Tuned vs. Desired RF Frequencies (niTuner_setFreqScanList)

[Ken W]

I'm using Labwindows/CVI 8.0 along with the PXI-5600/PXI-5620 cards to capture RF frequencies using scan lists. The returned (from niTuner_setFreqScanList) actualRFTunedFrequencies are so far removed from my input desiredRFfrequencies that I'm not getting the data I want.  For example, if I pass a desiredRFFrequency of 6.25E6 with a span of 500.0E3 the resulting desiredRFfrequency is 5.0E5.  With a span of 500.0E3 that won't collect my desired frequency.

I tried this using the MultiSpan example program and it doesn't work.  It's fine if you capture over large spans (e.g. 20 MHz), but I can't seem to capture a narrow span around a desired frequency.

Any suggestions?
Thanks,
Ken

[Ken W]

I see a typo in my original post.  It should have said "if I pass a desiredRFFrequency of 6.25E+6 with a span of 500.0E+3 the resulting desiredRFfrequency is 5.0E+6"

[Andy_Hinde]

Hello Ken,
The PXI-5600 has an RF tuning resolution of either 5 MHz or 1 MHz, meaning the RF center frequency of the PXI-5600 steps in 5 MHz or 1 MHz increments.

If you set the span input to < 10 MHz, the 5600 will tune in 5 MHz steps.
If you set the span input to >= 10 MHz, the PXI-5600 will tune in 1 MHz steps.

This is to optimze phase noise performance of the PXI-5600.

Since the RF tuned frequency is not exact, the compensation is made via the IF center frequency out. For example, if an RF center frequency of 100 MHz is requested, since 100 MHz is a multiple of 5 MHz, the IF center frequency out of the 5600 will be exactly 15 MHz.

In your case you request a RF center frequency of 6.25 MHz. The 5600 tunes its RF center frequency to the nearest 5 MHz increment, which is 5 MHz. 5 MHz RF input is converted to a 15 MHz IF output frequency, so your 6.25 MHz signal will be at 16.25 MHz out of the PXI-5600. Your IF center frequency is 16.25 MHz in this case. The NI-Tuner VIs should output this to you.

Hope this help.

Regards,
Andy Hinde
National Instruments

[Ken W]

Thanks Andy, that helps explain what I was observing.  So basically, I need to use a large enough span so that my desired frequency is within +/- 1/2 the span of the 5 MHz tuning resolution?  To capture 6.25 MHz, I'd need to use a span of at least 2.5 MHz so that the range of frequencies returned from my niScope_Read call will include everything from 3.75 MHz to 6.25 MHz?  Is this correct?

I was previously using a 20 MHz span (and a 10 KHz BW) and it wasn't a factor. Let me explain my requirement and hopefully you can further enlighten me.

I'm trying to detect peak levels on 12 frequencies spread over multiple 20 MHz bands (normall 6 bands, as each band may contain 1 - 3 of the desired freqs).  Some initial testing determined that I need to use a much smaller bandwidth (between 300 Hz and 1 KHz) to get my required resolution. But collecting six 20 MHz spans of data at those bandwidths has caused a big performance hit.  Each call to tune (SmtConfigureSpectrumForRFSA) requires about 40 msec, each call to niScope_Read 20 - 50 msec, each SmtZoomFFT 80 msec, and each SmtSpectrumPeakSearch 10 msec.  When that's done 6 times (for each 20 MHz band) it isn't quite meeting my performance requirements (sampling each frequency 2 times/sec).  That's when I started trying to speed up things by collecting smaller spans.

1) Does my approach sound reasonable or is there a significantly faster approach I could take?

2) Is it true that the 5600 will collect the full 20 MHz every read and the software specified span doesn't change that?

3) Is there any way I can improve the performance of the niScope_Read?

4) It seems like my best performance gain would be to work on my calls to SmtZoomFFT so I'm only zooming on subsections of the entire span (those of interest)?  Am I heading in the right direction with this thought?

Thanks,
Ken

[Andy_Hinde]

Hi Ken,
You shouldn't have to change the span of your power spectrum. I should clarify that the span input to the niTuner VIs does not affect the actual bandwidth of the PXI-5600 downconverterd IF output. This output is always 20 MHz wide, with the span setting used simply to set the PXI-5600 tuning increment. The IF center frequency will move around according to how I described earlier.

Let me see if I can answer your questions:
1) Does my approach sound reasonable or is there a significantly faster approach I could take?
I don't know of any faster approaches you could take here.

2) Is it true that the 5600 will collect the full 20 MHz every read and the software specified span doesn't change that?
The PXI-5600 is like an analog signal conditioner that performs downconversion and signal gain/attenuation. It is analog in and analog out and will always pass out of the IF output a 20 MHz wide signal centered at 15 MHz. This 15 MHz center is sometimes the actual IF center frequency of the signal of interest, and sometimes the IF center is shifted to account for the 5600 tuning steps. But, the actual analog output of the 5600 into the 5620 is 20 MHz wide.

The SW specified span controls the span used in the Zoom FFT algorithms and makes the FFT/power spectrum as efficient as possible. The power spectrum span is not always 20 MHz. The spectrum span is the span used in the Configure RFSAfor Spectrum routine.

3) Is there any way I can improve the performance of the niScope_Read?
The performance of the read routine will be linked to the requested RBW. The resolution of a power spectrum is dictated by the total acquisition time, so a low enough RBW will set a minimum acq time. What RBW are you using, and what spectral window? You can see what the NI-Scope Read sample rate and number of samples is (i.e. acquisition time) and see if this correlates with the amount of time that the Read function is taking.

4) It seems like my best performance gain would be to work on my calls to SmtZoomFFT so I'm only zooming on subsections of the entire span (those of interest)?  Am I heading in the right direction with this thought?
That sounds reasonable.

Can you send me the frequencies of the peaks you need to measure? Are you just acquiring data, converting to frequency domain, and then doing peak searches? What dynamic range / noise floor do you need (you mention earlier you had to tweak the resolution)? I can look and see if I have any recommendations.

Regards,
Andy Hinde
National Instruments