rubidium frequency standard or gps disciplined ones?

[ddavidebor]

hi guys and ladies

i have been hunting around the last month for a frequency standard for my lab.

for the same price, you can have
http://www.ebay.it/itm/TRIMBLE-10mhz-1pps-GPS-Disciplined-oscillator-A4-size-board-/290730738839?pt=US_Radio_Comm_Microphones&hash=item43b0e70497&_uhb=1#ht_5332wt_1133

or

http://www.ebay.it/itm/FE-5680A-Rubidium-Atomic-Frequency-Standard-Oscillator-Transceivers-10Mhz-Out-/270873476863?pt=LH_DefaultDomain_0&hash=item3f11513eff&_uhb=1#ht_6134wt_1133

the first is a disciplined frequency standard from gps, the second a rubidium standard.

the pros of the gps is that don't need calibration. the bad side is that need an antenna, but i alredy had one.



as you know the short term reference is given by the OCXO and the long term stability from rubidium or 1pps gps signal.

so, if you are me, what you would like to buy for normal lab use and why?

[jpb]

This is something I've been debating myself.

The advantages of GPS are:

Its neater (no worries about Rubidium tubes aging if bought second hand from e-bay), doesn't get hot like the Rubidium ones and I don't think it requires as much current as a Rubidium standard.
As you say, it doesn't require calibrating. (Having said that, if it contains an OCXO then I suppose this will get a bit hot.)

The disadvantages are:
You have to site the aerial and you're reliant on transmissions which may not always be reliable. Having an aerial lead trailing around the place might be a pain. I've also
heard that if you're a radio ham then you have to be careful not to overload the GPS from your radio transmitter (this doesn't affect me).

On the whole I'm tending towards a GPS solution - I rather like the one produced by James Miller (at least on paper, I've not bought one)

http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm

At present I'm playing around with building my own GPS one, I have a GPS module with 1pps at 10nsec rms and an oven controlled XO I got cheap off e-bay but I've not got very far yet!

One thing I'd say though is the board you linked to is A4 in size! That is pretty enormous.

[ddavidebor]

Oh god, you let me remember  that i alredy have a 1pps LEA-4t module from ublox, made for precision timing!


So i only need a board and a vcxo!

I'm a idiot!

I've no problem with radio emission in the labs... And if i put the module outside i can reach up to 6-7 sat.

[casinada]

Rubidium to Keep a frequency reference for your lab.
GPS if you want to sync time with atomic clock. It requires the antenna.

[ddavidebor]

This is something I've been debating myself.

The advantages of GPS are:

Its neater (no worries about Rubidium tubes aging if bought second hand from e-bay), doesn't get hot like the Rubidium ones and I don't think it requires as much current as a Rubidium standard.
As you say, it doesn't require calibrating. (Having said that, if it contains an OCXO then I suppose this will get a bit hot.)

The disadvantages are:
You have to site the aerial and you're reliant on transmissions which may not always be reliable. Having an aerial lead trailing around the place might be a pain. I've also
heard that if you're a radio ham then you have to be careful not to overload the GPS from your radio transmitter (this doesn't affect me).

On the whole I'm tending towards a GPS solution - I rather like the one produced by James Miller (at least on paper, I've not bought one)

http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm

At present I'm playing around with building my own GPS one, I have a GPS module with 1pps at 10nsec rms and an oven controlled XO I got cheap off e-bay but I've not got very far yet!

One thing I'd say though is the board you linked to is A4 in size! That is pretty enormous.

I've bought a cheap 15€ ocxo from china, year 2011, only 5v so i can power it from a 12v battery with a cheap linear regulator.

[AlfBaz]

I'm attempting to read up on both the FE5680A and the trimble. Haven't got very far but so far I'm thinking that you may be able to get the FE5680A with gps discipline built in... In Dave's teardown video you see a small coax connector and a bunch of unpopulated parts and I'm thinking that this may be for the GPS aerial.

