New Kilogram, Volt and Ohm are ready for SI-2018

[Dr. Frank]

All expected publications for the precise determination of the Planck and Avogadro constants are now online.

You may find the documents, most of them for free download, in metrologia: http://iopscience.iop.org/journal/0026-1394/page/CODATA-2017_adjustment
 
The necessary criterions are fulfilled to the due date of 30th June 2017, obviously:
R1: At least 3 different experiments agree within < 50ppb (7 experiments)
R2: At least 1 experiment  is uncertain to < 20ppb (3 experiments)

http://www.bipm.org/utils/common/pdf/SI-roadmap.pdf#page=3

I compiled these results in a graph, which is copied from Derek Mullers latest VERITASIUM video, about the redefinition of the kilogram and the NIST-4 Kibble balance.




The different numeric results are summarized in this table:



A little bit disappointing is the result of the latest PTB measurements, done on new, improved Si - spheres.
The numerical value is a bit far outside, and not overlapping with the other most precise WB experiments, from NIST and NRC.
Anyhow, the elder results from 2015 agree better.

This means, that in a few months time, CODATA 2017 will be published, with adapted values for h, e and N(A) respectively, which will then used for the final fixing of the new kilogram, the Volt and the Ohm, by end of 2018, when the according bipm congress, CGPM, will decide to implement SI-2018. I now believe for certain, that this will happen without any doubts.

These three units, plus the Kelvin, will then have quantum based definitions, with zero uncertainty in the SI.
Currently, Volt and Ohm both have about 0.2 ppm  uncertainty, and the prototype kilogram is drifting 0.05 ppm in hundred years, or so.

I'm very excited but also pleased about this outcome, looking forward to the new units.

Frank

[lukier]

Good news!

So by fixing the constants, we'll get 0 ppm error V = nhf/2e - by definition.
But the practical realization, that is the JJ array, still has uncertainty, noise floor (due to phase noise etc.). Am I correct?

[Dr. Frank]

Good news!

So by fixing the constants, we'll get 0 ppm error V = nhf/2e - by definition.
But the practical realization, that is the JJ array, still has uncertainty, noise floor (due to phase noise etc.). Am I correct?

That depends on the environment, where you look on it.
Comparing two JJAs, in cryogenic environment, at 4.2K and using superconducting devices, they agree within 1E-19.
That is on the same order as other quantum standards, like Cs or Sr clocks.

If you compare a JJVA voltage at room temperature, thermal voltages and noise limit the uncertainty to about 1nV, that is about 1E-10 relative to 10V.
That will also be the practical uncertainty limit. But currently, we do not know any 'analog' voltage reference (zener) uncertain to better than about 1E-7.

In comparison, time can easily be maintained to 1E-13 or better at everybody's home, out of thin air.

Frank

[alm]

Obviously any realization will have some uncertainty. But the advantage of defining in terms of fundamental physical constants is that the definition is stable over time (assuming the laws of physics do not change ) and can be reproduced from first principles.

[quarks]

thanks for your hint and summing up

[brucehoult]

Obviously any realization will have some uncertainty. But the advantage of defining in terms of fundamental physical constants is that the definition is stable over time (assuming the laws of physics do not change ) and can be reproduced from first principles.

It's a bit concerning that it depends on the local value of g, which changes both over time and in different parts of the room!

What about oscillating the kg at right angles to local gravity, using the precisely known V, I and calibrated (as in the video) coil?

[HighVoltage]

That is an impressive amount of work that is put in to the new kg definition.
Thanks for sharing this.

[Dr. Frank]

Obviously any realization will have some uncertainty. But the advantage of defining in terms of fundamental physical constants is that the definition is stable over time (assuming the laws of physics do not change ) and can be reproduced from first principles.

It's a bit concerning that it depends on the local value of g, which changes both over time and in different parts of the room!

What about oscillating the kg at right angles to local gravity, using the precisely known V, I and calibrated (as in the video) coil?

As far as I understood, you have no other chance than to precisely measure local G  g.
The whole principle relies on measurements parallel to the gravity force.
If you would try the measurement perpendicular, you would have no effect to measure.
Frank

[zhtoor]

Obviously any realization will have some uncertainty. But the advantage of defining in terms of fundamental physical constants is that the definition is stable over time (assuming the laws of physics do not change ) and can be reproduced from first principles.

It's a bit concerning that it depends on the local value of g, which changes both over time and in different parts of the room!

