Topband: Polarity and Phase

[Jim Brown]

On Wed, 14 Apr 2004 23:00:42 -0700, Michael Tope wrote:

>
>> shift, at every frequency. We know, of course, that the phase
>> relationships are a big part of what produces the patterns of our
>> antenna arrays.
>>
>
>I think this last statement is a red herring, Jim. While it is true
>that a feedline will have a different phase shift at every
>frequency, that doesn't mean that this phase vs. frequency
>behavior is a prerequisite for synthesizing an antenna pattern.

No, it does not. But it IS one way that it is done. And my point is
that the phase shift achieved in that line is DIFFERENT from changing
the polarity.  I could also have used the example of a passive network
(LC, RC, or RL) designed to produce a phase shift. The phase shift of
all of those networks is a function of frequency. Yes, we can design
all-pass networks with some constant delay or phase shift within a
given range. But my point is that ALL of these are DIFFERENT from a
polarity change. Not better or worse -- that depends on what you're
trying to achieve with any given system/design/application. But they
are DIFFERENT. 

>It simply means that real antennas will in most cases have finite
>pattern bandwidth.

You seem not to understand an important part of how many directional
antennas work.  Take a simple 2-tower AM broadcast array. The
directivity of that antenna in the horizontal plane is entirely the
result of the fact that every listener receives signals from both
towers, and that the two signals are out of phase with each other to
varying degrees.   The designer will have achieved the desired
directivity by his/her placement of the towers -- the angles between
them, their relationship to desired directions -- by the RF level fed
to them (how deep will the nulls be), and the phase difference between
the RF fed to them. You can move the nulls around by doing nothing more
than varying the phase.  

Take a horizontal dipole at some distance above a reflective earth. It
is the phase (and amplitude) difference between the direct radiation
from the antenna and the radiation reflected by the earth that provides
the vertical directivity of that antenna! 

>Yes, everything you say is correct, but as Tom points out, you
>can still null two "in-phase" signals by flipping wires 

Cummon -- say polarity. It doesn't hurt. :)

>at the
>point where the baseband signals are combined the same
>as you can null the signals by adding a 1/2 wavelength long
>transmission line in front of one of the receivers or at the
>output of one of the two LO's.

Yes, you can. But as my AM broadcast directional antenna example shows
(you drive through the null and the carrier nulls but the sidebands
don't), polarity is different from phase. If the null were achieved by
adding two coherent signals with a polarity reversal of one of them,
ALL of the signal would have nulled, except for any differences between
them. An interesting way of hearing this is to listen to the playback
of a mono LP with a stereo needle with the two channels combined to
mono. If you add them in polarity, much of the record scratch and
tracking distortion goes away. If you add them out of polarity, the
music goes away and all you hear is tracking distortion, record
scratch. What music remains is the result of small imbalances between
the two channels -- amplitude and phase differences in the preamp
gains, alignment of the stylus, etc.  

>> This is also incorrect. When you the oscillators (in a CW or SSB radio)
>> are on different frequencies, you are "uncorrelating" the signals, and
>> "phase" has no meaning. When they are free running but fairly stable,
>> you may, by some happy accident, be able to get them close (even rather
>> precisely) to the same frequency.
>>
>
>Not really, Jim. Take an old rig like a Drake R-4C, put it in
>SSB mode, and then tune it to an AM signal. If you are
>careful, you can adjust the VFO for an almost perfect zero
>beat. As the phase/frequency of the PTO in the receiver starts
>to drift from this point (which it invariably will do), you'll be able
>to hear the carrier phase start to rotate relative to the PTO
>frequency. At first it starts out very slow when the frequency
>offset is sub-hertz then gradually speeds up as the PTO drifts
>off frequency. When the offset gets large enough, you'll no
>longer be able to count peaks and troughs in the detected
>audio and you'll start to hear a steady beat note (probably at
>10 to 20 Hz). I find it helpful to picture a rotating vector. The
>position of the vector is the phase of the signal, and the
>angular velocity of the vector is the frequency of the signal.
>In any case, that slow beating you hear when the offset is
>very small is the phase difference between the AM carrier
>and the PTO rotating around the unit circle. In this case, the
>phase definitely has meaning. At large offsets, the phase
>is rotating so quickly that all you hear is the "beat note".

There is a trig identity that shows that there is a beat note produced
by this combination. I wouldn't call that a phase difference. Phase is
the angular difference between two vectors (phasors) that are rotating
at the same frequency. When they aren't at the same frequency, you've
got some complex relationship between them, but it isn't phase.  In the
above example, the carrier produces one beat, and each modulation
component produces another. The result is akin to modulation. 
>
>> The use of "phase" when we mean "polarity" is an unfortunate
>> anachronism, and laziness, and it confuses people. Yes, when you and I
>> grew up, it was common usage. But as engineering has matured, we have
>> learned to use the right word to describe polarity separately from
>> phase.
>
>Mathematically, flipping polarity is the same as adding a 180
>degree phase shift to one of the local oscillators. The two are
>indistinguishable. In other words if I have two phase locked
>SSB receivers, flipping the "polarity" of the audio output
>of 1 receiver has the same effect as shifting phase of one
>of the two LO's by 180 degrees.

ONLY if we are talking about a sine wave of a single frequency.

>> There is also the issue of time. Add some delay, and you have added
>> phase shift that is directly proportional both to time and frequency.
>> This is not polarity either, nor is it phase. It is time.
>
>Yes, you are correct that a constant time delay produces a
>phase shift that varies linearly with frequency. But, if the
>bandwidth of the signal is small relative to the carrier
>frequency, then a time delay element (such as a matched
>transmission line) produces - to a good approximation -
>a fixed phase shift.

Again, it depends on the bandwidth of the signal and/or the system.  

>> Polarity is not a function of space, or of time, or of frequency. These
>> are fundamental concepts, and understanding how they interrelate is key
>> to our understanding of how antennas (and lots of other systems) work.
>> It is unprofessional to make that understanding more difficult by using
>> the wrong words to describe what's going on.
>>
>
>Yes, I agree. The use of precise language in engineering is
>very important, especially when large sums of money are
>involved :)

And it also helps when we are trying to understand how our antennas
work, how diversity systems can be made to work, etc. K9DX and W8JI
have noted some ways that diversity systems can be used, and I'm
enjoying the opportunity to learn from them. W0UN posted some really
neat stuff. That's why we're all here -- to learn new stuff from each
other. But to do that, we have to understand what we're doing and
saying as precisely as possible. It ain't just when big bucks are
involved -- it counts when we design our antennas and our stations to
squeeze the last couple of dB out of a path that is right on the edge.
:) 

73,

Jim