On Thu, 29 Apr 2004 02:17:12 GMT, "The other John Smith"
wrote:

"Richard Clark" wrote in message
.. .
On Wed, 28 Apr 2004 22:18:19 GMT, "The other John Smith"
wrote:

A good tip and one I'll adopt. I do have three 50 Ohm loads. I don't have
mismatched loads, but I can parallel 50 Ohm loads to make a 25 Ohm load.
Come to think of it, 75 and 93 Ohm terminators are available through Mouser.
Perhaps I should invest in some.

This does not replace testing however. Care should be taken to not
heat the loads as this will cause them to drift in value. Use them
with 1/10 rated power dissipation capacity.

Further, you should also swap the dual coupler input/output and the
side arm ports to confirm it is in fact operating at fixed ratio (this
says nothing of the presumed ratio, but is still a necessary step).
Then repeat the paragraph above concerning the 50 Ohm load rotation.
You can then proceed to confirming the coupling coefficient which may
surprise you (you haven't given the pedigree of this particular
beastie).

Yes. I did not mention that I would do that, but my plan was to compare the
forward and reverse coupled outputs and record the ratio of the two.

This goes further than that, reverse the input/output ports as well.
ALL ports must be tested against every other port. Along with the
loads, this becomes an increasingly larger matrix of readings - it is
all necessary. This is about Orthogonality because as a bridge this
cannot be taken for granted.

I don't
think the absolute value of coupling (supposedly -30 dB) is needed for now.
The Narda coupler model is not listed at Narda's Web site, so I'm going by
the marking on the case.

OK, so it is a Narda. They are typically very good from 10% to 90%
over the band of operation (if you also use the correction factors).

I have no way of calibrating the Vector Voltmeter.

Such is life. As you are only employing it to make relative
measurements, what you DO need to calibrate is the Gain/Attenuator
range control. Construct or obtain a good step attenuator. This, of
course re-introduces the issue of calibrating it, but if you can
reduce variables and check its DC and low AC operation, you are fairly
down the road ahead of bonehead problems. If you have fixed
attenuators that are of quality, you can do the RF side through
substitution.

I presume you have undamaged leads for your meter (I won't even
presume they are calibrated, but for these purposes, testing should
reveal problems if they are not sufficient to the task).

They are undamamged as best as I can tell. Both channels read the same and
agree with a homebrew power meter. I realize the power meter is probably
less accurate than the voltmeter.

So many mistakenly infer the opposite.

I presume your source offers a 50 Ohm output Z. There is nothing like
a mismatched source looking at a mismatched load to increase confusion
by the square.

What is the output impedance of an amateur transmitter?

50 Ohms at rated output - just like the manufacture specifies. Other
useful power settings are not terribly far off (somewhere between 30
and 80 Ohms).

I see arguments all the time about this.

Nonsense is free of charge, and worth every penny spent.

And even so, what will be the result if it is not
matched? If there are reflections from the source, it will show up in the
composite forward voltage which will be the reference anyway. Is this not
so?

In SPADES! A source that is 2:1 mismatched feeding a load that is 2:1
mismatched runs the minimum error of your measurement upwards to 20 or
30% [It could be worse, I'm too encumbered to find my reference to
quote accurately.] This is why I said your low-ball acceptance of
10-15% error was stardust.

Is my equation correct and are the comments associated with it correct?

I'm focusing on technique here. I presume you've done your math
right. The tougher part is realizing it accurately in a world of
error.

It appears that you have given this some thought. Thank you.

Hi John,

I spent a few years as a Metrologist specializing in RF power
measurement. Very, very few know the term (even engineers) - think
National Bureau of Standards (or NIST). In practical terms, this
means I have calibrated every item in front of you (and much more).

So, with all the caveats in place, you can easily step right into it
by abandoning them all and noting some values. Then you can implement
each of these issues and observe how much the data changes. This is
the merit of knowing your equipment and your skill. No measurement is
right the first time, nor the second, and rarely the third time.
Besides, it is more fun to refine your bench and testing and close in
on the actual value by hammering out error. The bench artist will
even disturb the set up with a known error, and observe how it
propagates through to the final determination.

73's
Richard Clark, KB7QHC