Author Topic: Wide Bandwidth Signals and Spec An RBW  (Read 1213 times)

0 Members and 1 Guest are viewing this topic.

Offline g0mgx

  • Regular Contributor
  • *
  • Posts: 56
  • Country: gb
Wide Bandwidth Signals and Spec An RBW
« on: September 14, 2017, 06:51:57 pm »
Hi All

I'm relatively new here and am looking for help understanding something I have seen on my bench today.

I have been constructing a Digital Ham TV trasmitter, the bandwidth of the output signals are very wide - typically 2-4 MHz.

When I look at the signals on the Spec An I am seeing a drop in signal amplitude displayed by the instrument when the RBW is less than the signal bandwidth. I can't understand why this is. All on-line examples I can find use RBW to resolve close frequency signals, but my case I just have one very wide bandwidth signal. As my Rigol Analyser doesn't have a RBW wider than 1 MHz available, does that mean I cant see the correct amplitude of a 2 MHz wide signal at all?

Can anyone cleverer than I help me understand this concept?

Many thanks.

Mark
G0MGX
 

Offline danadak

  • Super Contributor
  • ***
  • Posts: 1875
  • Country: us
  • Reactor Operator SSN-583, Retired EE
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #1 on: September 14, 2017, 07:12:57 pm »
This might help -


http://www.ni.com/white-paper/3983/en/


What Rigol scope do you have, model # ?


Regards, Dana.
Love Cypress PSOC, ATTiny, Bit Slice, OpAmps, Oscilloscopes, and Analog Gurus like Pease, Miller, Widlar, Dobkin, obsessed with being an engineer
 

Offline g0mgx

  • Regular Contributor
  • *
  • Posts: 56
  • Country: gb
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #2 on: September 15, 2017, 05:06:36 am »
Hi Dana

That article is one that i already found in my search for an understanding. It talks about resolving close together signals and the impact on a wide RBW when the signals are closely spaced.

It doesn't talk about what happens when the signal itself has a wider bandwidth than the RBW and how that impacts the amplitude. My understanding is that the signals within the RBW as the spec an scans will be integrated to a single amplitude which is the value plotted. This has the effect of "spreading" the actual signal; the illustration used by W2AEW in this video is excellent:

https://youtu.be/Ffhs9Ny03lM

Once again, however, it doesn't consider the scenario where the signal bandwidth itself is wider than the RBW filter.

I have a Rigol DSA815TG and also two HP 22GHz analysers - 8562A and a 8593A. All of these instruments exhibit the same symptoms of showing the amplitude of the signal to be lower than reality when the RBW is narrower than the signal itself, but I really don't think I understand why.

My world of madness is documented here:

http://g0mgx.blogspot.co.uk/

Mark
G0MGX

 

Z80

  • Guest
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #3 on: September 15, 2017, 08:50:39 am »
Hi, it is essential to understand how the spectrum analyser works and what the trace is displaying to make sense of what is on the screen.  You have seen W2AEW's video which does a good job of describing what the instrument does so that should give you a clue as to why your readings are different (hint RBW has nothing to do with the bandwidth of signal you can display).  By default, the trace displays the average power contained within the current RBW window and that is the level displayed.  It should be clear if you think about this for a second that the trace isn't therefore giving you a detailed 'level' indication at each spot frequency, but a smeared out average of everything it can 'see'.  To get a more accurate representation of the actual signal you would reduce the RBW to narrow down the window which gives less averaging for each spot frequency, in fact to get the actual level would require an infinitely small RBW but in reality this isn't possible and the error with a realistic RBW is usually good enough.  The reason you have variable RBW by the way is for speed.  Narrow filters are slow to respond so wide ones are used when scan speed is required (try running a 1GHz span at 10Hz RBW and you will see).  Depending on what you are trying to measure, there are other types of detector available like peak, minimum etc to get round the resolution problem, you need to know the instrument and the measurement you are trying to make otherwise it will fool you.  The Rigol actually has reasonable help screens which give a description of the various functions which are ok for getting an idea of what the settings are for and it's well worth spending a bit of time to explore them.
 

Offline EverySignal

  • Newbie
  • Posts: 1
  • Country: us
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #4 on: September 16, 2017, 01:40:37 am »
Hi G0MGX,

I don't know what the signal power you expect to see, say x dBm and what is the drop, say y dB. I understand you find the signal power is  a few (??) dB down from your expectation, or the real value, when the signal BW (2-4 MHz) is much wider than your RBW.

Here is one case I have. I use Keysight SA to check WiFi 20MHz signal. My RBW is usually set as 100KHz. The actual occupied BW of this 20MHz signal is about 17MHz. Let's say this WiFi signal power is -10dBm (it is generated by an EXG). I will not see -10dBm on the SA. What I will see is -10 - 10*log10(17000000 / 100000) = -32.3dBm.

I'm not sure if it helps. But this is a case where RBW is much less than signal BW.
 
The following users thanked this post: g0mgx

Offline metrologist

  • Super Contributor
  • ***
  • Posts: 1832
  • Country: 00
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #5 on: September 16, 2017, 01:22:19 pm »
What is the modulation format you're using, something like 8VSB?

Channel power measurements typically use Average/RMS detection rather than peak detection and are integrated over the entire bandwidth. Some analyzers have this function built in.

As your analyzer sweeps the span, the detector samples the signal within the RBW filter at that particular frequency. If you are modulating, the signal level at a particular frequency is not constant over time, so when you sample, there may or may not be as much signal level at that time. When you increase the RBW, you open the sample window and will see more signal, along with more noise.

These are somewhat related and may provide some insight.

http://literature.cdn.keysight.com/litweb/pdf/5966-4008E.pdf
http://www.repeater-builder.com/test-equipment/aeroflex/pdfs/accurate-power-measurements-using-spectrum-analyzers.pdf
https://cdn.rohde-schwarz.com/pws/dl_downloads/dl_application/application_notes/1ef48/1EF48_2E.pdf
http://defenseelectronicsmag.com/site-files/defenseelectronicsmag.com/files/archive/rfdesign.com/images/archive/302Gorin32(1).pdf

 
The following users thanked this post: g0mgx

Offline g0mgx

  • Regular Contributor
  • *
  • Posts: 56
  • Country: gb
Re: Wide Bandwidth Signals and Spec An RBW
« Reply #6 on: September 16, 2017, 03:22:49 pm »
Thanks for all the replies.

The fundamental issue here is the type of signal I am trying to measure. The DVB-S digital video signal is effectively noise, so I tried using a simple noise source on my Spectrum Analyser and that produces the same results.

The attached image shows the same signal source at 300 KHz RWB in yellow and 3 KHz RBW in purple.

The difference is 10 Log (300000 / 3000) = 10 log 100 = 20dB and that is near enough the delta in amplitude I am seeing.

Because of the signal type, when I reduce the RBW the amount of energy in that width reduces also hence the integrated amplitude displayed by the instrument follows.

I think I understand now; the basic understanding I was missing was the type of signal I was measuring. Clearly, had the signal been a solid RF carrier then the amplitude wouldn't alter.

Mark
G0MGX
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf