Tom Rauch wrote:
>> Many other types of directional coupler such as the Bruene bridge work
>> on the same principle - the only difference is in the methods used to
>> get the capacitive and inductive signals, and to combine them with a
>> choice of phases.
>
>It is electrically impossible to have a single-point measurement in a
>transmission line that measures forward and reflected power
>without sampling both voltage and current.
>
Quite - no argument here.
>I am amazed that some claim current can flow two directions at the
>same instant of time in a conductor,
Obviously the real-world physical current at any given instant is only
flowing one way - but it can be *mathematically* resolved into
components of "forward" and "reflected" current flowing simultaneously
in opposite directions. Likewise the instantaneous net voltage can be
mathematically resolved into components due to the "forward wave" and
"reflected wave".
This mathematical stuff had better work, because without it you can't
prove the statement about forward and reflected power readings!
>> In the Bird, these two separate RF signals add or subtract at the
>> output of the pickup loop, and the resultant voltage is detected by
>> the diode. The instrument is designed so that when it is terminated in
>> exactly 50 ohms, and the slug is rotated to the 'reverse' position,
>> the two voltages are exactly equal and opposite, so the meter displays
>> zero reflected. That's what gives the instrument its directional
>> properties.
>>
>> So far, so good, but...
>>
>> When the slug is turned round to the 'forward' position, the two RF
>> signals add in phase. We agree that one signal is proportional to
>> voltage, and the other is proportional to current - but the meter
>> displays the rectified SUM of these two signals. To give a resultant
>> that is truly proportional to power under all circumstances, they
>> would need to be multiplied - which they ain't.
>>
>> That's why I still don't understand how the Bird can indicate a true
>> difference between indicated forward and reflected power with all
>> kinds of terminations. As I said at the start, I don't dispute Tom's
>> experimental data, but I still don't think we have a solid theoretical
>> backup for it.
>
>The original Bruene description will explain it.
Thank you - looking for that reference in Walt Maxwell's 'Reflections'
(1st edition) pages 3-17 to 3-18 actually have most of what we need.
This may be more accessible than Bruene's article, referenced as April
1959 QST, or Johnson's book on transmission lines which is also
referenced.
It makes sense now. I can now see why any correctly functioning and
calibrated directional wattmeter really *does* show the net power flow
as the difference between the forward and reflected power indications,
regardless of the terminating impedance.
But it certainly wasn't obvious! Not physically obvious, anyway. It's
one of those surprisingly simple results that occasionally fall out of a
rather complex-looking mathematical analysis.
There are a few loose ends, like the need to prove that it works the
same at any position along a transmission line, not just at voltage
maxima and minima where E and I are non-reactive... but I'm content to
believe that's provable too. Also Maxwell uses a term Zc which he
describes as the "line impedance", which doesn't cover the case where no
line exists, either inside or outside of the directional coupler... but
it still works if Zc is described as the impedance for which the
directional coupler was designed (to give zero reflected indication when
matched) and at which the power calibration was made.
OK, I'm happy to have learned something - except...
If you want to convince anyone that "Net power flow is the difference
between forward and reflected power readings, regardless of the
terminating impedance", you first need to say: "This isn't immediately
obvious, but it can be shown that..."
--
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
New e-mail: g3sek@ifwtech.co.uk
New website: http://www.ifwtech.co.uk/g3sek
|