Jump to content
  • 0

Resolution of Spectrum Analyzer


StanS

Question

When changing the start and stop frequencies of the spectrum analyzer the resolution does not seem to change. I'm using the software in the demo mode. Is this how the actual software works. One would hope that the resolution frequency would decrease (become a smaller frequency) as the frequency span is decreased. Thanks, for any responses.

Link to comment
Share on other sites

21 answers to this question

Recommended Posts

Hi @attila. I'm satsfied and have placed an order for the Discovery2 and some accessories. Looking forward to the performance of the real thing. Thanks again for your patience and help in my understanding of this device. StanS

Link to comment
Share on other sites

Hi @attila

I'm OK with the results. However it could perhaps be improved Let me share two screen shots of the same settings except for the fft span:

The top one shows fft results of a span of 1kHz and the bottom of a span of 2 kHz. I hope you can see the difference in the frequency display. The top show frequencies from 0 to 1 kHz and the bottom from 99 kHz to 101 kHz. I;m using wf 3.7.15 beta 74 bit Qt5.6.3 Windows 10. Thanks, StanS

image.thumb.png.6b531c840c653c39a383d053116d2065.png

image.thumb.png.0e2edf9745f2f8763ac283a4e83930c7.png

Link to comment
Share on other sites

Hi attila.

I got the  results below twice. I reran the wf gen in between, but no change. I understand the fft frequency scale must be adjusted so that its center is at the specified center frequency. I like your scans a whole lot better. Why the difference?

StanS 

 

image.thumb.png.5f09526001b2611f31c55969428a46dc.png

Link to comment
Share on other sites

This is all very interesting, attila. I don't get the "distortion" you show and thus don't get the fft results either. Interestingly (perhaps) the fft baseline frequencies remain centered at 100kHz until the span is reduced to under 1kHz and at 500 Hz span jump to the numbers on your shot - 0 to 500 Hz. Where do the responses at 100 and 350 Hz come from? Why does the center frequency change from 100 kHz to 250 Hz as the span is reduced? It appears that there are some bugs in the software. But this is really pushing pretty hard on this little box. It's truly impressive in other regards - nothing that I know of like it. Is the software updated from time to time and can it be uploaded to the Discovery 2? Thanks for your help, StanS

Link to comment
Share on other sites

HI @StanS

The FFT resolution is the sampling rate divided by the number of samples or top frequency per bins. The number of samples can be power of two values and bins is half of the time domain samples.
Under the Spectrum gear options (top right edge) you can select CZT algorithm which lets you select arbitrary number of bins or resolution, but this provides only cosmetic improvement.

Under WaveForms Settings/Device Manager select the 2nd device configuration which provides 16k scope samples, the first-default provides 8k. This will improve the resolution by 2.

You could use the FFT in Scope and record more samples for better resolution. 
Enter power of two samples number otherwise CZT will be used which might introduce some artifacts in the result.
The recording should work up to 1-2MHz, above this the data stream might fail. 
Under Channel options set sample mode Decimate because the Averaging attenuates the signal.

image.thumb.png.c0abf40f3789609ffb92d74a09d73d69.png
image.png.44fcd61506cc8e3d7b515c78afab486f.png

Link to comment
Share on other sites

I understand about bins and windows and appreciate their value. What I have observed is that with a constant number of bins the apparent bin spacing does not change in proportion to the span. I'm using the "auto" mode of the analyzer. I'm interested in making narrow bandwidth analysis of signals up to 1MHz, meaning down to a few Hz. Will Discovery 2 do that?

Thanks again for all the interest shown. Stan

Quote

 

 

Link to comment
Share on other sites

@JColvin, @attila,

FFT resolution is nominally given by the sample rate divided by the FFT size--not quite by the number of samples but the "number of samples" is usually close enough.  This metric is based upon a rectangular window (filter) which has *horrible* sidelobe performance.  Better sidelobe performance can be achieved with a better filter, often at the expense resolution--examples include Hanning and Blackman windows which will widen the response of any FFT to a tone by between 5 and 9 FFT bins at only 50dB below the carrier.

Better windows are available if you'd like them,

Dan

Link to comment
Share on other sites

Hi @StanS,

My understanding (though clearly limited) is that the resolution is calculated by taking the sampling rate and dividing it by the number of samples, as per @attila's comment in this thread. I am not certain if the sampling rate (or some sort of sampling divisor) changes at all though; I'll defer to attila on that information.

Thanks,
JColvin

Link to comment
Share on other sites

Thank you both for your two very different replies. I took some data to illustrate my problem. This is still demo mode.

I set the center to 100k, varied the span and recorded the resolution:

span 50k  resolution 30.5 Hz;   20k, 26.9;   10k, 25.6;   5k, 25.0;   2k, 24.7;   1k, 24.5.

So changing the span by a ratio of 50 changed the resolution from 30.5 Hz to 24.5 Hz. That's not what I expected. I expected to see the same ratio between changes in span to changes in resolution. There must be come limiting factor at work.

Please help me to understand the situation.  Thanks, Stan

Link to comment
Share on other sites

Hi @StanS,

I apologize for the delay in responding. The resolution does change in the Spectrum Analyzer when connected to a real device when changing any of the frequency range parameters (start, stop, center, span, BINs), though this does not appear implemented in demo mode.

Let us know if you have any other questions. (and defer to Attila for the most accurate information since they have infinitely more experience with WaveForms than myself).

Thanks,
JColvin

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...