Spectrum Analyzers different displays

[notaroketscientist]

What explains the different in trace shapes between a traditional CRT SA, a digital LCD SA, and the SignalHound?
The skirts of the center frequency appear differently in each type of scope.
So is this the result of the driving circuit's properties,  the displays, or differences in data acquisition methods?
How does one interpret the apparently different results obtained looking at identical signals?
Are there differing applications which would lead to choosing one from the others?

[Psi]

You'd probably get the same range of differences testing two different CRT SA's (or two LCD SA's)

I doubt the fact that its CRT or LCD has any bearing on it. Its just different design specs / SNR etc..

[KJDS]

The shape is determined by the filter function used and by the phase noise of the analyzer and input signal.

Change the Residual bandwidth on an analyzer and you'll see the shape change.

[cyr]

Some good reading on the subject:

http://cp.literature.agilent.com/litweb/pdf/5965-7920E.pdf

[notaroketscientist]

So the resolution bandwidth is responsible for the parabolic trace at the center frequency so often displayed on a DSA815?
Should I conclude then the analog CRT based scopes have a much finer resolution bandwidth explaining the very sharp spike like display and
the SignalHound must then have the finest resolution bandwidth to create its almost single line display of the center frequency?
In each case the noise component is distinctive but in the analog crt scopes often displays with an increasing average in the vicinity of the
center frequency.

How does one avoid misinterpreting displayed regions on either side of the center frequency as the amplitude diminishes?

[notaroketscientist]

Some good reading on the subject:

http://cp.literature.agilent.com/litweb/pdf/5965-7920E.pdf

Thank you for this very helpful paper understanding the elements which determine the display characteristics.
From this it would seem the analyzer which has the finest display would be most helpful in characterizing the noise level and level of
close in signals.

[KJDS]

Some good reading on the subject:

http://cp.literature.agilent.com/litweb/pdf/5965-7920E.pdf

Thank you for this very helpful paper understanding the elements which determine the display characteristics.
From this it would seem the analyzer which has the finest display would be most helpful in characterizing the noise level and level of
close in signals.

Providing that the internal phase noise is up to the job, and that has nothing to do with the display.

[ConKbot]

So the resolution bandwidth is responsible for the parabolic trace at the center frequency so often displayed on a DSA815?
Should I conclude then the analog CRT based scopes have a much finer resolution bandwidth explaining the very sharp spike like display and
the SignalHound must then have the finest resolution bandwidth to create its almost single line display of the center frequency?
In each case the noise component is distinctive but in the analog crt scopes often displays with an increasing average in the vicinity of the
center frequency.

How does one avoid misinterpreting displayed regions on either side of the center frequency as the amplitude diminishes?

check out

   for how the filters affect the measurements.


Spec-ans with the same specs should produce the same result.   The DSA815 has a rather dismal RBW of 100 Hz, so you wont be able to resolve any finer than that.   The signal hound USB-SA124B has an RBW of 0.1 Hz, so you should be able to get in real close to narrowband signals.  Not sure what your analog CRT supports, I'd expect down to 10-1 Hz or so.  That will give you the inverted parabolic shape.  Once youre in closer than that, the slope coming down the sides of the signal is the phase noise.  Either the phase noise of the source, or the phase noise of the SA if you have a nice low phase noise source.   


The signal hound and a lot of modern SA's have a FFT input, where if youre under a certain span, it digitizes the downconverted signal samples it for a bit, and runs a FFT on the acquired data. I.e. the signal hound boasts "I/Q Data up to a 240 KHz bandwidth" So they probably have a 1-2MSPS ADC (maybe faster, didnt check ), and they can run a FFT on the data on that.   For wider sweeps than that, they can break it into 250KHz chunks and get a FFT on each, or revert to a traditional sweeping SA method. 

[notaroketscientist]

So the resolution bandwidth is responsible for the parabolic trace at the center frequency so often displayed on a DSA815?
...How does one avoid misinterpreting displayed regions on either side of the center frequency as the amplitude diminishes?

check out

   for how the filters affect the measurements.


Spec-ans with the same specs should produce the same result.   The DSA815 has a rather dismal RBW of 100 Hz, so you wont be able to resolve any finer than that.   The signal hound USB-SA124B has an RBW of 0.1 Hz, so you should be able to get in real close to narrowband signals.  Not sure what your analog CRT supports, I'd expect down to 10-1 Hz or so.  That will give you the inverted parabolic shape.  Once youre in closer than that, the slope coming down the sides of the signal is the phase noise.  Either the phase noise of the source, or the phase noise of the SA if you have a nice low phase noise source.   


The signal hound and a lot of modern SA's have a FFT input, where if youre under a certain span, it digitizes the downconverted signal samples it for a bit, and runs a FFT on the acquired data. I.e. the signal hound boasts "I/Q Data up to a 240 KHz bandwidth" So they probably have a 1-2MSPS ADC (maybe faster, didnt check ), and they can run a FFT on the data on that.   For wider sweeps than that, they can break it into 250KHz chunks and get a FFT on each, or revert to a traditional sweeping SA method.

Perfect ConKBot that is the information I was searching to find. It clears up my confusion. 

[w2aew]

The Resolution BW and Video BW will determine the "shape" of the signal peaks and the cleanliness of the surrounding noise - which can also be affected by the detector used.  Beyond that, differences in the skirts of the noise surrounding the signal peaks will be governed by the phase noise of the analyzer.