On Sun, 13 Jun 2010 21:48:58 GMT, Owen Duffy wrote:

Steps 1 and 2 are quite explicit.

I have reviewed Walt's additional material at
http://www.w2du.com/r3ch19a.pdf .

Walt's load at step 1 is stated as 50+j0 without tolerance, but I accept
that Walt is thorough and would have confidence that it is low in error.

Convention has it that any report is understood to vary by 50% of the
least significant digit. There are other standards that do not wander
far from this simple expectation.

Hence 50±0.5+j0±0.5

Walt has an excellent GR 1606-A Bridge - same model as mine.

To give everyone some idea of what "Qualified" means in my former
trade, my bridge came with a traceable certificate of measurement that
tells me that me my accuracy:

" Resistance:
±[3.0 + 0.0024 · f² · (1 + R/1000)]% ± [(X/f · 10000) + 0.1] Ohms
" Reactance:
±4.0% ±(1.0 + 0.0008 R · f) Ohms
"This certificate is valid only when the bridge is balanced at the
connector reference plane and when all measurements made at this plane
are corrected as outlined in para 4.4 and 4.5 in the manufacturer's
operating manual wit the exception that the capacitive reactance to
ground (Fig. 6) is computed from a value of 2.2pF. Fig. 8 should be
used for the resistance dial multiplying factor."

The correction factors easily leave behind the nominal value of 1.0
for high resistance (such as Walt's 1400 Ohms) with 5% error in the
20M band, and 60% error in the 6M band. Unless, of course, you apply
all corrections (the whole point of having the bridge calibrated).

He reports Pf=100W and Pr=0, again accepted as indicating that the load
and calibration impedance of the '43 are similar, and probably quite
close to the nominal impedance, close enough for the purpose at hand.

Understanding what is available at Walt's bench, Pf could be off by
5W or 5%. The 0 reading also suffers the same 5W indeterminacy if it
is measured on the same scale (assuming he is using a Bird wattmeter
with a 100W slug).

My bridge was not remarkably far off, roughly only 1% more error in
indicating the true value for resistance following corrections.

If I were to try to measure 1400 Ohms in the 20M band, I could expect
it might read 1333 Ohms by dial with an error of roughly ±50 Ohms.

So, between a Bird wattmeter, and it measuring power into a load with
corresponding percentages of error, the we have 5% for the Bird and 3%
for determining the load.

A metrologist can report these two (and there are more) sources of
error by one of two ways.
1. Absolute worst case = 8% error;
2. RSS = 3.9% error
Which one picked is a function of other factors.

It doesn't take long to accumulate error and driving out variables is
found only in the skill of the practice.

At Step 8

Is a step I have no current interest in.

73's
Richard Clark, KB7QHC