RF meter vs spectrum analyzer: which one do you need?

If you want to know about the radio signals in your home or on your street, you'll come across two kinds of tool: RF meters and spectrum analyzers. They answer different questions, they react differently to the same signal, and their numbers are often not comparable. This guide explains how each one works, why two instruments in the same spot can disagree by a factor of ten or more, and how to choose the right one for what you actually need to know.

Three instruments, three different questions

Most confusion comes from treating these tools as interchangeable. They aren't. Each is built to answer one kind of question well and other questions badly.

Keep the three questions in mind as you read. "Where is it strongest?" is a meter question. "What is it?" is an analyzer question. "Is it within the rules?" is a question for a calibrated instrument and someone qualified to use it.

How a broadband RF meter works

Inside a typical broadband meter there are three parts: a small antenna, a detector and a display. The antenna picks up whatever is on the air across its range. The detector, usually a diode, turns that radio energy into a DC voltage. The meter then scales the voltage into a field-strength or power-density figure and shows it as one number.

The key word is broadband. The detector doesn't know or care what frequency the energy arrived at. An FM station, a cell site, your own phone and the microwave oven in the kitchen all push the same needle. The meter adds them together.

What that single number hides

• Which signal is which. A reading of 0.5 V/m could be one strong source or twenty weak ones. You can't tell from the number.
• Your own devices. A phone in your pocket, a smartwatch or a laptop on Wi-Fi can dominate the reading. You end up measuring yourself.
• Timing. Many signals are bursts, not steady tones. The number on the display depends on how the meter handles bursts, which is covered below.
• Frequency weighting. The meter is not equally sensitive at every frequency. Two signals of equal strength can read differently because one sits where the meter is less sensitive.

What it does well

A broadband meter is quick. You switch it on and walk. If the number jumps when you step towards a window, that window faces something. For finding the strongest spot in a room, or checking whether moving a router changed anything, that speed is genuinely useful. Just treat the reading as a relative indicator, not an absolute measurement.

Frequency-selective meters: the middle ground

Some meters add filters so they can report separate readings for separate bands, for example one figure for the cellular downlink bands and another for 2.4 GHz Wi-Fi. Professional versions go further and split the range into many service bands, each with its own reading.

A band-selective meter answers a better question than a plain broadband one: "how much of the total comes from cellular, and how much from Wi-Fi?" It still can't show you individual channels or the shape of a signal, so it can't distinguish two cell sites using the same band, or tell a Wi-Fi router from a Bluetooth device that shares the 2.4 GHz band. Think of it as a meter with a few drawers, not a full picture.

When a meter's specification says "frequency-selective", check what that means in practice. It might be two or three fixed bands, or it might be a full receiver with adjustable bandwidth. The latter is effectively a spectrum analyzer with a calibrated antenna attached.

Isotropic vs single-axis probes

A radio wave's electric field points in a particular direction, called its polarisation. A simple antenna is most sensitive to fields lined up with it and much less sensitive to fields at right angles to it. Indoors, reflections from walls, floors and furniture scatter signals so that their polarisation at any one spot is hard to predict.

Single-axis meters

A single-axis meter has one antenna element. Its reading depends on how you hold it. Point it one way and it may show a strong reading; turn it 90 degrees and the same field may read far lower. That's not a fault. It's how a single antenna behaves.

Isotropic (three-axis) probes

An isotropic probe has three small sensors at right angles to each other, usually labelled x, y and z. The instrument measures each one and combines them into a total that, ideally, doesn't depend on how you hold the probe. The combination is the root-sum-square:

E = √(Ex² + Ey² + Ez²)

Worked example

Suppose a three-axis probe reads Ex = 1.0 V/m, Ey = 0.5 V/m and Ez = 0.3 V/m at one spot. The total is √(1.0 + 0.25 + 0.09) = √1.34 ≈ 1.16 V/m. A single-axis meter held in line with the z sensor would show only about 0.3 V/m, which is roughly 12 dB below the true total. Held along x it would show about 1.0 V/m, close to the total.

If you only have a single-axis meter, you can approximate the isotropic total by taking three readings with the meter turned along three directions at right angles, without moving its position, then combining them as above. In a steady field that works reasonably well. With signals that change from second to second, the three readings are taken at different moments, so the result is rougher.

A single-axis reading taken in one arbitrary orientation can be badly low. If a comparison matters, either use an isotropic probe or take the maximum over several orientations and record which orientation you used.

