I spent three hours last Tuesday on a ridge in the Catskills, staring at a waterfall of noise that looked exactly like a signal, only to realize I’d forgotten to account for my own local oscillator leakage. It’s a rite of passage, but it’s also incredibly frustrating when you’re trying to figure out how to use a spectrum analyser to actually validate your antenna’s performance rather than just chasing ghosts. Most of the manuals will tell you to adjust the span and the RBW like you’re following a recipe for sourdough, but they never tell you why the noise floor suddenly decides to jump up ten dB just because you moved your hand near the chassis.
I’m not here to give you a lecture on theoretical signal processing or walk you through a menu of buttons you’ll never press. Instead, I’m going to show you how to actually read the data so you can distinguish a real station from a momentary fluke of the ionosphere. We are going to talk about real-world settings, the impact of your attenuation, and why your measurement is only as good as the ground you’re standing on. If you want the truth about what’s happening in the air, let’s get to work.
Table of Contents
- Mastering Frequency Domain Analysis Beyond the Manuals Lies
- Signal Amplitude Measurement That Actually Matches Reality
- Five Ways to Stop Chasing Ghosts and Start Seeing Signals
- The Reality Check: Three Things to Remember When You’re Looking at the Screen
- Stop Chasing Ghosts
- Putting the Data to Work
- Frequently Asked Questions
Mastering Frequency Domain Analysis Beyond the Manuals Lies

The manual will tell you that a peak on the screen is a signal. That’s a dangerous assumption. When you’re performing actual frequency domain analysis, you have to distinguish between a legitimate carrier and a transient spike caused by your own switching power supply or a neighbor’s poorly shielded LED driver. I spent three weeks chasing a phantom signal on the 40m band last summer, only to realize my analyzer was picking up the internal clock noise of my laptop because I hadn’t properly grounded the chassis. If you aren’t looking at the noise floor relative to your target, you aren’t measuring anything; you’re just looking at pretty pictures.
To get any real value out of signal amplitude measurement, you need to stop treating the span and resolution bandwidth (RBW) as static numbers. If your RBW is too wide, you’re going to smear the signal and lose your ability to see what’s actually happening in the weeds. I always tell people to narrow that RBW down until the noise floor settles; if the peak stays sharp while the floor drops, you’ve found something real. If the peak widens and starts bleeding into the surrounding noise, you’re likely looking at a low-amplitude signal struggling to survive a bad environment, not a clean transmission.
Signal Amplitude Measurement That Actually Matches Reality

The biggest mistake I see people make—and I’ve seen it in plenty of club meetings—is assuming the peak on the screen is an absolute truth. It isn’t. When you’re performing signal amplitude measurement, you have to account for the fact that your settings are constantly shifting the goalposts. If you haven’t adjusted your reference level or your attenuation, that “strong” signal might just be your analyzer hitting its ceiling and clipping. I once spent an hour trying to troubleshoot a “noisy” transceiver on a portable setup, only to realize my input attenuation was set so low that I was essentially blinding the front end.
You also can’t ignore the relationship between your span and your resolution bandwidth (RBW). If your RBW is too wide, you’re just averaging the signal with the surrounding noise, which gives you a bloated, inaccurate reading. For real signal-to-noise ratio optimization, you need to narrow that window down. I always tell my students: don’t trust a peak until you’ve tightened the RBW and watched how the noise floor reacts. If the floor stays steady while the peak grows, you’ve found something real; if they both move together, you’re just looking at a mathematical artifact.
Five Ways to Stop Chasing Ghosts and Start Seeing Signals
- Watch your Span, not just your Center. If you’re zoomed in too far, you’ll mistake a single carrier for a wideband signal; if you’re too wide, you’ll miss the sidebands that tell you if your amplifier is actually linear or just screaming into the void. I’ve spent too many evenings chasing a peak that turned out to be nothing more than a single, lonely tone because I forgot to look at the bigger picture.
- Stop trusting the “Auto” button. Most modern analyzers have an auto-setup feature that’s designed to make the screen look pretty, not to give you accurate data. It’ll adjust your reference level and span to make the noise floor look low, but it’s often masking the very interference you’re trying to find. Set your reference level manually—about 10dB above your expected signal—and work your way down from there.
- The Noise Floor is your real baseline. Don’t just look for a spike; look at what’s happening in the “empty” spaces. If your noise floor is jumping around, it’s either your equipment’s thermal noise or you’ve got something nearby—like a cheap switching power supply—polluting your local environment. If you can’t see the floor, you can’t trust the signal.
- RBW (Resolution Bandwidth) is the difference between a signal and a smear. If your RBW is too wide, two closely spaced signals will look like one big, fat blob. If it’s too narrow, you’ll be sitting there for twenty minutes waiting for the sweep to finish. I usually start wide to find the neighborhood, then tighten the RBW to actually see the shape of the signal.
- Account for your attenuation before you fry something. I’ve seen enough people walk straight into a high-power transmitter with a sensitive spectrum analyzer and wonder why their screen went blank. Always use a calibrated attenuator if you’re testing anything more than a low-power SDR or a battery-operated rig, and for heaven’s sake, check your input limits first.
The Reality Check: Three Things to Remember When You’re Looking at the Screen
Stop chasing ghosts by ignoring your noise floor; if your signal peak is barely distinguishable from the baseline, you aren’t seeing a transmission, you’re seeing a statistical fluke or a bit of atmospheric luck.
Never trust a measurement in isolation from your physical setup; a signal that looks perfect on the analyzer means nothing if your antenna is only 2 meters above the ground and your pattern is completely distorted.
Ditch the “textbook” settings for real-world hunting; adjust your span and resolution bandwidth until the signal actually looks like a signal, not a blurry smudge that tells you nothing about the actual bandwidth or stability.
Stop Chasing Ghosts
A spectrum analyser isn’t a magic wand that tells you what’s happening; it’s just a window into the noise. If you aren’t adjusting your span and your reference level to actually see the floor, you aren’t measuring a signal—you’re just watching a pretty pattern dance around a number you guessed.
Wren Castellano
Putting the Data to Work

