EEVblog #456 - CSIRO Rubidium Frequency Standard

[ddavidebor]

I don't understand a thing....

Gps provide pps, but is pps used to correct the rubidium or only is a separate thing?

And what tipe of equipment use ppm input?

[ddavidebor]

Oh, i can't understand a word.

What you say when you say something like "conformally coted" referring to the board inside the rubidum standard?

[Ed.Kloonk]

Oh, i can't understand a word.

What you say when you say something like "conformally coted" referring to the board inside the rubidum standard?

The clear, shiny coating that covers all components on the board.

[Ed.Kloonk]

ddavidebor,

When you say

Oh, i can't understand a word.


or

I don't understand a thing....

It sounds like you are not understanding anything.

Better:
I don't understand one particular word, I don't understand something.

Hope that helps.
 

[bingo600]

I don't understand a thing....

Gps provide pps, but is pps used to correct the rubidium or only is a separate thing?

And what tipe of equipment use ppm input?

Yes the PRS10 takes an external 1-PPS and can dicipline the 10Mhz rubidium output against the GPS provided 1-PPS.
So you get a "selfcalibrating" 10Mhz RB.

I have an X72 Rubidium , bought on the 'Bay for that purpose , but it (1-PPS diciplining) was implemented in firmware 5.x.
And the 'Bay seller forgot to say my X72 had 4.x firmware in it .... So watch out if trying to get an X72 to do it.

/Bingo

[tinhead]

Oh, i can't understand a word.

What you say when you say something like "conformally coted" referring to the board inside the rubidum standard?

http://en.wikipedia.org/wiki/Conformal_coating

[free_electron]

I suspect that these were intended to be used with the Rubidium standard rather than the inbuilt oscillator.

The reason is that there is an option for this machine to have a real oven installed inside. That is why there is so much empty space.

These counters use a switching power supply from delta (custom hp part number) that is always on. The power button is a soft power button that only toggles power to the counter section.

In the big cavity between front and power supply you can install an 11806 oven oscillator. Those beasts are dead stable. They go on ebay for about 100$ but you need the interface board. (Working on that.. I have a 54131 too and i have an 11806... ) the interface board is a few connectors and two co parators that signal the state of the oven to the counter( temp ok and voltage ok)

On the right hand side , above the fromt end (between front and the bix xilinx ) there is the room to install the third channel. Depending on the freque cy band chosen this module either screws onto the right hand support strut , or it is a block that drops in.

So there you go. That is why this thing looks so empty and is built the way it is built.

The two empty flatcable connectors are to connect the oven and prescaler for channel3.

If you have a standard , like daves rubidium, just hook it up to the back panel ref-in. The counter will detect it , disengage from its built in reference and use the incoming signal. There is some monitoring circuitry that detects the signal falling away and then the machine switches back to its own source.

There is a reason for using 10Mhz. They are the easiest ovens to make and agilent started , back in the hp dayss, to provide an extra input for 10mhz on all their time related equipment ( scopes, signal gens, spectrum analysers etc...) when you buy an oven for a machine it comes with an extra bnc that brings the ref out.

Originally they sold the oven in an external case as well.

Every machine has a ref in and ref out and you can daisy-chain the machines. There is a buffer inside each machine. Of course you incurr skew every time you pass through an instrument.

That is why they sell distribution amplifiers. These are carefully designed to also be phase aligned outputs. You do need to use cables with identical length though !

In my ADSL days i had an agilend gps disciplined reference ( now owned by symmetricom) and such an amplifier. All machinery in the lab was connected to it. We had two network analysers. Prior to connecting them we saw differences between the machines. Connect both to the standard and the became spot on. ( they were calibrated by agilent in their lab using such a standard , so the moment we connected to ours they were performing at their peak.

[Zad]

Daummmm .... Dave
A used PRS10 is 750+ US$ on the 'Bay ... alone

You are looking in the wrong place then. They have been going recently for $400

www.ebay.co.uk/itm/SRS-STANFORD-RESEARCH-SYSTEM-PRS10-RUBIDIUM-FREQUENCY-STANDARD-/130891002337

[gxti]

When dave did the tour of the HP cal lab and looked at the cesium clock it was stated that the cesium source eventually wears out or is consumed as the atoms are boiled off. Do rubidium clocks suffer the same type of life-span limit? If no, how do they die? Those FE-5680 units can look thoroughly knackered from external appearance.

