It's not clear to me what noise you're trying to measure. If you're
trying to measure the thermal noise of a good resistor, you're SOL --
anything you try to measure it with will add its own noise, which you'll
end up measuring instead. You can easily calculate the thermal noise,
though. If you're trying to measure the internal noise of the spectrum
analyzer, the answer is on the screen and, as you've found, it's a
function of the filter bandwidth. If you're trying to measure the noise
coming from a receiver, it has to be considerably greater than the SA's
internal noise, or you'll never see it. That means you'll have to
amplify it until you can see it, with an amplifier that's quiet enough
that the noise it adds isn't significant. The noise figure of the SA
itself is typically much worse than a decent receiver -- it's been
traded for dynamic range.

Once you get whatever noise you're trying to measure up to the level you
can see it on the SA, you'll find that its magnitude also varies with
the SA filter bandwidth, if the noise is broadband. You can get any S/N
ratio you want, if you can tolerate an arbitrarily narrow bandwidth. A
real receiver has some particular bandwidth to accommodate the incoming
signal. It's usually this bandwidth at which the S/N ratio is measured.

Roy Lewallen, W7EL

Bill B. wrote:
Hello,
I need some help in figuring out how to properly use a Spec. Analyzer.
I have done several hours of research online now, and have somewhat of
an idea of what I am doing, but I just can't get down one piece of the
puzzle. In my line of work, we use simple devices that give us the
noise floor level in dB. I (stupidly) expected to be able to get this
same information easily out of a SA. As you may have guessed by now, I
ran into a problem when switching the RBW value. After much... much
reading, I fully understand *WHY* it changes (Filters increasing in
size cause a greater internal noise level, etc.) but I can't figure
the best way to get a "base" noise level reading. And as much as I
would like to fully understand the theory, what I really need is a few
more examples... IE: If you are looking a 100 MHz span and your RBW is
1MHz simply ___Fill in the Blank____ to find the base noise floor. To
further explain what I am trying to do, we are setting up a link that
requires a SNR of at least 6dB... I need to get the noise floor level
to compare with my projected signal strength for this link.

Please be nice... This is my first post grin

Thank you! Thank you!

Bill B. - N1SNI

"Roy Lewallen" wrote in message
...
[...snip...]
If you're trying to measure the noise
coming from a receiver, it has to be considerably greater than the SA's
internal noise, or you'll never see it.


Assuming classical "noise" as we are all probably thinking...
First, a minor correction to the above "considerably greater", Roy. If the
EXternal noise is just 6dB above the SA's INternal noise (I guess 4 times
can be considered 'considerable'), then the resulting displayed value will
be only 1 dB higher than the incomming noise actually is. Noise power (when
combining in one receiver) simply adds. To see the math, think this way. A
1 dB increase amounts to 1.25 times the power. That extra 25 percent is one
fourth the original power. 25% of the power is 6dB down from the original.
SO adding some power which is -6 dB (or 25% the original) will raise the
total 1 dB.

However, The poster used the term "noise", but doesn't know the source of
his problem. His data hits can be caused by other than what we are thinking
is noise. It can be (sounds like since it recently started) spurious
signals in his channel OR it could be IM being caused from strong signals
getting into his receiver. The spurious in-band stuff could be narrow
signals or wideband. Typical CDMA signals behave much like simple wide-band
noise as we think of noise. A narrow band signal/spur is different.

From a later post I see that he is trying to measure some junk that has
appeared on the link frequency causing interference. In any case, the
front-end noise floor (noise figure) of the SA will determine how far down
you will be able to see and if it is not low enough, then you won't see the
interference. I can not comment on the Advantest sensitivity. If you have
access to a low noise amplifier of the type used in your system, it can be
used in front of the SA and *MAY* raise the sensitivity (lower the noise
figure/floor) enough to see yout interference.

Good luck. We've had a strange low level (around -123 dBm)spur on our
repeater input since last fall and can't locate it...wish I had lotsa time.
--
Steve N, K,9;d, c. i My email has no u's.