What about oscillating the kg at right angles to local gravity, using the precisely known V, I and calibrated (as in the video) coil?

As far as I understood, you have no other chance than to precisely measure local G.
The whole principle relies on measurements parallel to the gravity force.
If you would try the measurement perpendicular, you would have no effect to measure.
Frank

hello,

an analogy for mass and "volt vs JJA" could be:-

establishment of a unit of mass based on measurement of energy in a mass-energy transformation reaction, (E=mc2)
so that the unit of mass is determined in terms of c, a known constant (or is it?).

ie; m = integ(vi)dt/c2

regards.

[JoeO]

I always wondered what physicists do.   

[brucehoult]

Obviously any realization will have some uncertainty. But the advantage of defining in terms of fundamental physical constants is that the definition is stable over time (assuming the laws of physics do not change ) and can be reproduced from first principles.

It's a bit concerning that it depends on the local value of g, which changes both over time and in different parts of the room!

What about oscillating the kg at right angles to local gravity, using the precisely known V, I and calibrated (as in the video) coil?

As far as I understood, you have no other chance than to precisely measure local G.
The whole principle relies on measurements parallel to the gravity force.
If you would try the measurement perpendicular, you would have no effect to measure.
Frank

I whipped up a quick spreadsheet to estimate the rate of change of local g due to the apparent movement of the Sun and Moon.

https://docs.google.com/spreadsheets/d/1KM6urh6FdOvXzF-nFcXSEUT2DH5a1Cda5evWhQFkZoo

The Sun adds an acceleration of +/- 6 mm/s^2 with a period of one day. A little less at Paris, but I'll ignore that. The Moon's effect is around 180 times less.

Adding the vectors together, the magnitude of the total changes the fastest when the body is at 90 degrees to the local gravity field i.e. rising or setting.

For the Sun, the relative change in the total local gravity is 4.4e-08 per second. For the Moon, 2.4e-10 per second.

If you're trying to determine local G -- and then 1 kg -- to 1e-8 then you're going to have to measure g and then use g within a quarter of a second. Or at the very least, include compensation in your calculations. Or work only at noon or midnight, at new moon or full moon, when things will change more slowly.

[alm]

Local G sounds quite contradictory to me, since by the common definition g is the local gravity constant, and G the global gravity constant (i.e. approximately \$6.67 \times 10^{-11} \cdot { m^{3} \over { s^{2} \cdot kg } }\$).

Local g can be measured, so it is 'just' another calibration step that could be measured continuously and would only add a small uncertainty (measurement uncertainty and short term fluctuations).

[texaspyro]

[quote author=brucehoult link=topic=91703.msg1255439#msg1255439
It's a bit concerning that it depends on the local value of g, which changes both over time and in different parts of the room!
[/quote]

Here is a plot of the effects of solid earth tides and the vertical gravity offset (due to the sun and moon) over 30 hours.  The day this run covered was not a particularly wobbly day.

If you have a REALLY good (think >$250,000) pendulum clock you can see the effects of gravity changes.

[Dr. Frank]

Sorry, I of course meant local g, not the Gravity Constant G.

It's explicitly explained in the Veritasium video, that g has to continuously be measured, and that even the heavy permanent magnet inside the balance has its influence.

The Kibble balance equation explicitly contains g, at the bottom, to the right :





Frank

[alm]

But again, I do not see what the big deal is about measuring g. Sure, it makes the experimental setup more complicated, but it is just another quantity to measure. Apart from a (probably small) contribution to the experimental uncertainty, it does not make the experiment any less valid.

[Henrik_V]

Good news!

So by fixing the constants, we'll get 0 ppm error V = nhf/2e - by definition.
But the practical realization, that is the JJ array, still has uncertainty, noise floor (due to phase noise etc.). Am I correct?

The uncertainty is 'just' shifted to the uncertainty of how exactly we know the values of these constants. As far as we know these constants are fixed. But our knowledge of the true* value will change, with the consequence that  the values even of  the most stable artifacts will change too.   

*) 'true' values only exist in math 

[Dr. Frank]

Good news!

So by fixing the constants, we'll get 0 ppm error V = nhf/2e - by definition.
But the practical realization, that is the JJ array, still has uncertainty, noise floor (due to phase noise etc.). Am I correct?

The uncertainty is 'just' shifted to the uncertainty of how exactly we know the values of these constants. As far as we know these constants are fixed. But our knowledge of the true* value will change, with the consequence that  the values even of  the most stable artifacts will change too.   