Cesium oscillators consume cesium as they operate, it's an irreversible process. Rubidiums do not have such a problem. They do deteriorate with age as the rubidium metal deposits on the wall of the chamber but it is possible to 'rejuvenate' it with heat.

[bingo600]

Daummmm .... Dave
A used PRS10 is 750+ US$ on the 'Bay ... alone

You are looking in the wrong place then. They have been going recently for $400

www.ebay.co.uk/itm/SRS-STANFORD-RESEARCH-SYSTEM-PRS10-RUBIDIUM-FREQUENCY-STANDARD-/130891002337

Hmmm ... I hope there are more to come ...

Thnx

/Bingo

[EV]

Here is Self made GPS clocks 10 MHz signal as input to Agilent 53131A counter channel 2 (Pic-1) and self made Rubidium standard as input to Agilent 53131A counter Ref input (Pic-2). Measurement result is in Pic-3. Gate time is 60 seconds.

[jpb]

if i had a GPS disciplined, rubidium frequency standard on hand
i would do the only right thing for a hacker to do...

take my egg timer apart. and replace the 32.768 khz crystal with a divide by 305.17578125... circuit
and feed it the 10mhz reference...

finally go cook some eggs up with a only a few ppm's worth of precision and consistency

challenge posted!.   



haha..


/Kyndal

You'd also need a multiple laser interferometer to get the egg's precise volume and  surface area and a very accurate set of scales and measuring the air pressure would also need to
be done to very high precision to determine the boiling point of the water.

On the timing side, provided you wanted a whole number of seconds (say precisely 3 1/2 minutes) then the 1ppm from GPS is typically in the 10s of nsecs range,
which over 210 secs would be reduced further. This may be good enough.

[free_electron]

You'd also need a multiple laser interferometer to get the egg's precise volume and  surface area and a very accurate set of scales and measuring the air pressure would also need to
be done to very high precision to determine the boiling point of the water.

On the timing side, provided you wanted a whole number of seconds (say precisely 3 1/2 minutes) then the 1ppm from GPS is typically in the 10s of nsecs range,
which over 210 secs would be reduced further. This may be good enough.

That's all easy stuff. It's getting the calibrated chicken that's the hard bit...

[codeboy2k]

That's all easy stuff. It's getting the calibrated chicken that's the hard bit...

Negative Feedback. Using the aforementioned laser measuring tools, each egg is measured at the factory as its laid.  When it's high or low out of spec, the information is fed back to the hen that laid it, via a much higher power laser Once this loop stabilizes only hens laying eggs in spec will be alive in the hen house

[c4757p]

Negative Feedback.

Dammit! I was going to say that. 

[codeboy2k]

Dave,

Those are nice in-house designed interface boards for the PRS10 and the GPS receiver.  They are not that difficult to design and build and even troubleshoot, but it's nice to (a) have the schematics, and (b) read the schematics of a proven design. 

Given that your Australian tax dollars funded the development of those boards, I think they should release the schematics for it to you, if they haven't already made them available.

[hairykiwi]

You are looking in the wrong place then. They have been going recently for $400

www.ebay.co.uk/itm/SRS-STANFORD-RESEARCH-SYSTEM-PRS10-RUBIDIUM-FREQUENCY-STANDARD-/130891002337

Thanks for sharing Zad. And /Bingo - It looks like they're relisted again.

[codeboy2k]

Chip scale atomic clock developed at the NIST in the USA

Lots of information there, but the web page is not clear about what can be clicked on.  Most of the numbered lists on their pages are actually click-able links, but have nothing to indicate that.

http://tf.nist.gov/timefreq/ofm/smallclock/CSAC.html

[hairykiwi]

codeboy2k - the specs on that device are really AWESOME! Though, if it were several orders of magnitude larger, I'd say it definitely looks like it came out of a well used DeLorean.

Back to affordable frequency standards, it would be really interesting to hear what those of you with a Rubidium Frequency Standard laying around think of the 'Network Frequency Transfer' project.  It's advertised as a 'better than 1ppm' frequency standard by it's Danish creator, Kasper / KKP. Here's the link to his public announcement of the project back in 2009:
http://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=82842&highlight=ntp

According to one guy on that thread - alank2, it's accuracy is pretty impressive:

I finally got around to testing my rubidium frequency reference against a NTP server using your excellent NFT utility.
I ran it for 24 hours and it came out 0.00 ppm for first/last and all samples! I thought it would be some off...