As for the trimble, there's a graph showing the oscillator's drift above and below the exact value and apparently the control system wont do anything about it until it starts "walking away" from centre frequency. Unfortunately I don't know what the y-axis is as its just labelled magnitude and it shows it going from -0.5 to +0.5

Graph is on page 5-5, fig 5-2
http://trl.trimble.com/dscgi/ds.py/Get/File-10001/ThunderBoltBook2003.pdf

[AlfBaz]

I think I may have jumped the gun. It's looking like the FE5680A's with the extra sma connector (freq output) are the programmable ones

[bingo600]

Afaik there is only 2 "stock" RB's that can be diciplined by 1-pps.
As in they come with a 1-pps input for diciplining , from factory.
They are : PRS10  and X72 (X72 must have fw > 5.x).

Other RB's like the LPRO and the likes have an external C-field input that can be used for adjustments  , like the way you do on an OXCO.

The FE-5680A has an rs232 input  that can be used for tuning the RB , and i think i saw someone make a mod' that enables C-field adjustment.

/Bingo

[grumpydoc]

Rubidium frequency references are OK but consideration needs to be given to the fact that they are trimmed onto frequency by a calibration source....so you really need a GPSDO to calibrate them ! Why have both if the GPSDO may be used (you need an external aerial)..

I use the rubidium (a FE-5680A) for day-to-day use and the GPSDO (a Rapco 1804M modified for 10MHz output) occasionally. This is mostly because I don't run the GPSDO continuously and my experience is that it takes 10 days to two weeks to really settle down.

Eg I've had both on for a couple of weeks while I bed in a new OCXO for a Racal 1992. The rubidium locks in less than 5 minutes and drifts very slightly for a few minutes after that but then is stable.

Given adequate signal the GPDSO achieves "fine" lock within a few tens of minutes to a couple of hours. Actually I don't have an adequate antenna, just a puck style thing; a decent one is on the "would like" list but what I have has to do for now.

Fine lock means within a couple of parts in 10¹⁰, the manual doesn't say exactly. However I checked last week and they were about 3 in 10¹¹ apart.

Had another look this evening and the relative phase of the GPSDO output and the Rubidium hasn't changed at all in the last two hours so we're now very close to each other, probably 1 in 10¹² or better.

Since I doubt very much that the rubidium unit has drifted in the last two weeks (they're extremely stable while they work) I'm assuming that it is reasonable to interpret this as
 1) The GPDSO has been slowly converging on something better than 1 in 10¹¹ over that past two weeks.
and
 2) The rubidium was spot on - at least as close as I can measure.

But, as I said above the kicker is that the rubidium can give me this in 10 minutes after switch on.

Of course I now need a third time standard to be really confident about the other two................. 

[jpb]

If you're happy with something slightly worse than 10^-12 precision, then the easiest way to do a cheap DIY GPS disciplined one is probably to do it based on frequency counting rather than a PLL. The little module I have claims the 1pps has an rms error of 10nsecs so presumably  a count over 100 secs would be down to the 1nsec level and a 10MHz signal counted over 100 secs is 10^9.

This is what I'm aiming at, i.e. 1 part per 10^9 rather than 1 part per 10^12.

Earlier GPS units (Jupiter) produce 10kHz signals which are ideal but they are harder to find now -
http://gpsdo.i2phd.com/
 the 1pps signal is much easier as you can just buy a £30 new module.
Here is a 1pps disciplined circuit which is rather more accurate than the simple counting I envisaged:
http://www.qsl.net/zl1bpu/MICRO/SIMPLE/SimpleGPS.htm

[ddavidebor]

Well, i don't see the problem to leave the gps on for month, it can only improve itself.

My board can provide up to 1khz output signal, but i've tryed it now and don't work. I think i've killed the cheap crap antenna.

[grumpydoc]

Well, i don't see the problem to leave the gps on for month, it can only improve itself.

