Resolution bandwidth (RBW) is the smallest frequency difference a spectrum analyzer can distinguish between two signals. It’s determined by the width of the internal filter the analyzer uses as it sweeps across a range of frequencies. A narrower RBW lets you see finer detail in the frequency spectrum, but it comes with trade-offs in measurement speed and complexity.
How RBW Works Inside a Spectrum Analyzer
A spectrum analyzer works by sweeping a filter across a range of frequencies and measuring the power at each point. The width of that filter is the resolution bandwidth. Think of it like looking at a landscape through a window: a wide window shows you a broad, blurry view, while a narrow window lets you pick out individual details but takes longer to scan the whole scene.
When the analyzer encounters two signals that are close together in frequency, the RBW setting determines whether they show up as two distinct peaks or merge into one. Two narrow signals can only be separated if the resolution bandwidth is smaller than the frequency gap between them. If the RBW is wider than that gap, the filter covers both signals at once as it sweeps past, and they appear as a single combined peak on the display.
RBW and the Noise Floor
Resolution bandwidth has a direct, predictable effect on the noise floor of your measurement, sometimes called the displayed average noise level (DANL). A wider RBW lets more noise energy through the filter, raising the baseline. A narrower RBW lets less noise through, lowering it.
The relationship follows a simple rule: every tenfold change in RBW shifts the noise floor by 10 dB. Switching from a 10 kHz RBW to a 1 kHz RBW drops the average noise level by 10 dB, making it easier to spot weak signals that were previously buried. Going the other direction, widening the RBW from 1 kHz to 10 kHz raises the noise floor by 10 dB. This makes RBW one of the most powerful tools you have for pulling small signals out of the noise.
The Speed Penalty of Narrow RBW
Narrower isn’t always better, because RBW has a dramatic effect on how long a measurement takes. For traditional analog-style filters with a near-Gaussian shape, sweep time follows this relationship:
Sweep time = k × (span) / RBW²
The key detail is that RBW is squared in the denominator. That means reducing the RBW by a factor of 10 increases the sweep time by a factor of 100. If a measurement takes one second at 10 kHz RBW, switching to 1 kHz RBW pushes it to about 100 seconds. This is why you want to use the widest RBW that still gives you the frequency separation and noise floor you need, rather than defaulting to the narrowest setting available.
RBW vs. Video Bandwidth
Resolution bandwidth and video bandwidth (VBW) are both essential spectrum analyzer settings, but they do very different things. RBW controls frequency selectivity: how well you can distinguish closely spaced signals and how much noise passes through the measurement filter. It shapes the raw measurement data.
VBW, by contrast, is a post-processing smoothing function. It averages the displayed trace after the measurement has already been made, reducing the visual “fuzz” of noise on the screen. Narrowing the VBW makes the trace look cleaner and easier to read, but it doesn’t actually change the analyzer’s ability to resolve signals or its true sensitivity. RBW changes what you measure. VBW changes how the result looks on screen.
Choosing RBW for Pulsed Signals
Continuous signals are forgiving when it comes to RBW selection, since the signal is always present and fully charges the internal filter regardless of its width. Pulsed signals are trickier. The RBW filter needs enough time to respond to the signal during the pulse, which means the pulse has to be long enough to fully “fill” the filter. If the RBW is too narrow, the filter never reaches the signal’s true amplitude before the pulse ends, and your amplitude reading comes in low.
On the other hand, setting the RBW too wide raises the noise floor and can mask low-level components like harmonics and spurs. The practical approach is to start wide and narrow the RBW until the resulting filter response fits within the pulse width without exceeding it. For example, when measuring a pulsed radar signal, a 30 kHz RBW might produce a filter response of about 74 microseconds, which fits neatly inside the radar pulse. That would be the narrowest usable RBW for that particular pulse width, giving you the best balance of sensitivity and accuracy.
Filter Shape Options
Not all RBW filters behave identically. Most spectrum analyzers offer a choice of filter shapes, each suited to different measurement goals. The Gaussian filter is the most common default. It provides good frequency separation between closely spaced signals, moderate amplitude accuracy, and high dynamic range thanks to its low side lobes. For general-purpose measurements, it’s the standard choice.
Flat-top filters sacrifice some frequency resolution in exchange for better amplitude accuracy. If your priority is measuring the exact power level of a signal rather than separating it from a nearby neighbor, a flat-top filter is more appropriate. The trade-off is straightforward: Gaussian for resolving signals, flat-top for measuring their precise amplitude.
Practical Guidelines for Setting RBW
In most situations, you’re balancing three things: the ability to separate closely spaced signals, the noise floor, and measurement speed. Start with the default RBW your analyzer selects automatically (most instruments choose one based on your frequency span), then adjust from there.
- To separate two close signals: Set the RBW narrower than the frequency spacing between them.
- To find weak signals near the noise floor: Reduce the RBW. Each tenfold reduction drops the noise floor by 10 dB.
- To speed up a measurement: Widen the RBW. Each tenfold increase cuts sweep time by roughly a factor of 100.
- To measure pulsed signals accurately: Match the RBW so the filter response fits within the pulse duration.
The core principle is that RBW is never a “set and forget” parameter. The right value depends entirely on what you’re trying to measure, how fast you need the answer, and how much noise you can tolerate in the result.