Peak, average and RMS: how the detector decides the number

A detector turns a rapidly changing radio signal into a figure you can read. There are several ways to do that, and they give different answers for the same signal.

"Average" and "RMS" are often confused. For a steady sine wave, the relationship between them is fixed. For real signals, especially digital ones with high crest factors (a big gap between peaks and average), only an RMS detector gives the true average power. Some instruments average the logarithm of the signal, which reads lower than true power for noise-like signals. For pure noise, a log average reads about 2.5 dB below the true power average. That's one reason two analyzers with different default settings can disagree on the same channel.

Diode detectors in cheap meters

A diode detector behaves differently at different levels. At low levels it responds roughly to power, which is what you want. At higher levels it starts to respond more to peak voltage. With several signals present at once, or with digital signals whose peaks are much higher than their average, the reading can drift away from the true power. Better meters correct for this; basic ones may not.

Why pulsed signals read differently

Many modern transmitters don't send continuously. They transmit in short bursts with gaps in between. The fraction of time a signal is on is its duty cycle. When the duty cycle is low, the peak and the average are far apart.

So a meter showing peak and a meter showing average can disagree by 20 or 30 dB on the same signal, and both can be working correctly. When you report a figure for a bursty signal, always say which detector produced it.

Wi-Fi

An idle Wi-Fi access point still transmits short beacon frames, commonly about ten times a second (the usual beacon interval is 102.4 ms). Between beacons the channel may be silent. When someone streams video, the channel fills with traffic and the average rises sharply. The peak level of each burst barely changes. That's why a peak-reading meter shows similar numbers whether the network is idle or busy, while an averaging meter shows a big difference.

5G and LTE using time-division duplex

Time-division duplex (TDD) systems share one channel between the base station and the phones by taking turns in time. The base station transmits in some time slots and listens in others. The average level at your location then depends on the slot pattern, on how much traffic there is, and, with beam-steering antennas, on where the users being served happen to be. Even with no traffic, the base station broadcasts short synchronisation and control bursts at regular intervals. A peak reading captures those bursts; an average reading over a quiet period can be much lower. See 5G frequency bands explained for which bands use TDD.

Radar

Radar is the extreme case. It sends very short pulses with long gaps, so its duty cycle is tiny. Many radars also rotate, so their beam passes over you for a moment every few seconds. A peak detector shows a large figure for a fraction of a second; an average over minutes is very much lower. Weather radars in the United States use frequencies around 2.7–3.0 GHz, inside the range of the HackRF One (in Consentiva, the Full plan covers 2.7–6 GHz). A slow or averaging meter may show almost nothing; a spectrum analyzer in max-hold mode shows a clear spike that appears only when the beam sweeps past.

Missing bursts entirely

A swept spectrum analyzer looks at one part of the band at a time. If a burst happens while it's tuned elsewhere, it isn't seen. A software radio sweeping in steps has the same limitation. The fix is to run many sweeps and use max hold, or to narrow the span so each part of the band is revisited more often. This "probability of intercept" problem is why one sweep is never enough to say a signal is absent.

Frequency response, flatness and calibration

No antenna and detector are equally sensitive at every frequency. Manufacturers describe this as frequency response or flatness, often as "±X dB" over a stated range. Outside that range, sensitivity usually falls off, though strong signals just outside the range can still register at a reduced level.

What a tolerance in dB means

Decibel tolerances sound small but aren't. A ±3 dB uncertainty means the true power could be anywhere from half to double the reading. For field strength in V/m, ±3 dB is a factor of about 1.41 either way. A ±6 dB uncertainty is a factor of about four in power and two in field strength. When a specification gives flatness over a wide range, the error at any particular frequency could be anywhere in that band.

What calibration actually means

Calibration means the instrument was compared, at stated frequencies and levels, against a reference whose own accuracy is traceable to national standards. The calibration certificate lists the frequencies tested, the corrections or errors found, and the measurement uncertainty. A calibration is valid for a period the maker or the lab recommends, after which it should be repeated.

A meter that says "accurate" without a certificate hasn't been calibrated in that sense. It may still be consistent from day to day, which is enough for comparing places. It isn't enough for comparing a reading with a regulatory limit.

Calibration for software radios

A software-defined radio reports levels relative to the full scale of its digitiser, in dBFS. The relationship between dBFS and the real power at the antenna socket depends on frequency, the gain settings and the individual unit. You can estimate an offset by measuring a known signal source, which is what Consentiva's calibration offset does: once you enter an offset, levels are shown in dBm instead of dBFS. That makes readings at the antenna socket more meaningful, but it doesn't calibrate the antenna, so it still isn't a field-strength measurement. See dBm, V/m and µW/cm² explained for how these units relate.