And here's Kaspers original post for convenience. (His pdf and exe links are still working.)

If you have traceable frequency standards and several counters to choose from, this is likely not for you.

I have kludged up some software to let someone, with no frequency standard, and no counter, verify an oscillator to better than 1 ppm using an Internet connection. Then I thougt maybe this would be generally useful, and spent another day polishing it.

It works like this: The user's hardware sends characters* to a Windows PC's serial port at regular intervals. Both native and USB works. This drives a virtual clock in the pc, and this clock is compared against an NTP server on the Internet. After some statistics and regression math, out comes the offset. It takes about 15 minutes for 1 ppm.

http://n1.taur.dk/nft/nft.pdf manual with screenshots
http://n1.taur.dk/nft/nft.exe win32 exe

In order for this to work, an NTP server with a healthy refclock is needed. I've put in mine as default; If you are far far away from Denmark, you may want to find another, though mine may work perfectly well for you**.

Also, this is prototype software, so if it spits out unbelievable data, then it's probably broken.

License: This software is under Beerware license. If you find it useful, and we meet, you may buy me a beer.

/Kasper

*or a pulse the PC will interpret as a character. The value does not matter.
**unlike for timekeeping use, for nft static latency does not matter.

[grumpydoc]

Neat.

It's certainly better than no reference at all but to be fair 1ppm is only really useful for a straightforward crystal oscillator, or perhaps a TCXO. Good ovened oscillators are stable enough to try to set them up to 1x10^-8 to 1x10^-9, Rubidium units should be in the in the 10^-9 to 10^-11 range.

[EEVblog]

It's certainly better than no reference at all but to be fair 1ppm is only really useful for a straightforward crystal oscillator, or perhaps a TCXO. Good ovened oscillators are stable enough to try to set them up to 1x10^-8 to 1x10^-9, Rubidium units should be in the in the 10^-9 to 10^-11 range.

I agree that 1ppm is pretty useless.
You can get a lot better than 1ppm by a DIY  TXCO design (but it's messy of course).
I don't really consider anything a decent reference standard unless it's at least rubidium. e.g. > x10-9
Sure you can get some good oven based oscillators, but they just aren't really the same level.

[codeboy2k]

I don't have a rubidium standard here, but I did buy a couple of old 10Mhz ovens from eBay a few months back.. I'm working on a DIY GPS disciplined 10Mhz standard for my own home lab.

After several hours (I think it's 24 to 48 hours) you can get its Allen Variance down to between 10^-11 or 10^-12. Thats still good for my lab. I don't have anything here that can measure that good yet.

And I agree, I would rather have a rubidium standard as the reference clock to discipline. The oven clock will drift more and faster than the Rb standard if it loses GPS lock. So for my uses, I am hoping that my ovenized standard will be ok.. I'm thinking about putting a lock LED and a 48HR LED. One light locks when it has GPS lock, the other light locks after 48 hours of consistent lock.  Then I'll know when the output has been "in training" long enough to assume it's good (like I said, I can't measure it).

[olsenn]

Out of curiosity, has anyone ever tried making a circuit that takes the average frequency of say 10 or 20 standard (pick of the bunch) crystal oscillators to get higher accuracy at low power levels? How about circuits that compensate for the temperature-frequency variation curve?

[c4757p]

Out of curiosity, has anyone ever tried making a circuit that takes the average frequency of say 10 or 20 standard (pick of the bunch) crystal oscillators to get higher accuracy at low power levels?

I'm no statistician, but I don't think this will be worth much. It's entirely possible for all of them to be out in the same direction - and not entirely unlikely.

How about circuits that compensate for the temperature-frequency variation curve?

You mean a TCXO?

[olsenn]

I'm no statistician, but I don't think this will be worth much. It's entirely possible for all of them to be out in the same direction - and not entirely unlikely.

Possibly, but I reckon not probably. Dave did a video on this before using resistors instead of crystals. However, I could be wrong; for all I know, drift could always be in one direction for crystal oscillators.

You mean a TCXO?

I was under the impression that TCXO is an oscillator that is either kept at a specific temperature with an oven, or contains housing (or possibly even is a certain cut of crystal) that is insensitive to frequency variations due to temperature. Do these oscillators actually come with active circuitry built in?