At the end of the day, a spectrum analyser is just a very expensive way to confirm what your ears might already be telling you, provided you aren’t being fooled by your own settings. We’ve covered why you can’t trust a single peak without checking your span and resolution bandwidth, and why a signal that looks massive might just be an artifact of a poorly set noise floor. Remember: if your measurements don’t account for the actual environment—the height of your antenna, the local interference, and the current state of the band—you aren’t measuring physics, you’re just measuring your own mistakes. Stop chasing ghosts and start verifying the baseline before you trust a single decibel reading.
I know it can feel overwhelming when you first move from looking at a simple signal strength meter to staring at a complex waterfall display, but don’t let the math intimidate you into staying in the dark. The goal isn’t to become a human calculator; it’s to develop the intuition to know when a reading is real and when the ionosphere is just playing tricks on you. Radio is one of the few places left where you can actually see the invisible, and once you learn to read the spectrum for what it truly is, the whole world opens up. Now, get off the screen, go outside, and see what’s actually happening out there on the airwaves.
Frequently Asked Questions
If I'm looking at a signal on my SDR, how do I know if that noise floor is actually my local environment or just the limits of the hardware itself?
You have to baseline the hardware. Turn off your antenna, disconnect the feedline, and cap the input—or better yet, use a 50-ohm terminator. If that floor stays exactly where it is, you’re looking at the quantization noise or the thermal limits of your SDR’s ADC. If it drops, you’re seeing local RFI. I’ve seen people chase phantom signals for hours, only to realize their cheap LNA was just screaming at its own internal noise.
When I'm trying to find a weak signal, should I be adjusting the span or the resolution bandwidth first to avoid missing a transient peak?
Start with the span. If your span is too wide, you’re just looking at a haystack and wondering why you can’t see the needle. Get the window tight around your area of interest first so you actually have the visual resolution to see anything. Once you’ve centered on the right neighborhood, then you tighten the RBW to pull that signal out of the noise floor. Just don’t expect a miracle if your noise floor is already sitting too high.
How much does the actual physical placement of my antenna—and its height above the ground—skew the amplitude readings I'm seeing on the screen?
It skews them more than most people realize. If you’re measuring signal strength with a spectrum analyzer but your antenna is sitting ten feet off the ground, you aren’t seeing the true potential of that frequency. Ground plane interaction and near-field effects will absolutely mess with your amplitude readings. I’ve seen people think they have a dead band, only to find that raising the antenna another six meters brings the signal out of the noise floor entirely.




































