SPECTRUM ANALYZER - Best choice of RBW and VF?
Mario Emmanuel
I would like to know how to use the RBW control and the VF filter in a
spectrum analyzer. I don't understand well how such controls affects
measurements.
Is normal that amplitude measurement varies with one or other RBW selection?
How should I measure for maximum accuracy?
Thanks for sharing your experience.
John Miles
Check out Tektronix's "Fundamentals of Spectrum Analysis" app note at
http://209.95.122.40/jm/492/fundsa.pdf. It should answer your
questions.
-- jm (What this country needs is a Napster for out-of-print app notes!)
------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------
Rick
> Hi.
> I would like to know how to use the RBW control and the VF filter in a
> spectrum analyzer. I don't understand well how such controls affects
> measurements.
The resolution bandwidth filter is like the selectivity filter in a radio;
it determines what bandwidth of signal at the analyzer's present frequency
gets fed into the power detector.
The Video Bandwidth filter is the filter *after* the power detector. it's
easiest to see the effect of the video filter by setting the analyser
to zero span with a sampling detector (not peak!) and just changing the
video bandwidth. You don't even need to have a signal going into the analyzer,
just look at the effect on the noise-floorr; it gets filtered just like
a normal audio signal would be, but the level shouldn't change.
> Is normal that amplitude measurement varies with one or other RBW selection?
Only if you are measuring a modulated signal. A pure tone only has
one frequency component, so widening or narrowing the RBW filter just couples
the same amount of signal into the detector (with a bit more or less noise).
But if you are measuring a modulated signal, then widening the RBW filter means
that you are coupling more of the signal's power into the detector, so the measurement
increases. If you narrow the RBW filter, you're coupling less of the signal's
power into the detector, and the measurement will go down.
> How should I measure for maximum accuracy?
If you're just trying to measure signal power, then you need a wide RBW, so that
all of the energy contained in the signal gets into the power detector. Note that
if the RBW is sufficiently wide, then anything you feed through it just looks
like an impulse, and what you see on the analyzer screen then looks like the
impulse response of the filter - i.e. you lose the "detail" of the signal
you're measuring.
If you're trying to measure the modulation itself, then you need a RBW
filter thats narrow compared to the signal bandwidth, otherwise you just end
up measuring the total (wideband) power as above. Many signals are specified
in terms of a resolution bandwidth at a certain offset from the centre frequency.
> Thanks for sharing your experience.
--
Rick
Gerrit Barrere
selection?
> Only if you are measuring a modulated signal. A pure tone only has
> one frequency component, so widening or narrowing the RBW filter just
couples
> the same amount of signal into the detector (with a bit more or less
noise).
> But if you are measuring a modulated signal, then widening the RBW filter
means
> that you are coupling more of the signal's power into the detector, so the
measurement
> increases. If you narrow the RBW filter, you're coupling less of the
signal's
> power into the detector, and the measurement will go down.
This is not correct. Even with a single tone, if your RBW is too narrow for
the sweep rate you're using your amplitude and even peak frequency
measurement will be in error. The ideal frequency line will shorten,
broaden, and "lean over" as you reduce RBW at a given sweep rate, because
the IF filter doesn't have time to respond to the fast sweep.
Gerrit Barrere
Don K.
>
> Hi.
>
> I would like to know how to use the RBW control and the VF filter in a
> spectrum analyzer. I don't understand well how such controls affects
> measurements.
RBW sharpens the bandwidth and lowers the noise floor.
VF smoothes out the noise floor without changing its average value.
As you narrow the RBW or VBW, you will have to slow the sweep speed.
> Is normal that amplitude measurement varies with one or other RBW selection?
It's always a good idea to perturb the instrument settings of VBW, RBW,
sweep rate, and attenuator to verify the readings make sense and are
consistent.
> How should I measure for maximum accuracy?
First it's nice to set RBW and VBW wide enough so you can have a fast
sweep so you can see things in "real time" with no flicker.
