🚀 go-pugleaf

In sci.electronics, usenet@lonex.rl.af.mil writes:

    Short of buying a pad, does anyone know a good (cheap) way of matching a
    cable tv coax (75ohms) to the input of a HP spectrum analyzer (50 ohms)?
    I'm interested in the 500MHz to 1GHz range.


    Thanks
    BRO'HERN (Electronics Engineer)

For narrow-band applications, it's possible to design impedance transformers
using trasmissions lines (e.g., a quarter wave transformer or open/shorted
stubs).  However, if you really are interested in the full 500MHz-1GHz band,
it is not easy to make a *good* impedance transformer.  The easiest way
of doing this would be to use lumped trasformers wound on ferrite cores
(TOKO makes some good relatively high frequency parts).  The problem will
be getting a good low-loss transformer with flat frequncy response.

Have you considered just tieing the 75 ohm cable directly to the 50 ohm
input?  This will give you a return loss of -14 dB , which isn't terribly
bad, depending on how accurately you're trying to measure the 75 ohm signal.
(The non-perfect match at the spectrum analyzer will create a ripple pattern
 in the response versus frequency.  Worst case ripple will be 
 +/- 20*log(1- .2 * rho) where rho is the reflection coefficient of the
 75 ohm source)

Another idea -- don't laugh -- is to tie a 25 ohm resistor in series with
the input of the spectrum analyzer.  What this does is provied a perfect
load match for the 75 ohm source but worsens the source match as seen
by the 50 ohm load.  The 50 ohm load will see a source match of -10 dB
which is poor, but this isn't a problem as long as you keep the distance
between the 50ohm input and the 25 ohm resistor very small.  The 25 ohm
resistor will give you 1.8 dB of added loss.

In short, if you really want *good* measurements without messing around
with these schemes, I'd go with the minimum loss pad.  The -5.7 dB loss is
a small price to pay for a good, flat response and good return loss.

Chris