R
Member
Bursty signals are where spectrum analyzer settings quietly lie to you. I spent a morning chasing an intermittent spur that only showed up on some sweeps, and the culprit was my own RBW/sweep-time combination. Notes from the bench:
First, understand what sweep time actually does. With a swept analyzer, each frequency point is only observed for a fraction of the sweep. If the burst duty cycle is 1% and your sweep is fast, most sweeps simply miss the burst. The analyzer is not broken; it is just looking elsewhere when the signal fires. Before touching anything else, check the duty cycle of what you are hunting.
Second, RBW sets the noise floor and the dwell time. Narrow RBW lowers the displayed noise and helps you see small spurs, but it also forces a longer sweep (the analyzer needs more time per bin). Wide RBW sweeps fast but raises the noise floor. My rule of thumb: start wide and fast to find the burst, then narrow down once you know the frequency. Do not start narrow — you will sweep past the signal while it is silent.
Third, use max hold, but understand its cost. Max hold over many sweeps is the simplest way to catch a burst, but it also accumulates noise peaks and makes everything look worse than it is. I run max hold for detection, then switch to a single sweep with a triggered capture to measure the real level. Two different jobs, two different settings.
Fourth, video bandwidth matters more than people admit. With VBW much narrower than RBW, the trace gets smoothed and short bursts get averaged down — sometimes below the noise. For burst hunting, keep VBW at least equal to RBW until you have the signal pinned down.
Fifth, if the analyzer supports it, use a real-time or FFT-based capture for intermittent work. A swept measurement fundamentally samples in time; an FFT capture watches the whole span at once. The price is a higher noise floor per bin, but for "is it there at all" questions, that trade is usually worth it.
One open question for the group: when you hunt intermittent emissions, do you trust max hold plus a slow sweep, or do you go straight to a triggered time-domain capture? What has actually caught the elusive one in your lab?
First, understand what sweep time actually does. With a swept analyzer, each frequency point is only observed for a fraction of the sweep. If the burst duty cycle is 1% and your sweep is fast, most sweeps simply miss the burst. The analyzer is not broken; it is just looking elsewhere when the signal fires. Before touching anything else, check the duty cycle of what you are hunting.
Second, RBW sets the noise floor and the dwell time. Narrow RBW lowers the displayed noise and helps you see small spurs, but it also forces a longer sweep (the analyzer needs more time per bin). Wide RBW sweeps fast but raises the noise floor. My rule of thumb: start wide and fast to find the burst, then narrow down once you know the frequency. Do not start narrow — you will sweep past the signal while it is silent.
Third, use max hold, but understand its cost. Max hold over many sweeps is the simplest way to catch a burst, but it also accumulates noise peaks and makes everything look worse than it is. I run max hold for detection, then switch to a single sweep with a triggered capture to measure the real level. Two different jobs, two different settings.
Fourth, video bandwidth matters more than people admit. With VBW much narrower than RBW, the trace gets smoothed and short bursts get averaged down — sometimes below the noise. For burst hunting, keep VBW at least equal to RBW until you have the signal pinned down.
Fifth, if the analyzer supports it, use a real-time or FFT-based capture for intermittent work. A swept measurement fundamentally samples in time; an FFT capture watches the whole span at once. The price is a higher noise floor per bin, but for "is it there at all" questions, that trade is usually worth it.
One open question for the group: when you hunt intermittent emissions, do you trust max hold plus a slow sweep, or do you go straight to a triggered time-domain capture? What has actually caught the elusive one in your lab?