Then make sure the spectrum analyzer is not overloaded.
Toggle 10 dB of attenuation in and out and make sure nothing changes by
more than 10 dB.
Make sure the signal you want to measure is at least 10 dB above the
noise floor. You can lower the noise floor either by changing the
attenuator, or narrowing the RBW. Also, if signals are too close together,
narrow the RBW.
If you want to read the value of the noise floor, narrow the VBW.
Don
--
To reply by email, delete the antonym for walk in d...@homerun.com
Mario Emmanuel Martinez
Thanks for your comments. I've read the PDF John suggested and carefully
read the comments of Rick, Gerrit and Don K.
I understand (more or less) the function of each control and how they could
affect measurements.
Now I think my spectrum analyser is bad calibrated (or just a piece of
rubbish) because I have the following behaviour:
With the sweep generator selected for -40dB (for example), I have the
following measurements for the selected RBW:
RBW 1MHz -45dB
RBW 100KHz -40dB
RBW 10KHz -30dB
Sweep rate is set very slow (5 seconds) and SPAN is indiferent (I tried with
values beetween 0 to 2MHz with similar results)
I get a similar (not exactly equal but similar) behaviour (decrease RBW ->
amplitude increase) reading peak amplitude of a home made colpitts
oscillator.
Please tell me if I'm wrong but I think this behaviour is not covered by any
of your comments. I think IF filters gain controls could be uncalibrated.
Can this be right?
I own a cheap spectrum analyser bought to an italian magazine, which means
they haven't the same quality calibration control that any normal company
should have. For its cost I'm happy with it but of course I would like to
get maximum available performance of it, so if someone thinks it is a
calibration problem I would be happy to know it. Any suggestion on how could
I test it better is welcomed.
Thanks.
John Miles
>
> Please tell me if I'm wrong but I think this behaviour is not covered by any
> of your comments. I think IF filters gain controls could be uncalibrated.
> Can this be right?
Sounds likely. Even if you have an unstable test oscillator (sweep
generator?), you'd expect a lower amplitude reading at the narrower
resolutions, since the signal's energy would be spread out over a wider
bandwidth. You're getting the opposite effect above, so I'd say the
analyzer either needs calibration or was never intended to offer
absolute amplitude accuracy in the first place.
-- jm
Don K.
and record a matrix of correction factors to apply for the various
combinations of RBW, VF, frequency, and span rate.
Don
Mario Emmanuel Martinez wrote:
> I own a cheap spectrum analyser bought to an italian magazine, which means
> they haven't the same quality calibration control that any normal company
> should have. For its cost I'm happy with it but of course I would like to
> get maximum available performance of it, so if someone thinks it is a
> calibration problem I would be happy to know it. Any suggestion on how could
> I test it better is welcomed.
--
Mario Emmanuel Martinez
Sounds very interesting. But I would also need a very nice RF generator
without armonics? Or should I just use my sweep generator with 0 span?
I would also ask if you recommend me to build myself a power metter using
any schottky diode. I've been reading some Agilent (I think it was Agilent)
application note but I find that linearity may be a serious problem.
Thanks
"Don K." <d...@homerun.com> escribió en el mensaje
news:3B68C05A...@homerun.com...
Don K.
Since you don't have a calibrated power meter, then you can at least make
a matrix of corrections in terms of relative power rather than absolute.
That is you can probably assume your slowest sweep, widest RBW gives
you the best relative dB accuracy. Make the matrix relative to that.
Since you can't control absolute power, just record the dB change for
each RBW, VF, and sweep speed relative to the reference setting.
Yes, set the sweeper to CW or 0 sweep width. If you were setting its
absolute power with a broadband power meter, you would want its
harmonics 25 or 30 dB down. But you're not, so harmonics are irrelevant.
I'm not sure that using a Schottky diode detector as a power meter
would do you any good. You still have no way of calibrating its absolute
accuracy.