Jeff wrote:
. . .
My comments were not about properly made connectors, that will have very low
loss, but were a caution that at high UHF and above care has to be taken and
it is very easy to introduce loss. This is particularly the case when
launching from connectors onto pc boards with microstrip or joining on to
coax cable.

Having been involved professionally for over 10 years in the design of
microwave equipment I can assure you that there is an endless list of the
ways that loss can creep into a system at high frequencies.

I sense a communication problem here.

"Loss" usually implies dissipation of some of the desired power in the
form of heat, and that's the way I use the term.

However, people who spend a lot of time in a test equipment environment
often use the term to mean loss as described above, "mismatch loss", or
some combination of the two. And I've often seen people mistake
"mismatch loss" for dissipative loss. But it's important to separate the
two, since they have different causes and cures.

Dissipative loss is caused by current flowing through a resistance,
which causes power in the amount of I^2 * R being lost as heat. When
dealing with dielectric loss, it's often easier to calculate it as V^2 /
R. The mechanism is somewhat different but the end result, dissipation,
is the same.

"Mismatch loss" is entirely different, although someone making
measurements with a network analyzer or other test system won't readily
be able to distinguish it from dissipative loss. "Mismatch loss" works
like this:

Suppose we have a signal generator with a 50 ohm fixed resistive output
impedance which delivers 1 watt to a resistive 50 ohm load. If we were
to insert, say, an inductor with a reactance of 100 ohms in series
between the source and load, only 500 mW will be delivered to the 50 ohm
load. Now, this is for a lossless inductor, so there's no dissipative
loss in the inductor. Yet the inductor is said to have an "insertion
loss" of 10 * log(1/0..5) ~ 3 dB because the amount of power delivered
to the load is less than the amount which would be delivered under
matched conditions. Consider another example -- a perfect, lossless
transformer is inserted between the generator and 50 ohm load. If the
transformer has a 2:1 turns ratio (4:1 impedance ratio), 640 mW will be
delivered to the 50 ohm load. So the transformer is said to have an
"insertion loss" of ~ 1.9 dB.

The way to reduce dissipative loss is to reduce the I^2 * R or V^2 / R
product one way or another. Power companies do this by raising the
voltage for long distance transmission, thereby reducing I. We often use
a larger diameter coaxial feedline, which reduces the conductor R -- or
use an open wire line which reduces I because of its higher
characteristic impedance.

"Mismatch loss" can be eliminated entirely by adding an impedance
matching network. In the examples above, a matching network anywhere
between the generator and load which causes the generator to see 50 ohms
resistive will reduce the "mismatch loss" to zero -- that is, it will
raise the power in the load resistor back to its maximum possible value
of 1 watt.

Now when Jeff says that loss can be introduced by coax-to-microstrip
transitions, he's speaking of "mismatch loss", not dissipative loss.
There's no mechanism which would cause the I, V, or R to become high
enough in the region of a transition to cause an appreciable amount of
dissipative loss. It is, however, extremely difficult to make a
transition which doesn't introduce a different impedance in the
transition region. (And it's very nearly always a higher, or inductive,
impedance due to the fields escaping or fringing from the regions
immediately between the conductors.) I'm very familiar with this, having
designed transitions which had to maintain the proper impedance from DC
into the tens of GHz for very sensitive time domain equipment. It can,
however, usually be compensated by introducing complementary impedances
in the immediate vicinity -- this is the equivalent to providing an
impedance match. The result is a clean transition with negligible loss
-- "mismatch" or otherwise. It's not possible to compensate for
dissipative loss in this way. If such a transition shows loss in a
network analyzer measurement, for example, it's almost surely "mismatch
loss" and not dissipative loss. It won't burn anyone's fingers if a kW
is applied, and the cure wouldn't help dissipative loss at all.

"Mismatch loss" is useful mostly to people who have to stay in a fixed
50 ohm environment with no opportunity to apply impedance matching.
That's not usually the case in amateur antenna systems. So when someone,
especially someone with a test laboratory background, says "loss", it's
important to ask whether they mean dissipative loss or "mismatch loss",
since we can make the latter disappear but not the former.

So, I'll ask: Was the 2 dB connector loss dissipative (caused by an
exceptionally high resistance series conductor and/or exceptionally
lossy dielectric), or was it "mismatch loss" (caused by a dramatic
impedance change within the connector due to a changed relationship
between conductors -- say, missing dielectric or something like a
helical conductor)?

Roy Lewallen, W7EL