Topband: Stub measurement error

[Tom Rauch]

> Here are some measured numbers.  The transmission measured 
> frequency of this
> RG-213 stub is 28.535 MHz.  The reflection measured 
> frequency is 28.785 MHz.

By transmission and reflection, are you referring to S21 and 
S12?

If so, how could it be different with a symmetrical T? The 
fact it is different, if you are using a symmetrical T, 
means something is wrong in the test setup.

> The reflection measured frequency with the tee is 28.380 
> MHz.  The stub has
> 35.8 dB loss at the null frequency.  If the reflection 
> measurement is used
> to cut the stub, the attenuation at the desired frequency 
> will be about 3 to
> 4 dB less.  That is not acceptable in my book.

That 35 dB wouldn't be true in an antenna system or even 
into a dummy load with a transmitter and receiver, because 
the real system does not behave anywhere like a pure 50 ohm 
source and load like a good generator or S parameter test 
set looks like.

Let's look at an example. Say we have a 14.025 signal and 
want to null the second harmonic out of a 28.050 receiver. 
The 14MHz shorted stub would indeed see around 50 ohms at 14 
with a good antenna, but at 28 the impedance where the stub 
is connected could be anything. Since the source of the 
harmonic isn't 50 j0 or anywhere close at 28 MHz, and since 
the load isn't 50j 0 or anywhere close, the attenuation 
would be totally different than we measured in an S 
parameter test set anyway. So why pick nits?

If we really want to squeeze the last dB out a stub has to 
be tuned in the actual working system. If we do that and 
move 200kHz on 14 MHz,  it's all out of whack anyway because 
the harmonic moved 400kHz..

The bottom line here is if we need 35 dB attenuation and are 
depending on a single stub and can't live with 30dB, we'll 
only get that through pure dumb luck in the working system. 
It won't repeat what we measure no matter what because the 
system radically changes.  If a degree of line length 
affects the system's performance....it's time to look at the 
design. A degree is about one inch on ten meters, or a foot 
and a half on 160.

73 Tom