Nonlinear RFSA 5665

[1111111111111111]

Hello!

I work with spectrum analyzers NI 5665 and 5663e.I have a question about FFT-width. I understand from the specification to the devices FFT width is fully controlled.However, for different values of the width of the FFT I get different results. (See image). That is, when FFT width 50 MHz I watch the hills.

From which they come, I do not understand. This fact affects the accuracy of my measurements.For me, it would be optimal to use the FFT width of 50 MHz (my devices support this feature.) FFT width of 50 MHz yields the best results in speed of processing for me.

For research, I used the generator R&S SMF100A.Frequency step 1MHz(500kHz,100kHz). Time to step 300ms. Level-20dBm.

For RFSA windows type - flat top(other settings on images).

Cable to connect the analyzer and generator used the tried and tested on the network analyzer to the R&S ZVB.

Very good that your company provide the software to LabVIEW. Very nice and convenient to manage devices using LabVIEW 🙂 

[1111111111111111]

Continued images

[1111111111111111]

I'll get the answers?

[dbell0971 · Accepted solution]

Hi 111...,

Thank you for the compliment on LabVIEW, I do prefer LabVIEW and it's ability to manage a large number of devices from even just one VI.  I've always loved LabVIEW, even before working here at NI.

To your question: if you look closely at the "hills", you'll notice that they occur about every 50 MHz.  If you set the acquisition to acquire more than a 50 MHz span, the 5663E and 5665 will take several acquisitions with 50 MHz bandwidth and combine them to make the larger acquisition.  For more information on this, check the first paragraph after the first two "notes" in this article:

RFSG Help: FFT Width Property

http://zone.ni.com/reference/en-XX/help/372058H-01/rfsapropref/pnirfsa_fftwidth/

No filter (or device for that matter) has a perfectly flat response.  The ideal filter would be the gray line below, while a practical FFT looks more like the black line.  According to the product page of the 5663E and 5665, they have a 3 dB 50 MHz instantaneous bandwidth.  The 3 dB is key; this means that the signal at 50 MHz is attenuated at 3 dB relative to what it's actual value is.  If you take a look at your screenshots, it looks like the roll off on the hills is only around 0.2 dB, much better than the expected 3 dB.

Hope this helps!

Thanks,

David B

National Instruments

Applications Engineer

[1111111111111111]

"...The increase in bandwidth results in faster signal acquisitions, but amplitude accuracy is decreased for spectrum acquisitions and magnitude and phase accuracy is decreased for I/Q acquisitions..."-It seems that this problem is unbeatable 🙂

Do you want fast - lose accuracy. Want accuracy - then wait 🙂

Thanks for the competent answer, David!

I wanted to ask about the relationship instantaneous bandwidths and FFT-wide, but you gave a link to the desired material.

Good luck and success in your work!