If I had a properly mounted and weatherproofed antenna then I'd agree but as it is I don't like leaving it exposed to the elements for too long.

[jboard146]

I started looking at this myself. What you really want is like on the one dave picked up. A GPS synced rubidium source.

I'm probably going to make one up myself, but you have to get the right rubidium reference. There is the common fe-5680a and the less common  fe-5652a. When i say common i mean common to find on ebay. FYI - I've noticed that the fe-5680a have jumped from $40-$50usd to ~$75-100.

According to the data sheet the fe-5652a takes a 1pps input form a GPS. The bottom left col on what is labeled pg 6.

http://www.freqelec.com/pdf/rfs_12pg.pdf

I'm still trying to track down the pinout for the two connectors on the fe-5652a.

I figured I'll track down the right rubidium source first then get a GPS module with a 1pps output.

This is a dual project to make a ntp time server linux based on a raspbery pi and a frequency standard for the lab all in one. I'll planning on driving the 1pps into one of the gpio pins in conjunction with the serial time output from the gps unit.

Here is an interesting page on using gps for ntp.
http://www.lammertbies.nl/comm/info/GPS-time.html

[ximac]

Hi,

One has to ask oneself, "Does the difference in accuracy really matter for the application I have in mind", as generally we are not competing with MSF or WWV. I have both standards but use my Racal 9475 rubidium one just now and again although most of the time it really doesn't matter which. Don't be put off by Dave's problem with the shakey oscillator in his Agilent counter as either a GPS or rubidium standard would be more than enough. It simply comes down to price, convenience and, of course,  personal choice.

Cheers - Joe

[ddavidebor]

I know i don't need accuracy NOW, but i prefer living me expansion possibilites

[ejeffrey]

i have been hunting around the last month for a frequency standard for my lab.

How are you going to use it?  That matters a lot.

as you know the short term reference is given by the OCXO and the long term stability from rubidium or 1pps gps signal.

AFAICT The FE-5680A does not use an OCXO for the short term signal.  This may or may not matter to you.

[ddavidebor]

I want to use with various old equipment, precise but uncalibrated that i have.

Fr the rubidium without ocxo thaks,i missed it.

One question.

Is necessary to use a sine wave clock for all the istrument, or  can use a square wave?

[jpb]

One question.

Is necessary to use a sine wave clock for all the istrument, or  can use a square wave?

A lot of instruments specify the external clock in terms of 50% duty cycle and TTL level voltage which implies a square wave rather than a sin wave. I think a square wave is the better option, you can always low-pass filter it if sin is really needed and I suspect a square wave will give less phase noise or jitter as it should be faster rising (else it would just be a sin wave anyway).

[ddavidebor]

Yes thanks, not only, square wave are super simple and cheap to trouble with... I can amp. A square wave with buffer.

[alm]

Usually a sine wave is used. A distributing a 10 MHz square wave all around your lab is a pain for EMI.

[ddavidebor]

mmh, it's in a earh referenced coaxial cable, and it's only 5v... how many problem can it do?

...

bad news (for me)
my ublox lea 4-t gps is DEAD... broken...

[ddavidebor]

Well, but use some 79xx buffer is cheapest and simplest i think, if all instrument accept 5v ttl.

[ddavidebor]

Well, how can i block dc if ttl is dc?

Is a nonsense.


But you're right, a lot of strange things will happens using ttl.

But, from the generation of the wave to the distribution (escluded) ttl are really userful...

For example, if you want to do 5mhz from 10 (some old equipment has it) you only need a flip flop.

So i'm thinking...

If i generate the 10mhz ttl, next i use buffer, then i filter it (at this point i should have something near a sine, right?, but don't know the best way to do it, any suggestion is appreciated)  and then i use a resistive divider made that between the out and ground there are 50ohm... I have a 50 ohm out with 1vpp ugly sine wave, right?