*) 'true' values only exist in math 

That's not correct, in a metrological view.

Currently, only the frequency, or time is defined exactly, with zero uncertainty.
That's a fixed value for the Cs emission frequency.

h and e currently are not fixed, and have to be measured, implying some uncertainties: 
h= 6.626 070 040 x 10-34 J s; rel. uncertainty = 1.2E-8
e= 1.602 176 6208 x 10-19 C, rel. uncertainty = 6.1E-9 and .

By the new SI-2018, h and e will be assigned fixed values with zero uncertainty.
This will then never change any more, as these constants can not be measured any more by means of other units like formerly by the Prototype Kilogram.
Instead h and e themselves will define the other units, like the kilogram.

This change of metrological baseline is the biggest intellectual change in the new SI-2018.

Such an evolution already happened many years ago, in 1983, when the meter was defined by using fixed values for c (speed of light) and s (second), with zero uncertainty.

Frank

[alm]

And to get rid of the constants altogether, we could redefine the SI base units as fractions of fundamental constants, just like how it is common to use c (speed of light in vacuum) as unit of speed (so c = 1) in some parts of physics. Plus it would let us use those obscure metric prefixes like exa and zetta .

[Vtile]

Isn't the defining uncertainty in all these measurements the uncertainty of definition of Ampere, isn't it. Everything else today is derived from it (everything is measured with precise ampere derived measuring instruments aka electronics).  As a mere mortal it is something I wonder at times, though I'm sure there is some relatively simple solution (which I possibly could see, if I truely thought the subject over 5 minutes).

[saturnin]

Dr. Frank, thank you for very interesting and informative post! It is amazing how accurate measurements they are able to achieve...

Currently, only the frequency, or time is defined exactly, with zero uncertainty.

I would not say that today's definitions are not exact. Let's take kilogram as an example:
"The kilogram is the unit of mass; it is equal to the mass of the international prototype of the kilogram (IPK)".

The definition itself is exact. 1 kg is (exactly, with zero uncertainty) the mass of the IPK. (Uncertainty of unit realization is of course something different and always non-zero.) I know that most evidence suggest its mass varies in reality, but as follows from the definition it is exactly 1.000000... kg whatever happens (cleaning, washing - it is a real science how to maintain the IPK): http://www.bipm.org/metrology/mass/units.html#si-brochure. Still, its drift ~0.05 ppm/century is just remarkable! What other artefact standard does have such stability?

The funny consequence of the kilogram definition is: When the IPK is cleaned (and its mass is changed little bit), mass of all objects in the universe expressed in kg units is changed too (there is no real change of course ) It is evident the kilogram definition is not suitable for modern applications, so the solution is to get rid of drifting IPK, measure elementary physical constants as accurately as we can now and assign them fixed values...

[branadic]

Since the volt will be newly defined in 2018 is it worth to calibrate multimeters at the end of this year / at the beginning of next year or just wait till the new definition is implemented by all calibration labs during 2018? How long will that take?

-branadic-

[glarsson]

Since the volt will be newly defined in 2018 is it worth to calibrate multimeters at the end of this year / at the beginning of next year or just wait till the new definition is implemented by all calibration labs during 2018? How long will that take?

How many digits of precision does your multimeters have?

[Andreas]

which will then used for the final fixing of the new kilogram, the Volt and the Ohm, by end of 2018,

Hello branadic,

I fear that your question is (at least) one year too early for the annual calibrations.
And the question is how much will the change be compared to the (>=4 ppm?) uncertainity of your instruments.

With best regards

Andreas

[Dr. Frank]

The SI-2018 Volt will change by 0.00844 ppm, and the Ohm will change by 0.00015ppm, compared to the current definition of the Josephson and the Klitzing constants.

The new fixed values of constants of nature will be:
h = 6,62607015E-34
e = 1,602176634E-19   
k = 1,380649E-23
N(A) = 6,02214076E+23

These will have no error any more.

This CODATA-2017 Special Adjustment has been published 20th October 2017 in Metrologia:
http://iopscience.iop.org/article/10.1088/1681-7575/aa950a

There's also the final recommendation to change the SI to these new constants, and to draw a new baseline, eliminating the IPK. This will then be officially decided end of 2018.

Hurray!

Frank

[EEVblog]

The SI-2018 Volt will change by 0.00844 ppm, and the Ohm will change by 0.00015ppm, compared to the current definition of the Josephson and the Klitzing constants.

Calibrated 3458A's will be going cheap on ebay soon