Field strength vs power density

RF meters report either electric field strength (volts per metre, V/m), magnetic field strength (amperes per metre, A/m) or power density (watts per square metre, W/m², or the older µW/cm² and mW/cm²). They describe the same wave in different ways, but they are only interchangeable under certain conditions.

The far-field relationship

Far enough from an antenna, the electric and magnetic fields are in a fixed ratio. That ratio is the impedance of free space, about 377 ohms. Power density and electric field strength are then related by:

S = E² / 377, with S in W/m² and E in V/m.

Two things stand out. Power density goes with the square of field strength, so tripling the field multiplies the power density by nine. And the unit conversions are easy to slip on: 1 W/m² equals 100 µW/cm², and 1 mW/cm² equals 10 W/m².

Why a meter that shows both may be guessing

Most consumer meters measure the electric field and then calculate power density using the far-field formula. That calculation is only valid in the far field. Close to an antenna, a cable or an appliance, the formula doesn't hold and the displayed power density is not a real measurement.

Near field vs far field

Around any antenna there are regions where the fields behave differently. Their size depends on the wavelength and on the size of the antenna.

• Reactive near field. Very close to the antenna, roughly within one wavelength divided by 2π for a small antenna. Energy is stored and returned rather than radiated. Electric and magnetic fields are not in the 377-ohm ratio, and either one can dominate.
• Radiating near field. Further out, the wave is radiating but the beam hasn't fully formed. Readings can rise and fall over short distances.
• Far field. For an antenna that is large compared with the wavelength, the far field is usually taken to start at about 2D²/λ, where D is the largest dimension of the antenna and λ is the wavelength. Here the field falls off smoothly with distance and the 377-ohm relationship holds.

The reactive region of a small antenna is tiny at cellular and Wi-Fi frequencies. Large antennas are different. For a hypothetical panel antenna 1.3 m long working at 1.9 GHz (wavelength about 16 cm), 2D²/λ comes to about 21 m. Inside that distance, readings can vary in ways the simple distance rule doesn't predict.

What this means in practice

• Close to an antenna, power density shown by an E-field-only meter is unreliable. Professionals measure both E and H fields there.
• Holding a meter right next to a router or phone measures the near field of that device, which says little about the room as a whole.
• Rooftop and close-in measurements near base station antennas are specialist work, for both accuracy and safety reasons.

How a spectrum analyzer shows the airwaves

A spectrum analyzer splits what it receives by frequency. The horizontal axis is frequency, the vertical axis is level, and every transmitter appears at its own place. FM stations cluster at 88–108 MHz, land mobile radio sits in the VHF and UHF bands, cellular downlinks occupy their bands, and Wi-Fi shows up at 2.4 and 5 GHz. Because each signal has its own position and shape, you can usually say what kind of service you're looking at. For the basics, see what is a spectrum analyzer.

To use one well, you need to understand its settings. They change what you see, and two people using different settings will get different pictures of the same air.

Centre frequency and span

The span is the width of the frequency range on the screen; the centre frequency is the middle of it. A wide span (say 30 MHz–2.7 GHz) gives an overview. A narrow span (say 10 MHz around one signal) shows detail. With a fixed number of display points, a wider span means each point covers more spectrum, so narrow signals merge together.

Resolution bandwidth (RBW)

The resolution bandwidth is the width of the filter the analyzer looks through at each point. It's the single most important setting.

• A narrower RBW separates signals that are close together and lowers the noise floor.
• The displayed noise floor drops by 10 dB for every tenfold reduction in RBW.
• A narrow, steady signal (narrower than the RBW) reads the same whatever the RBW.
• A wide signal (wider than the RBW) reads lower as the RBW narrows, because the filter captures only a slice of it.

Worked example. A Wi-Fi channel is roughly 20 MHz wide. Viewed with a 1 MHz RBW, each display point captures only about one twentieth of the channel's power, so the trace sits roughly 10 × log10(20) ≈ 13 dB below the channel's total power. Switch to a 100 kHz RBW and the trace drops another 10 dB. The signal didn't change; the window did. To get the total power of a wide signal, an analyzer integrates across the channel, usually with a "channel power" function.

Video bandwidth (VBW)

The video bandwidth is a smoothing filter applied after detection. Setting VBW lower than RBW averages out the jitter in noise and makes weak steady signals easier to see. It doesn't change the true noise power; it makes the trace calmer. It also slows the sweep. For bursty signals, heavy video filtering pulls the displayed level down towards the average.

