On Tue, 21 Dec 2004 21:16:03 GMT, (Robert
Lay) wrote:

It has been suggested that a
virtue of the T pad would give a clean 6 dB loss instead of some
"not-so-nice" loss, like 5.7 dB. Well, that's not a problem because
the L-Pad can also be designed for exactly 6 dB. In fact, it can be
designed to provide ANY loss you want, so long as it's AT LEAST 5.7
dB. You see, the 5.7 dB L-Pad happens to be the MINIMUM loss design
for that particular mismatch.

Hi Bob,

The difference between 5.7 and 6dB is immaterial if neither is
calibrated. As for the desire for a nominal 6dB pad, that too is
hardly of great merit for quality measurements with a good Spectrum
Analyzer.

Just like directional coupler design, precision applications focus on
directivity and sacrifice round numbers in coupling to achieve better
separation of ports. A 6dB attenuator will isolate your precision
gear from the unknown better than a 3dB attenuator, but not as well as
a 10 or 20dB attenuator.

Now, as to the term isolation. It has a variety of meanings which in
this case means that your measurement is less perturbed by the literal
unknowns of your proverbial unknown being measured. That is, in your
attempt to find a value (the proverbial unknown) your accuracy can be
upset by variables whose magnitude can affect that accurate
determination. Large attenuators obviously de-sense the
instrumentation, but if you have sufficient dynamic range, then that
is not a debit, but actually an asset. Hence de-sense or isolation is
benign. When this large attenuator is placed on the source, it
reduces the load's influence to pull or mismatch there too.

This says nothing of actual mismatches, it simply presents what is
called swamping. That is, you introduced known and controlled losses
to buffer the measurement. Later you can subtract out the losses to
find your proverbial unknown.

I already alluded to the virtue of using attenuators to increase the
power tolerance to the input of a Spectrum Analyzer, aside from this,
the only practical use of attenuators is to introduce controlled loss
to isolate the unknowns' influence.

To answer "Why Match?" returns us to isolation. Once this is
achieved, the measurement can be trusted to be faithful in proportion
to that isolation. With this example of a simple 50 to 75 Ohm
conversion, that measurement's faithful accuracy is fairly good. As
for it being a ~6dB attenuator, by placing it into another test with
an unknown, it will offer mixed results - a T or PI configuration at a
higher attenuation would be far more flexible, and faithful.

73's
Richard Clark, KB7QHC