AnsweredAssumed Answered

MXA FFT mode spurs

Question asked by wheel on Jun 25, 2007
Latest reply on Jul 6, 2007 by wheel

Like • Show 0 Likes0
Comment • 7

Currently our company is evaluating the new MXA and after some short testing, we found some internal spurs showed up in low frequency under FFT mode. (ie. center 15.4Mhz, span 25Mhz, RBW 1k, FFT 25Mhz).   Our experiment was done with terminator, and the spurs are at least 10db higher noise floor. (the case above is not the worst case, we saw more serious ones)   We double checked with VSA connected with MXA and turned out the same result.   Also, the spurs will change in some frequency pattern when changing the center. So it is more like internal LO mixing product. However that did not happen on Sweeping mode.

The initial response from local representatives stating that this is not a issue and claim the MXA in FFT mode only reach -100dbm.   Maybe I am wrong, but someone can show me why an internal spurs is not an issue, and where the -100dbm comes from.

... wheel

No one else has this question

Outcomes

7 Replies

mike.wohlert Jun 25, 2007 4:53 PM
Mark Correct
Correct Answer
Hi,

I set up this measurement just as you specified: center 15.4Mhz, span 25Mhz, RBW 1k, FFT 25Mhz. While I noticed a spur at 15.8 MHz, it was at -115 dBm, which is 15 dB below the MXA's residual response specification of -100 dBm. There could be a number of things causing this including mixing products, something in the ADC, etc. Regardless, this spur is still within the guaranteed specification of the MXA (see page 38 of the MXA specs guide: You can find this by going to the Agilent web page and searching for MXA, and then clicking the product library Tab).
Also, I found that if I changed the span to 5 MHz or changed the center frequency to 16 MHz, the spur went away, giving me a nice clean noise floor.

-Mike
Like • Show 0 Likes0
Actions
wheel Jun 25, 2007 6:56 PM
Mark Correct
Correct Answer
So, is that means anything I measure under -100dbm is meaningless, since it can always be some internal spurs of MXA. How can we make the correct measurement ? Please let us know how to workaround.
Like • Show 0 Likes0
Actions
mike.wohlert Jun 27, 2007 9:05 AM
Mark Correct
Correct Answer
Absolutely not. According to the guaranteed specifications, the MXA can actually measure down to -150 dBm at 15 MHz.

The "internal spurs" that you're seeing are the LO getting into the IF filter. Since you're in FFT mode with an FFT width of 25 MHz centered at 15.4 MHz, the LO is getting into the skirt of the IF filter which has a 3 dB bandwidth of 22.5 MHz. To fix this, simply change the FFT width from 25 MHz to 7.99 MHz (the keystrokes are: sweep control, sweep setup, FFT width, <7.99 MHz). Narrowing the FFT width keeps the LO from getting into the IF filter and gets rid of the spurs you were seeing. Another fix is to increase the center frequency to 16.5 MHz. This moves your IF filter far enough away from the LO so that none of it gets into the IF filter.
Like • Show 0 Likes0
Actions
wheel Jun 28, 2007 6:57 AM
Mark Correct
Correct Answer
Thanks for the detail. It make us more comfortable to know that Agilent has full confidence how it behave and where it will appear. However, can Agilent provide kind of algorithm to help us to avoid those trouble points ? Or just rule of thumb, then we don't need to check every combination we might require to fulfill our measurement.

... wheel
Like • Show 0 Likes0
Actions
YS Jul 2, 2007 8:49 AM
Mark Correct
Correct Answer
Hi
Since the internal spurs cannot be removed, what will happen when we perform digital demodulation if the spur is located in the signal band?

Let's take the previous question for example, if we are dealing with a 256-QAM signal centered at 15.4MHz, Symbol rate=20MHz, roll-off factor= 0.18.

What will happen at this situation? Will the SNR be affected? (I do believe so) What kind of phenomena will us see on the constellation and other measurements?

At this condition, what can we do to improve the measurement accuracy?

Best Regards

YS
Like • Show 0 Likes0
Actions
Frank_BR Jul 4, 2007 2:38 PM
Mark Correct
Correct Answer
"
Since the internal spurs cannot be removed, what will happen when we perform digital demodulation if the spur is located in the signal band?

YS
"

The internal spurs (residual responses is the usual name for them) are a problem only with very weak signals (below -100 dBm) and BW smaller than 10 to 100 kHz. Remember that even an ideal spectrum analyzer, free from residual responses, will be limited by noise. That is, ALL spectrum analyzers have limitations for analyzing very week signals.

One way to circumvent residual responses issues when analyzing very weak signals is to use an amplifier between the signal source and the spectrum analyzer. Most spectrum analyzers have a preamplifier as an option.
Like • Show 0 Likes0
Actions
wheel Jul 6, 2007 10:52 PM
Mark Correct
Correct Answer
My problem is the spec and how to move on. Selling something 50db higher DANL spectrum should be much cheaper. I understand now internal spurs has its own spec -100dBm. And DANL is different and which is much lower. But if there is no document stating where those internal spurs will come out, then unless we do all combination measurement first with terminator on to record everything, otherwise the following measurement will be only meaningful > -100dBm. Not even mention how to resolve if we do need some measurement which does contain those spurs. Adding amp, changing center, changing FFT size are not responsible answer for a customer point of view.
I am aware such limitation and willing to live with them because the spec said so -100dBm, but we need to know more detail in order to increase measurement efficiency, usability and certainty.

... wheel
Like • Show 0 Likes0
Actions

MXA FFT mode spurs

Outcomes

Related Content

Recommended Content