Rubidium vs. GPSDO 10Mhz oscillator

[DC1MC]

Dear metrology gods, it is time for me to have a frequency reference because of TE piling-up. I wanted to ask you what will be the first 10MHz reference to get, a rubidium or GPSDO one ?
Aside for the obvious things, like the GPSDO needs GPS signal but uses less power and the rubidium one needs more power and the lamp may die and it's unbotanium but is independent of the outside world, are there any other criterion to consider ?
I know that in the end I'll get both, but being a poor mouse I can only start with one.

 Thanks,
 DC1MC

 

[ap]

It depends on what you need this for. If you truely need a calibrated source, a GPSDO is the only solution for you. It is pretty unpractical though, because it takes relatively long to lock, and you need the antenna setup as you mentioned, but its uncertainty is defined. If you buy a RO, you are not sure about its tolerances unless you have it calibrated. If you just need something arround 10MHz, and uncertainties are of no concern, even a TCXO will do. Thats probably not what you want, given the question. So younds like a GPSDO is what you need.

[kj7e]

Anymore, a quality GPSDO can out perform a Rubidium standard.  The major issue is getting the GPS antenna clear view of the sky.  Fortunately I own a home out in the country so its not an issue for me.  But if you lived an a high rise apartment building in down town Manhattan it may be a problem so a Rubidium may make more sense.  If you can get a good GPS signal, I would go for a good GPSDO first.  Then pick up a Rubidium so you can sanity check against each other.

I read this the other day, its a few years old but still lots of good info on this subject;
http://ws680.nist.gov/publication/get_pdf.cfm?pub_id=842479

[edpalmer42]

I agree with the comments made by ap and kj7e.

If you have good visibility towards the equator (you'll see more satellites), a GPSDO is the best starting point.  If your visibility is questionable, do you have a way to check the frequency of a Rb?  Is so, then that becomes first choice.  If you have poor visibility and no way to check the frequency of a Rb, then I guess you toss a coin!

Ed

[jfphp]

Don't forget two points.
GPSDSO exhibit large jumps (XX 10-11) from time to time even after the end of the selective availibility, even a GPS disciplined Fluke Rubidium. Comparison with a cesium is very interesting and I have the paper comparisons with a chart recorder between cesium and HP 58503B.
Phase noise of a GPSDSO is poor, as a rubidium and a cesium (except the high performance tube of the 5072A).
Running a SA or a counter with these standards is much more noisy than a selected 10811A( there are several pages of the 108811A with different PN, one selected for phase noise, one for aging, one for stability in g acceleration, one standard low cost for most of the counters , one for relaxed...).
Shortly said, the stability of a GPSDSO is only valid after several days of averaging, in short time is is worse than a rubidium or high performance oscillator and in very short tima (Allan variance/phase noise) is is worse than a very good OCXO. Rubidium, even Standford Research, and cesium except high performance tubes, are the poorest.

[glarsson]

When you give performance info about GPSDO do you mean a GPS receiver directly controlling a oscillator or controlling a quality OCXO?

[DC1MC]

When you give performance info about GPSDO do you mean a GPS receiver directly controlling a oscillator or controlling a quality OCXO?

I want to bit on this guy:
https://www.ebay.de/itm/TruePosition-10mhz-GPS-DISCiPLINED-CLOCK-GPSDO-more-information/292452398481

What is your opinion ?

 Thanks,
 DC1MC

[jfphp]

Not a basic GPS receiver controling a TCXO by the 1pps output but a HP 58503B :  very special OCXO (10811A with 2 enclosures !) and software enhancement originaly built to average the instabilities induced by the selective availibility now( officialy...) suppressed. Beware of the ZXXX GPSDSO : they were never intended as reference oscillators but as synchronizing tools in networks. OCXO is poor and the software modifications (mostly chinese vendors) on EBay can never put this rig on the level of the 58503A/B (totaly obsolete in the GPS domain with the number of satellites and the catastrophic sensibility).

[jfphp]

Dear DC1MC,

It seems thar it is again a basic GPSDO with the traditional 1pps comparison.
Have a look at the description of the Fluke 910 and 910R. The limitation of the DSO are clearly explained :


"Two high-stability operating modes to suit your application

Most users prefer automatic adjustment (known as disciplining) of their frequency standard, to fully eliminate long-term frequency changes (aging). This disciplined mode is also the default mode in the 910 and 910R. As long as there is a valid satellite signal, the internal local oscillator is monitored and adjusted and the mean 24-hour frequency offset is always virtually zero. However, in this mode, the inherent short-to-medium term stability of all local oscillators, except rubidium, is compromised. This is true for all GPS frequency references. The received GPS signal has relatively large short-term frequency variations, due to variations in atmospheric conditions. This means that when using the received GPS signal for disciplining the 910 (OXCO), the stability is reduced a little for averaging times of 100 s to 1000 s.

