cheap rubidium!

Not so fast. I looked them up. It's a real company. Look at their leaders, like CEO and Tech guy.

Joe

Payed US $71,99 on ebay for this in November 2012: FE-5680A Rubidium Atomic Frequency Standard 10MHz out

Not on 24/7 only every now and then.

And just mere $2,125, real bargain...

We have one like that, but we can't use ebay parts in products. And we'd need something small that would mount on a PC board.

Rubidiums are power hogs too, and have roughly 10 year MTBFs.

There seem to be two common pcb-mountable units, the $2K square ones and the $3K rectangulars. Most look identical and have the same phrases in their data sheets. Maybe they are really the same parts, or exact copies.

We need some new physics here. That the ribidium even works is an amazing collection of coincidences.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Note 3 on page 1 of the data sheet.

"The application should maintain thermal stability to obtain optimum performance. The use of a heat sink or copper plate under the device should be avoided. Device mounting should allow for a minimum of 1mm clearance from the printed circuit board".

They do seem to want you to leave a little air space as air insulation under the device.

What the data sheet says is "approximate MTBF is 100,000 hours, stationary". The schemes for calculating approximate MTBF numbers are very approximate. There's no obvious wear-out mode, and the device will probably outlast the customers application.

You do get nervous about stuff you don't understand. What are you going to advise them to use instead? Something based on the thorium--229 nuclear resonance?

Am 02.07.26 um 18:30 schrieb john larkin:

some MTI 270 series crystal dual ovens invade Rubidium turf, depending on the options. There are 2 of the 270 in my Lucent house reference as a redundant pair. Methinks they are not the real costly ones because they are locked to GPS anyway for good long term trend.

Gerhard

Yes. The MTBF is dominated by the physics package. The Rubidium is slowly absorbed into the fused quartz bulb inner wall material where the physics is done.

Physics package will be the same, and the stuff around will be optimized for this or that.

It seems that what you need is CSAC (Chip Scale Atomic Clock) which actually uses Cesium, but has Rb-level stability. Costs about $2900.

.

What are the requirements and use case?

Joe

John Larkin snipped-for-privacy@Glen--Canyon.com wrote: |--------------------------------| |"We need some new physics here."| |--------------------------------|

Why? (S.

Because rubidiums are big power hogs and have limited lifetimes. And are expensive.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

The physics package, a rubidium lamp and a separate quartz tube full of vaporized rubidium, tends to fail.

Caesium clocks just run out of caesium.

I understand it and I'm not nervous. I actually have a rubidium clock, and its schematic. It's clever and depends on a lot of luck, some physics and some numerical accidents.

What are you going

A $5 TCXO.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

You need better engineering, which is easier to find than new physics.

If you insist on hiring only engineers who are dumber than you are, you may be in trouble.

So it needs to be more carefully engineered. You need to know what the common failure modes are before you can work out how to make them less likely. It's difficult to see how a quartz tube full of vapourised rubidium is going to fail, unless some ham-fisted engineer breaks it. My glass-blowing days are long behind me, but I could probably still make one for you if I could hold of the tools required (which does include a working vacuum line).

So put in a bigger caesium reservoir

Your confidence in your own understanding is part of the egomaniac mind-set. Reality does leak through from time to time, but you don't like to admit it.

Is it going to be as stable? Have you any idea why they wanted rubidium stability in the first place?

Yes, GPS is a possibility

Been writing GPS related code the last few days. Needed a good GPS signal Indoors - no So made a 10 meter long USB extension cable to the window and put a GPS module outside. Did not work, GPS cable too long!!! Got one of these:

Typical GPS antennas have a 40 to 50 dB amplifier which needs power. Maybe your cable has a zero at 0 HZ? Cable RF attenuation usually does not play a role after +50 dB gain.

Don't expect to see a GPS signal on the SA. They are below the noise and get a positive SNR only by correlation.

Gerhard

I have several GPS modules, these

I also have one, without (USB converter) connecting directly to a Raspberry Pi serial port (via GPIO).

And then I have some of these, where the RF antenna is connected separately, also USB interface added:

There is a lot more to it, you can download xgpspc C code, but some things have been added not on my site yet.

Code to converet to heading, distance, speed, google map like latitude and longitue... more :-)

They have been just about portable and tolerable power requirements since the 1980's. It depends a lot on whether you need that sort of medium term clock stability. Today you can do it more cheaply with a GPS receiver for the majority of end user applications.

H-masers tend to be preferred in some critical applications like VLBI.

The nuclear Thorium-229 clock might be a way forward with high precision and impervious to most environmental effects but I doubt very much if it will be cheap. Iff they can make them work outside the lab.

There's an Australian over-the-horizon radar system that relies on whispering gallery modes around lumps of sapphire sunk in liquid helium

- it got written up in IEEE Spectrum a year or two ago. I don't have those kinds of contact with Australian defense industry.

Long term stability isn't wonderful, but the short term jitter is apparently very small indeed

Wikipedia doesn't think that they have yet.

Design a cheap, small, low power, long-life rubidium oscillator, and I'll buy it. Or any good clock oscillator that ages less than 1 PPB per year.

I don't insist. I have an intern this summer who may be smarter than anyone we've ever met. It's not fair for anyone to be that smart.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

The rubidium oscillator manual explains everything in detail. Locking the 10 MHz xo to the 6.834,682,612 GHz rubidium resonance is impressive.

The spec that we have requires a clock that is about 20 ppb accurate longterm. A good TCXO or OCXO ages about 1 PPB per day.

I'll just take exception to that part of the spec. They don't really need 20 ppb.

I just thought atomic clocks might be something interesting to discuss.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics