A measurement of 10 MHz down to 10 uHz

[John Heath]

Hi to everyone in the group.

Weather is fine in Canada , a nice 2 C / 36 F warm day. You have to be Canadian to understand why 2 C is a warm day.

I would like to publish my no good ideas in a journal however there will be nothing left but the feathers if the empirical evidence is not firmly stabled to the back of this paper. The empirical evidence must establish that a 5 K volt change in the voltage frame of reference of a stable 10 MHz crystal will cause a change in frequency of .000,005 HZ .  5 uHz change in a 10 MHz oscillator per 5 KV, ouch. If there is anyone one that would know the best way to accomplish this task it would be the EEVblog group which brings myself to these door steps. I would welcome any suggestions. I can say what I have built so far to put a finer focus on the VEPS project.

A] A 5 KV double Faraday cage with the inner cage changing from 0 volts to 5 K volts every 10 seconds.

B] The 5 K volt source is a small 12 volt BW TV with the HV flyback lead going to the inner Faraday cage.

C] The test board is powered by a 12 volt dry fit battery , 2 amp hour and the 10 second 5 K volt cycling is accomplished by a 555 timer and relay that switches the power to the small TV.

D] Normally the test board is plugged into power to charge the dry fit battery and make adjustments. In test mode for data logging the board is unplugged running only on battery power with one wire from the outer Faraday cage to earth ground.

E] The inner Faraday cage , 5 KV on and off , houses a 10 MHz hermetically sealed crystal oscillator, off the self type , with 5 volts in 10 MHz out with no temperature compensation nor capacity trimmer adjustments. A 9 volt rechargeable battery through a 7805 regulator supplies the power.

F] High voltage isolation from the inner Faraday cage is provided by a Toslink TTL data to fiber adapter that is also powered by the 9 volt battery for the 10 MHz test oscillator. A 20 foot toslink fiber cable provides the 10 MHz information to the data logging test board as well as the voltage isolation from the inner Faraday cage.

This is what I have so far. Everything works with the bugs worked. This brings us to this nasty business of testing for a 5 uHz change in a 10 MHz oscillator per 5 KV, ouch. If anyone one has thoughts on the best way to accomplish this it would be appreciated.

John

 

[T3sl4co1l]

...Huh?

Well...

Why would you expect that there would be a difference?

How do you expect to prove (p < 0.05) that the effect is of the claimed magnitude, or indeed, measurable at all, under any circumstances?

How can you prove that it isn't due to other effects, like power supply noise, or noise coupled into the signal cables, or... dust attracted to the enclosure by the high voltage?

A parts-per-trillion change is manifestly meaningless for the noise level of the type of oscillator used, so even if you can prove the hypothesis, and disprove possible alternative ones, what use could it possibly have?

At this level, the design of your experiment is almost irrelevant, while the design of your methodology is absolutely critical.

Tim

[TheSteve]

Can't help you on the frequency measurement but you must be in the east or something, 2 degrees is a cold day in the west!

[DimitriP]

I'd like to know what 10Mhz oscillator you have that is stable enough so you can measure a 5uH  change to its frequency.

[miguelvp]

Weather is fine in Canada , a nice 2 C / 36 F warm day. You have to be Canadian to understand why 2 C is a warm day.

Anything above freezing is a warm day in Chicago, although it has been a warm winter this year, currently 2 C/35 F it's t-shirt time

[awallin]

An off the shelf counter like SR620 or Keysight 53230A or similar has a noise floor of around ADEV(tau) = 2e-11/tau(s)
measuring a 5uHz change in 10MHz is 5e-14 fractional, so it takes at least 400 seconds for the counter-noise to average down to that level - assuming you have all other noise sources at a level below the counter.

If you need to measure a 5e-14 change faster you need a DMTD or other clever schemes.
There are both analog [1] and digital (e.g. symmetricom 5115A, 5120A, 3120A) implementations.
Quite recently there are some interesting SDR-based ideas for DMTD implementations which might be more affordable (e.g. Ettus USRP or similar)

[1] http://www.wriley.com/A%20Small%20DMTD%20System.pdf

The best BVA-type quartz crystals are at around 8e-14 or so:
http://www.oscilloquartz.com/files/1363164953-Br_%20OCXO%208607.pdf

Something more affordable (40euros on ebay) like a Morion MV89A is at maybe 2e-12 between 1s and 10s - but thats 40-fold worse than what you want to measure at 5e-14...
It will draw 1A @ 12V at warmup and maybe 0.4 A @ 12 V when warm - will not last real long on a 9V battery.

good luck! (not sure how serious OP is about this..)

[John Heath]

Why would you expect that there would be a difference?

Hi Tim

Boy is that a long answer that could easily go on for 20 pages. I will sum it up . There was an  engineer , Professor Simhony , that designed integrated circuits. His mind was such that he had to understand all there was about crystallization ,doping and electron flow. The end result was a book published describing the precise properties of a vacuum as positronium cubic matrix spacing 5 f meters.

http://simhonytribute.webs.com/epolagravity.htm

The VEPS project is my attempt to tap Professor Simhony's work a little to the left by using virtual positronium not real positronium as a building block for a vacuum. From this added flexibility time dilation from a change in a voltage frame of reference should be possible.  A GPS satellite has a projected voltage frame of reference of .5 to .7 M volts positive with a known time dilation .5 Hz at 1 GHz. This is where VEPS project 5 uHz change in a 10 MHz oscillator per 5 KV comes from. If a positive test result can be measured then a reasonable argument could be made that Professor Simhony was right. Also that the use of virtual positronium instead of real positronium would bridge Professor Simhony's work with General Relativity. That would be really cool. It is a long shot but you never know what is under a stone until you turn it over.

A parts-per-trillion change is manifestly meaningless for the noise level of the type of oscillator used, so even if you can prove the hypothesis, and disprove possible alternative ones, what use could it possibly have?

I can say from progress made that this is an understatement. Hydro 60 cycles per second can only be kept at bay by having everything run on batteries. It is possible that the entire experiment will have to be driven away from the city for reasonable error bars. From my jitter meter it seems it is out of range as well. I was thinking of additional 10 MHz crystals as a high Q band pass to sweeten up the jitter numbers. Will not know until it is tried.   

[alsetalokin4017]

...Huh?

Well...

Why would you expect that there would be a difference?

How do you expect to prove (p < 0.05) that the effect is of the claimed magnitude, or indeed, measurable at all, under any circumstances?

How can you prove that it isn't due to other effects, like power supply noise, or noise coupled into the signal cables, or... dust attracted to the enclosure by the high voltage?

A parts-per-trillion change is manifestly meaningless for the noise level of the type of oscillator used, so even if you can prove the hypothesis, and disprove possible alternative ones, what use could it possibly have?

At this level, the design of your experiment is almost irrelevant, while the design of your methodology is absolutely critical.

Tim

This ^.

This reminds me very much of the "Mikhailov experiment" and Mikhailov's conjecture about frequency change in a relaxation oscillator within a charged cage. In fact, except for the type of oscillator it sounds exactly the same. I spent many months on such an experiment around 15 years ago, and even blew out an expensive HP frequency counter doing it. You can take my word for it, the experimental difficulties are huge and the effect size is .... tiny. Yes, everything has to be run on batteries, the oscillator itself and its power supply have to be completely enclosed in the HV enclosure; getting data out under those circumstances involved optical means rather than wired connections; the lab itself was in an isolated location at least half a mile from other structures..... hoo boy. This is "big science" even if the apparatus itself can basically fit on a table-top.

[Rerouter]

Question, If the change is voltage dependent, why stop at 5KV, Its easier to make a higher grid bias than it is to measure such a small change, And i agree optical coupling would be a must, which introduces its own issues,

[HAL-42b]

Interferometry and Doppler techniques are the only thing I can think of that can measure with such sensitivity over such a range. Which means the experiment can't be conducted in air because sound waves would affect it.

Have a look at these guys http://www.thz.physik.uni-muenchen.de/research/lab/index.html

THz Time domain spectroscopy.

[KE5FX]

...Huh?

Well...

Why would you expect that there would be a difference?

How do you expect to prove (p < 0.05) that the effect is of the claimed magnitude, or indeed, measurable at all, under any circumstances?

How can you prove that it isn't due to other effects, like power supply noise, or noise coupled into the signal cables, or... dust attracted to the enclosure by the high voltage?

A parts-per-trillion change is manifestly meaningless for the noise level of the type of oscillator used, so even if you can prove the hypothesis, and disprove possible alternative ones, what use could it possibly have?

At this level, the design of your experiment is almost irrelevant, while the design of your methodology is absolutely critical.

Tim

There are plenty of OCXOs with an Allan deviation under 1E-12 at t=1 second, and it doesn't take anything that exotic to measure them. 

I've never heard of this effect, though... it sounds like it may require a lot of, um, tinfoil.

[alsetalokin4017]

Question, If the change is voltage dependent, why stop at 5KV, Its easier to make a higher grid bias than it is to measure such a small change, And i agree optical coupling would be a must, which introduces its own issues,

I didn't stop at 5 kV... I was going to 60 kV and even more sometimes. The setup was pretty dangerous now that I reflect on it. And I did wind up toasting a HP 5370B, after all.

it sounds like it may require a lot of, um, tinfoil.

Exactly......   

[uncle_bob]

Hi

Ok, the first issue is signal to noise. You want to be sure that your measurement is reading frequency and not something else. Oscillators hop and jump around at the same level that you are trying to detect. So, what is normally done?

Excite the system with an AC signal at a fairly low frequency. A 60 KV sine wave would do pretty well. You would need to make very sure that the AC signal does not get into your measurement setup. It would have to be at something other than the power line frequency or you would never sort things out. The measurement would need to be done on a fairly high quality OCXO, but nothing crazy.

If the "AC" is down around 1/10 Hz or so, there are a variety of things you could use to detect it. People like Symmetricom would be happy to sell you any of a number of boxes that would do the job. There are resources available to help with the operation of some of those boxes. If the frequency is low enough, a programmable power supply could be used to generate it.

Bob

[dom0]

I wonder how this effect works.

Is this the effect of the extremely (160? 200? dB) attenuated (by oscillator casing, hermetic metal crystal packaging etc.) E field on the crystal? Or something different entirely?

[alsetalokin4017]

The real question should be "is there an effect at all".  The "Mikhailov" effect that I was looking for was conjectured by Mikhailov to be due to the external E-field changing the mass of the electrons in the oscillator. When I finally got to the point where I was taking consistent data.... I got null results. Then I wound up toasting the counter due to an accident, and shortly thereafter the organization I was working for folded up entirely.

[Marco]

The completely ad hoc method I would try without any understanding of the underlying processes or even fully understanding the math and it's implications would be to subsample the oscillator, do FFTs of the central X seconds of each half period to find the dominant frequency, take the difference of the frequencies for each cycle, repeat and average Y times.

[zapta]

AmpHour had recently a guest that specialize in ridiculously accurate measurements. I would contact him and see if he has any advice. Drop the EEVblog and Dave's name, it may help.

[nctnico]

The completely ad hoc method I would try without any understanding of the underlying processes or even fully understanding the math and it's implications would be to subsample the oscillator, do FFTs of the central X seconds of each half period to find the dominant frequency, take the difference of the frequencies for each cycle, repeat and average Y times.

This could be a good approach. The jitter and/noise should spread across the spectrum but a frequency offset should stand out.

[uncle_bob]

The completely ad hoc method I would try without any understanding of the underlying processes or even fully understanding the math and it's implications would be to subsample the oscillator, do FFTs of the central X seconds of each half period to find the dominant frequency, take the difference of the frequencies for each cycle, repeat and average Y times.

This could be a good approach. The jitter and/noise should spread across the spectrum but a frequency offset should stand out.

Hi

You are looking for an effect that is "greeted with skepticism". The best way to prove that it is there is to use well know techniques. There is already a lot of gear out in the market that will make these measurements. Doing something odd opens you up to challenges. Defending the measurement at the same time as the results is *not* where you want to be. Making a measurement error and missing out on that Nobel Prize, also would be a major bummer.

Bob

[Marco]

You are looking for an effect that is "greeted with skepticism". The best way to prove that it is there is to use well know techniques.

I guess then just use a PLL FM demodulator and digital lockin amplifier, you should be able to get that from big brand name lab devices.

[Bud]

. There is already a lot of gear out in the market that will make these measurements.

I wonder what they are... There is an area of research and application of crystal resonators frequency shift called Quartz Crystal Microbalance where they measure shift in frequency caused by depositing molecules on the surface of the crystal (change in mass), and it seems it is a challenge to reliably measure the shift with better than 0.01 Hz resolution. BTW, i recommend OP to read about QCM, there may be a few good ideas to pick up from there.

[T3sl4co1l]

Ugh. Word soup.

This explanation,

Boy is that a long answer that could easily go on for 20 pages. I will sum it up . There was an  engineer , Professor Simhony , that designed integrated circuits. His mind was such that he had to understand all there was about crystallization ,doping and electron flow. The end result was a book published describing the precise properties of a vacuum as positronium cubic matrix spacing 5 f meters.

http://simhonytribute.webs.com/epolagravity.htm

The VEPS project is my attempt to tap Professor Simhony's work a little to the left by using virtual positronium not real positronium as a building block for a vacuum. From this added flexibility time dilation from a change in a voltage frame of reference should be possible. 
(...)

Means nearly nothing to me, is incoherent in terms of traditional physics, and Simhony's FAQ describing things in better detail (find it here http://www.epola.co.uk/faq/FAQframe.html ) is similarly self-inconsistent, poorly thought out, and simultaneously utilizing, and against, various parts of known standard physics.

I suppose my insistence on methodology will end up being ignored in the end, which would at least be consistent with the types of "experiments" those kinds of people like to do their childs-play with.  But whatever.  Best of luck whatever you end up doing, I suppose.

Tim

[TimFox]

An extremely sensitive way to measure small fractional deviations of much, much higher frequency is the Mossbauer effect for gamma-ray frequencies.  See  http://www.mossbauer.info/mossbauer.html.
For example, this was used to measure the difference in gamma-ray photon energy (i.e., frequency) due to gravitation over the height of the tower on the Harvard campus.  When the emitting and target nuclei are each constrained by a crystalline lattice (e.g., Co-60 in bulk cobalt), the effective width of the resonant interaction is incredibly narrow and frequency shifts can be measured in the audio range.
If your effect is real, it should be visible with the source in one Faraday cage at high voltage, and the target in a grounded Faraday cage where you can do the measurement.

[uncle_bob]

. There is already a lot of gear out in the market that will make these measurements.

I wonder what they are... There is an area of research and application of crystal resonators frequency shift called Quartz Crystal Microbalance where they measure shift in frequency caused by depositing molecules on the surface of the crystal (change in mass), and it seems it is a challenge to reliably measure the shift with better than 0.01 Hz resolution. BTW, i recommend OP to read about QCM, there may be a few good ideas to pick up from there.

Hi

Well the one next to me on the bench here is a Symmetricom 3120A. It will do the job quite nicely. They have many other instruments (as do others) that also will measure the expected impact of a 60KV shift.

Truth in lending, I don't work for Symmetricom. I did not work on the 3120A at any point in time. I'm just a happy customer. Yes, there is *slightly* more to the story, but it's not relevant.

Bob

[KE5FX]

. There is already a lot of gear out in the market that will make these measurements.

I wonder what they are... There is an area of research and application of crystal resonators frequency shift called Quartz Crystal Microbalance where they measure shift in frequency caused by depositing molecules on the surface of the crystal (change in mass), and it seems it is a challenge to reliably measure the shift with better than 0.01 Hz resolution. BTW, i recommend OP to read about QCM, there may be a few good ideas to pick up from there.

Hi

Well the one next to me on the bench here is a Symmetricom 3120A. It will do the job quite nicely. They have many other instruments (as do others) that also will measure the expected impact of a 60KV shift.

Truth in lending, I don't work for Symmetricom. I did not work on the 3120A at any point in time. I'm just a happy customer. Yes, there is *slightly* more to the story, but it's not relevant.

Bob

Warranty void above 10 kV...