Disagreements of preciseness of h in PDG and at Watt Balance

[exponent137]

Planck constant in
http://pdg.lbl.gov/2015/reviews/rpp2015-rev-phys-constants.pdf
determined to 12 ppb.

Planck constant in
https://www.sciencedaily.com/releases/2016/06/160621115645.htm
was measured with 34 ppb and it is a big step forward in replacing of Paris kilogram with more stable definition. They predict still a better measurement with this device.

A similar measurement was with watt balance was also made in Canada a few years ago and it was precisier. A do not find link now.

How it is possible that PDG have value 12 ppb, but value 34 ppb is good anyway? I suppose that value 12 ppb is measured also by watt ballance?

What is connection among these values?

 

[exponent137]

I will ask differently:

Planck constant in
http://pdg.lbl.gov/2015/reviews/rpp2015-rev-phys-constants.pdf
was determined to uncertainty of 12 ppb.

Planck constant in
https://www.sciencedaily.com/releases/2016/06/160621115645.htm
http://physicsworld.com/cws/article...-4-watt-balance-weighs-in-on-plancks-constant
was measured with 34 ppb and it is a big step forward in replacing of Paris kilogram with more stable definition. They predict still a better measurement with this device. They expect 20 ppb of uncertainty and then they will fix value of h, and kg will be determined with this.
and:
"The best watt balance measurement of Planck's constant so far comes from Canada's National Research Council, with an uncertainly of 19 parts per billion"
http://phys.org/news/2016-06-important-milestone-road-redefined-kilogram.html

But, why 19 ppb is such a big achievement, if 12 ppb was already achieved?
Is 12 ppb incorect, or it was not obtained by a Watt balance?

 

[mfb]

At that level of precision, you always have to check in which units the measurement is given, and what exactly has been measured, as unit conversions change the uncertainties.

The most precise measurement seems to be in terms of eV*s, the conversion to J*s happens via the elementary charge which has a similar uncertainty. Both measurements don't depend directly on macroscopic masses - I don't see how you could use those to measure the mass of anything.

 

[Vanadium 50]

The PDG 12 ppb value is an average over several measurements. The NIST 34 ppb value is a single measurement. How did I know this? I read the links.

 

[exponent137]

According to PDG link I suppose that h is calculated via fine structure constant and e. Because 2*6.1=12 This means that this is another measurement than Watt Balance?

The next question appears: how e is measured?
In
https://en.wikipedia.org/wiki/Elementary_charge
it is written, that the best measurement is by Watt Ballance.

V50:If you will look fig 15 in

http://scitation.aip.org/content/aip/journal/rsi/87/6/10.1063/1.4953825

you will see that average of such measurements cannot give only 12 ppb.

 

[mfb]

The PDG 12 ppb value is an average over several measurements. The NIST 34 ppb value is a single measurement. How did I know this? I read the links.

That would need at least 8 independent (!) measurement with a precision of 34 ppb each, and ~30 measurements for the 6 ppb value. I don't think that is the full story.

 

[Vanadium 50]

you will see that average of such measurements cannot give only 12 ppb.

A. Take it up with CODATA.
B. Nonsense.

Normally I would just ignore you, but I am getting pretty annoyed with your filling the forum with nonsense. I am getting equally annoyed by your refusal to do a lick of work yourself, instead dumping it on us. (If your feelings are hurt, they should be) Had you asked a question, I would have been more polite - but you made a statement. A statement that is completely wrong.

The cited measurement gets 148 +/- 34 (all values are h/h90-1 in units of 10^-9)
Ref 46. gets 189 +/- 18
Ref.47 measures the Avagadro constant (which is equivalent to measuring h, since N_Ah is known to 1 ppm) to +/- 36
Ref.48 measures the Avagadro constant to +/- 20
Ref. 49 gets 158 +/- 87
Ref. 50 cites three measurements: 29 +/- 19, 95 +/- 37 and 106 +/- 38

Averaging all those together gives an uncertainty of just over 12 ppb on the average.

 

[Vanadium 50]

I don't think that is the full story.

The full story is in the references, all of which have links. The 34 ppb measurement is not the best, or even particularly good. There are three measurements around 20: that gets you close to 12.

 

[exponent137]

Thus, uncertainty can be calculated as:
https://www.physicsforums.com/threads/uncertainty-of-an-average.612633/#post-3949676

But, what if scattering of results is larger than their uncertainties?
for instance:
189 +/- 18
29 +/- 19
https://www.physicsforums.com/threa...-pdg-and-at-watt-balance.876942/#post-5511555

1 How tu use this scattering in calculation? (I suppose that it is not used, but why not?)
2 Does this means that uncertainties were underestimated?
3 Is this consequence of drift of measurement masses?
4 But, why then to use average?P.S.
V50: This was intended as a question, not as a statement or as a claim:

If you will look fig 15 in
http://scitation.aip.org/content/aip/journal/rsi/87/6/10.1063/1.4953825
you will see that average of such measurements cannot give only 12 ppb.

 

[mfb]

But, what if scattering of results is larger than their uncertainties?
for instance:
189 +/- 18
29 +/- 19

Then something went wrong. PDG typically takes this into account by scaling the uncertainties up until some reasonable agreement can be seen.

2 Does this means that uncertainties were underestimated?

Or at least one measurement is just wrong.

 

[Meir Achuz]

PDG numbers are not from any particular expts. They are determined by a statistical fit of all constants to a large number of measurements, not only those specifically of h. This can give a smaller stat error.