Fw: Re: [Test-Equipment] Spectrum Analyzer BW measurement

[Gary Schafer]

Kees,

If you are looking at a signal that has its band edges roughly -55 db 
down then the -3 db points on the filter will still come into play. At 
the low level you are still using approximately the -3 db points on the 
filter. The wider part of the filter comes into play when the signal is 
strong out at the edge of it. If the signal was pretty much a "square 
block" (side bands are full amplitude and suddenly drop to zero at 
edges) then you would have to measure out on the filter skirt.

You can easily see what the skirt width is on your filters by looking at 
  an unmodulated carrier as discussed before. Find how many db down you 
are trying to measure  and then find the bandwidth on the filter at that 
many db down.

But keep in mind that it is going to be upside down (backwards?) from 
what it seems it would be. In other words, the farther down in db the 
side bands are that you are trying to measure the higher up you would go 
on the filter curve. This giving a narrower filter response at that point.

Does that make sense?

73
Gary  K4FMX



windy10605@juno.com wrote:
> Wait a minute, that "actual BW is roughly the measured BW minus the RBW
> applies only to the -3dB point". The -55dB point of the internal
> oscillator, when measured with a 500Hz RBW is 3.2Khz and when measured
> with a 5Khz RBW it's 30Khz. Approx a 10x relationship which is the
> specified RBW ratio. So now how do I correlate to a 3Khz RBW measured on
> a different SA ? 
> 
> 73 Kees K5BCQ
>   
> --------- Forwarded message ----------
> From: windy10605@juno.com
> To: garyschafer@comcast.net
> Cc: test-equipment@mailman.qth.net
> Date: Fri, 30 Jan 2004 13:32:56 -0600
> Subject: Re: [Test-Equipment] Spectrum Analyzer BW measurement
> 
> OK, now we're starting to get somewhere. I appreciate the input supplied
> by SEVERAL people, Jim, Bob, Tom, Gary, etc.
> 
>>from Gary:......With the 2 khz filter the band width of the signal
> appears to be 12 khz wide. With the 5 khz band width filter that same
> 
>>signal appears > to be 15 khz wide.
> 
> 
> So the Actual BW is very roughly: Measured BW minus the RBW. This is one
> reason why the RBW/BW ratio should be only a few %. If you are trying to
> measure a 15Khz BW, a 300Hz or 500Hz RBW is much preferred over a 3Khz or
> 5Khz RBW.
> 
>>from Gary ....To see the shape of the filter in the analyzer just tune
> into an unmodulated carrier or the zero frequency signal (local
> oscillator) >on the analyzer. With a 10 khz filter switched in you will
> see a signal that looks like it is 10 khz wide. Switch to a 500 hz filter
> and the signal >now looks to be 500 hz wide (at the -3 db points). A
> steady unmodulated signal has zero bandwidth (unless it is noisy) so any
> apparent >width that you see is the filter width. Good way to check your
> internal filters!
> 
> Sure enough, the -3dB points of the internal osc signal are approx 500Hz
> apart for a 500Hz RBW and approx 5Khz apart for a 5Khz RBW. Cool ! That
> also implies that RBW isn't adjusted/compensated for internally. The only
> "automatic" function is filter selection.
> 
> So I'm going to assume that the guys who spec'd -55dB for 20Khz BW using
> a 3Khz RBW did so 1) because that instrumentation is more readily
> available, 2) however, their BW reading is going to be approx 3Khz wider
> than actual.
> 
> 73 Kees K5BCQ
>