or better ? Or there's diminishing returns (I bet there is, but when approximatively ?) ?

There isn't a simple answer. First, the GPSDO gets more accurate timing results when it knows its position more precisely, most of these GPSDOs get their position to start with by doing a self-survey i.e. they get the GPS position measured over a longish time period (typically several days) and average it. So results will be more accurate after this bedding in period.
Secondly, GPS time signals (the 1pps) is pretty jittery but it averages over a long period to a very accurate value. Any one pulse might be 10 to 50 nsecs out so the accuracy might only be 10^-8. Averaged over a long period it will approach the accuracy of the GPS atomic clocks (getting down to the order of 10^-14).
The idea of the OCXO is to provide a good clock to interpolate over shorter time periods. The OCXO might be accurate (in terms of not varying rather than absolute accuracy) to around 10^-12 at 1 second - certainly better than 10^-10. But over longer times (100s to 1000s of seconds) it will drift off.
This is where the disciplining time constant comes in. For example with the Star4+ the default time constant is only 200secs. At 200 secs the GPS averaging will only have improved to 10^-10 or so while the OCXO is still around a few parts in 10^-12 so such a short time constant will make the OCXO worse. If the time constant is increased to say 1000 secs then the GPS averaging is perhaps down below 10^-11 and the OCXO drift is probably increased to approaching 10^-11 so the GPS is then almost good enough to improve the OCXO.

For the Star4+ I've found that a time constant of 3600 (1hour) gives a good compromise.

As far as accuracy is concerned, though the ADEV doesn't rise above 5 x 10^-12, the sd for the frequency difference between my two samples (on different antennas in slightly different locations) is around 1.3 x 10^-11 with the mean difference being 1.3 x 10^-13 so the accuracy after say a day of averaging is around 10^-13 but for any one seconds reading it is likely to be some where within around a few times 10^-11. Probably worse than that because I've just measured the difference between two samples - there will probably be some correlation with temperature and other effects so they may go up and down together.

If you're interested in this stuff I recommend the following web page:
https://www.wriley.com/

[jpb]

I'm also thinking at another thing, is it an issue if I use one of those extension cables https://www.amazon.fr/femelle-C%C3%A2ble-coaxial-dextension-32-80ft/dp/B079YT7NM3/ref=sr_1_6?__mk_fr_FR=%C3%85M%C3%85%C5%BD%C3%95%C3%91&keywords=cable+sma+extension&qid=1585583101&s=electronics&sr=1-6 with the puck antenna that comes with the gpsdo ? Will it decrease accuracy ? I have a clear sky view but 5 meters of cable might be a tad short. Also, I'm looking around to get a bigger antenna, but I have trouble to choose one, I don't really know what to look for (I read that some antenna can get fried by the GPSDO, as there's a voltage mismatch, or something like this) and where, on ebay it seems theres not a lot of options !

You want an active antenna (i.e. one with a built-in amplifier of say 30 to 40dB). Older ones often required 5V or more. More modern ones are more flexible - mine takes any voltage in the range 3 to 10V I think (I can't remember exact numbers).
It doesn't have to be very big but it is good if it is a timing antenna which are more geared to looking at higher satellites and ignoring lower ones where reflections etc may lead to inaccuracies.
I was very extravagant and bought one off digikey which was expensive - you don't need to do that.
The voltage mis-match you refer to is probably referring to older antennas wanting 5V while the GPSDO is only providing 3.3V say. You could get around this using a bias T and powering it externally but that is more expense and trouble.

I have my expensive antenna mounted outside on my fence but for my second GPSDO I just used the supplied puck antenna (which is an active one) in my window. The signal is much worse but it is still ok. I wanted to have the two on different antennas in different positions for measurement purposes.

As far as cable extension is concerned, you need to estimate (or look up) the loss of the cable you want to use (thinner cable is cheaper and easier to use but is more lossy) and check that the gain of your antenna is sufficient to overcome the loss. If you search the internet you'll find such calculations.

The other thing you may need to worry about if you mount it outside is some sort of lightening/surge protection especially if you're going to put it on your roof or somewhere high.

EDIT: The other thing to say regarding cables is that though everything should be 50 ohms, 75 ohm TV aerial co-axial cable can be quite low loss and people have reported good results even though there is a mis-match. I did use 50 ohm cable but, again, I probably spent more than I needed to.

[Lemonizer]

Thank you for your detailed explanation. It definitively brightens some points !
But I'm getting lost on one, I've read on many places that people calibrate their Rb standards with GPSDO's, and even Cesium standards (those also drift, even if its really low, so you do need to adjust them one day, right ?).
So what is the best, without regarding comfort (leaving something on 24/7, being dependent on a GPS signal, etc), a GSPDO ? Or a simple Rb standard ? A GPSRBO (if it exists) ?

Can a GPSDO be as accurate, or even more accurate than a local Cesium standard ?

[awallin]

fwiw, some gpsdo tests by nist
https://www.nist.gov/publications/characterizing-performance-gps-disciplined-oscillators-respect-utcnist
pdf link https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=50196

[jpb]

The GPS system has its time corrected from several atomic clocks based on earth (there are clocks in the satellites as well) so as an average measurement over a long period (say days) then it can certainly be used to correct Rubidium standards.
The second is defined from Cesium but, and I'm not an expert, I understand that there is some variation due to magnetic effects so perhaps some adjustment could be made to an individual caesium standard based on a measurement over a number of days or weeks. Though it might be difficult to decide if the cesium was a bit out or if the GPSDO measurement had gone wrong.

[Lemonizer]

If I'm right, they correct those cesium clocks (on sattelites) with hydrogen masers located on earth, as those are currently the most accurate time source. But, let's forget about those

Ok so the GPSDO will be perfect I think ! Thanks awallin, I'm reading it right now ! 

[jpb]

Hydrogen masers are the most accurate (or perhaps most regular) over a time period of something like hours to days? (I'm not sure). Cesium will always be the most accurate long term while the second is defined in terms of Cesium  transitions. I think that there is an even more accurate clock - the ytterbium optical atomic clock :
https://www.sciencedaily.com/releases/2019/04/190411101805.htm
though to take over the second would require a new definition.

Basically the most accurate clock/oscillator to use depends on the time period you want to measure over. For example, a BVA quartz oscillator is probably more stable over seconds to tens of seconds than a hydrogen maser is.

[Lemonizer]

Pretty impressive stuff ! 

Well, GPSDO's will do the trick I guess !