 

In this mode, the frequency deviation between the internal timebase oscillator and the received GPS-signal is used to continuously adjust the oscillator (disciplining). The resulting frequency offset and adjustment data is stored in non-volatile memory every 24 hours, to enable printout of the traceability record. The actual frequency offset (24-hour mean value) is calculated and displayed on the front panel.

Some applications demand superior short-medium term stability, especially for jitter and wander measurements in digital telecommunication networks.

The unique Manual Hold-Over Mode makes it possible to switch over temporarily from Disciplined to Hold-Over Mode during the actual measurement, thereby achieving a superior frequency accuracy at the start of the measurement and a superior stability through the measurement. Here, the internal oscillator is not adjusted. This mode is normally automatically entered when there is no usable received GPS-signal. This mode can also be selected manually by activating the Manual Hold-Over Key. If Manual Hold-Over is set together with a valid received GPS signal, the actual frequency offset is calculated, displayed and stored in non-volatile memory every 24 hours.

For the ultra-stable rubidium oscillator in the 910R, there is no measurable difference between the stability in Disciplined and Hold-Over mode, for averaging times up to 1000 s."

[DC1MC]

Dear jfphp, many thanks for the valuable advice, your little expose about the devices properties, strengths and weaknesses was very instructing, today I've learned. While the devices that you recommended are indeed professional stuff, the Fluke costs as much as my car and it's mostly unobtanium and the HP is still 10times more expensive then this poor dirty mouse that I've bought (please see the attached picture), there the GPS module seem to be Japanese and oscillator seem to be American, actually I don't think I've seen anything very Chinese/Asian on this board that comes from Iowa.

So if anybody has some experience with the "True Position" boards like the one in the picture and/or can recommend an acceptable priced GPS antenna, I will be very grateful. I have an excellent south view on my last floor located apartment.

I hope to get from Germany a "classical" Efratom Rubidium oscillator and I'll do a battle royal in between them .
 
Now that I'm here could you please recommend an (affordable) way to make a 10MHz square wave out of these sinus wave, with good edges and low jitter, what kind of comparator/circuit should I use ?

 Cheers,
 DC1MC

[jfphp]

1. Check with a voltmeter connected to the antenna center conductor if you find 5V, standard antenna power supply on the GPS. You have to buy a timing antenna not a generic one for GPS localization (not filtered). For example, but you don't need 40 db and prefer 26 if you put it in free space (will pick up a lot of junk in a city)
https://www.ebay.com/itm/Lucent-GPS-40dB-High-Gain-Timing-Antenna-Lightning-Surge-Protection-Mount-Kit/222789666215?epid=10008077643&hash=item33df4c85a7:g:-yEAAOSwjkdZ81n5

2. Try to find (vendor ?) the software to communicate with the GPS. It is very important to see, what is wrong, the number of satellites and to insert  a compensation for the delay introduced by a long antenna cable...

3. I don't really understand your problem : is the 10 Mhz sinus phase locked ? Why do you need a sqare wave ? Classical problem is : 1 pps out of the GPS (every one) and 10 Mhz sinusoidal. There are hundreds of schematics on the net for this classical problem of phase lock. The first one found by Google
http://www.ik0otg.net/index.php?option=com_content&view=article&id=53&Itemid=59&lang=en
Cmos IC have a lot of jitter but you have to heavy average/have a very long time constant (short to get phase lock then long : a switch is a silly solution).

[DC1MC]

1. Check with a voltmeter connected to the antenna center conductor if you find 5V, standard antenna power supply on the GPS. You have to buy a timing antenna not a generic one for GPS localization (not filtered). For example, but you don't need 40 db and prefer 26 if you put it in free space (will pick up a lot of junk in a city)
https://www.ebay.com/itm/Lucent-GPS-40dB-High-Gain-Timing-Antenna-Lightning-Surge-Protection-Mount-Kit/222789666215?epid=10008077643&hash=item33df4c85a7:g:-yEAAOSwjkdZ81n5

Thanks for the suggestion.


2. Try to find (vendor ?) the software to communicate with the GPS. It is very important to see, what is wrong, the number of satellites and to insert  a compensation for the delay introduced by a long antenna cable...
I have fond tons of threads and support here on forum and allover the internet, the communication is over a simple srial link and fully understood.


3. I don't really understand your problem : is the 10 Mhz sinus phase locked ? Why do you need a sqare wave ? Classical problem is : 1 pps out of the GPS (every one) and 10 Mhz sinusoidal. There are hundreds of schematics on the net for this classical problem of phase lock. The first one found by Google
http://www.ik0otg.net/index.php?option=com_content&view=article&id=53&Itemid=59&lang=en
Cmos IC have a lot of jitter but you have to heavy average/have a very long time constant (short to get phase lock then long : a switch is a silly solution).
Just wanted to make a 3.3V CMOS output out of it.

Cheers, DC1MC