Detector

Each display point usually summarises many measurements. The detector setting decides how: peak, sample, average or RMS, as described earlier. Peak is the usual default for a survey because it doesn't miss bursts. RMS is the right choice for measuring the power of digital signals.

Sweep time

A traditional swept analyzer needs time for its filter to respond at each frequency. Sweep time grows in proportion to the span and in inverse proportion to the square of the RBW, so halving the RBW makes the sweep roughly four times slower. Modern analyzers and software radios use fast Fourier transforms (FFTs) for narrow RBWs, which avoids much of this penalty, but the trade-off between detail and speed never disappears entirely.

Reference level and input attenuation

The reference level is the level at the top of the screen. Input attenuation reduces the signal before the first amplifier or mixer. Too little attenuation with a strong signal present overloads the input: the analyzer then generates false signals of its own, called intermodulation or spurious products, that aren't on the air at all. Too much attenuation raises the noise floor and hides weak signals. A good check is to add 10 dB of attenuation: real signals stay at the same displayed level, while signals generated inside an overloaded analyzer change or vanish.

Noise floor

Even with nothing connected, an analyzer shows a floor of noise. At room temperature, thermal noise is about −174 dBm per hertz of bandwidth. In a 1 MHz RBW that becomes about −114 dBm, and the analyzer's own noise figure adds to it. A receiver with a 10 dB noise figure, for example, would show a floor around −104 dBm in a 1 MHz RBW. Anything weaker is hidden.

Dynamic range

Dynamic range is the span between the strongest signal the analyzer can handle without distortion and the weakest it can see at the same time. It matters when a strong local signal sits near a weak one you care about. A nearby FM station or cell site can use up the available range, so weak signals elsewhere disappear into the floor or become polluted with spurious products.

Settings at a glance

Software-defined radios as spectrum analyzers

A software-defined radio (SDR) is a receiver that digitises a slice of the radio spectrum and hands the samples to a computer or phone. With sweep software, it steps across a wide range and builds a spectrum from many slices. The HackRF One, which Consentiva uses, covers 1 MHz–6 GHz.

What an SDR does well

• Wide coverage for little money. From 1 MHz up through the VHF and UHF radio bands, cellular, Wi-Fi and the 5 GHz band.
• Identification. Like any analyzer, it shows each signal at its own frequency with its own width.
• Recording. Because the data is digital, every sweep can be kept, time-stamped and compared later. That's the basis of a survey.

Its limits

• Uncalibrated levels. Out of the box, levels are in dBFS: relative to the digitiser's full scale. They're consistent for the same unit and settings, which is what change detection needs, but they aren't absolute power.
• Gain settings change the numbers. The HackRF has adjustable amplifier stages. Changing them shifts every reading. Keep them fixed for any comparison.
• Limited dynamic range. The HackRF digitises with 8-bit samples, so a very strong signal in the same slice can limit how weakly it can see others.
• Antenna. The supplied antenna is not calibrated and its sensitivity varies with frequency. It's fine for comparing a place with itself over time; it isn't a measurement antenna.
• Input protection. The HackRF's maximum safe input is low (the maker specifies −5 dBm). Never connect it directly to a transmitter output, and don't place its antenna right against a transmitting antenna.

What Consentiva adds to the SDR

Consentiva runs in Chrome on an Android phone, using WebUSB to talk to the HackRF directly, with no app to install (HackRF One Android setup covers the connection). It records sweeps with GPS while you walk, then learns a separate baseline for each block of roughly 150 m and each time of day: night, morning, afternoon and evening. Signals stronger than that learned baseline are flagged as unusual. That sidesteps the calibration problem for the question it's built to answer: not "how many watts are here?" but "is this different from what is normally here?" It doesn't measure exposure, and its levels are uncalibrated unless you enter a calibration offset.

From received power to field strength: why it needs a calibrated antenna

People with an SDR often ask whether they can convert its readings to V/m. In principle yes, if you know the power at the antenna socket in dBm and the antenna's characteristics. In practice that's exactly where the uncertainty lives.

The link is the antenna factor (AF), in dB per metre. For a 50-ohm system:

E (dBµV/m) = P (dBm) + 107 + AF

For an antenna with gain G (in dBi) at frequency f (in MHz), the antenna factor is approximately AF = 20 log10(f) − G − 29.79.

Worked example

Take an ideal isotropic antenna (G = 0 dBi) at 1,000 MHz. AF = 60 − 0 − 29.79 = 30.2 dB/m. If the receiver measures −40 dBm at the socket, then E = −40 + 107 + 30.2 = 97.2 dBµV/m, which is about 72,500 µV/m, or 0.073 V/m. In the far field, that corresponds to a power density of about 0.000014 W/m² (0.0014 µW/cm²).

Now look at what you had to assume: the exact gain of the antenna at that frequency and in that direction, a perfect match, no cable loss, an accurately calibrated receiver, and far-field conditions. A small whip antenna on a phone-held receiver satisfies none of these reliably. An error of a few dB in gain moves the answer by the same few dB. That's why serious field-strength work uses antennas with calibrated antenna factors, or purpose-built probes.

Calibrated field-strength meters and when you need an engineer

Professional field-strength meters combine an isotropic probe, a calibrated detector and, on better models, frequency selectivity. They come with a calibration certificate and stated uncertainty. Many show results as a percentage of a chosen exposure standard, applying the correct limit for each frequency.

The rules they're measured against

In the United States, the FCC's limits for human exposure to RF fields are in 47 CFR 1.1310. The limits depend on frequency and are expressed as time-averaged values. That time-averaging is one more reason a peak reading from a simple meter can't be compared directly with a limit. The FCC's OET Bulletin 65 describes how to evaluate compliance, including how measurements should be made. For a plain-English overview, see RF exposure limits explained.

Situations that call for a professional

• You need a figure to compare with the FCC limits, for a landlord, an employer, a regulator or a court.
• The measurement is close to transmitting antennas, on a rooftop or near equipment rooms, where both near-field effects and access safety matter.
• The result may be challenged, so the instrument's calibration, the method and the person's qualifications all need to stand up.
• Many sources overlap and you need each one's contribution, assessed properly against frequency-dependent limits.

An RF engineer will typically choose the instrument, document the calibration, follow an established method, and state the measurement uncertainty. A survey you've done yourself is still valuable in that conversation: it shows where and when to look, and which frequencies are involved.

How to take a fair comparison

Most "the reading went up" stories are really "I measured differently". If you want to compare two places, or the same place at two times, control everything except the thing you're testing.

1. Use the same instrument and settings. Same meter, same mode, same detector. For an analyzer or SDR, the same span, RBW, gain and attenuation. With Consentiva, keep the same receiver and gain settings across walks.
2. Mark the exact spot. A tape mark on the floor, a fixed point outdoors, or GPS coordinates. Readings can change over short distances because reflections add and cancel.
3. Fix the height. Pick a height, such as 1.5 m, and use it every time. A shelf, a non-metal tripod or a marked pole helps.
4. Fix the orientation. Hold a single-axis meter the same way, or take three orientations and combine them. Note which direction you faced.
5. Keep your body in the same place. Your body absorbs and reflects RF. Hold the instrument at arm's length, in the same position relative to you, or step away from a tripod while it measures.
6. Remove your own transmitters. Put your phone in airplane mode if it isn't the measuring device, and switch off nearby gadgets you control. Note anything you can't switch off.
7. Match the time. Traffic changes through the day. Compare evening with evening, weekday with weekday. Consentiva's separate baselines for night, morning, afternoon and evening exist for exactly this reason.
8. Measure long enough. Bursty signals need time. Take readings over at least a minute, or repeat the walk, rather than trusting one glance.
9. Repeat it. Take the reading again, on another day if possible. A difference that shows up every time is worth discussing; one that appears once may be chance.
10. Write it down. Date, time, location, height, orientation, instrument, settings, detector and what was nearby. Without the conditions, a number is just a number.

The full survey method, including baselines and control locations, is in how to survey the radio signals in your home or neighbourhood.

Which should you choose? A decision table by use case

Many people end up using two tools: a recording analyzer to find out what's there and when, and a professional measurement if a specific number against a limit is ever needed.

Common mistakes and how to avoid them

Troubleshooting strange readings

A short glossary

• Antenna factor: the conversion between the voltage at an antenna's output and the field strength around it, in dB per metre.
• Broadband meter: a meter that adds up energy across a wide frequency range into one number.
• dBFS: decibels relative to the full scale of a digitiser. Relative, not absolute.
• dBm: decibels relative to one milliwatt. An absolute power level.
• Duty cycle: the fraction of time a signal is transmitting.
• Dynamic range: the range between the strongest and weakest signals an instrument can measure at the same time.
• Far field: the region far enough from an antenna that the electric and magnetic fields are in the fixed 377-ohm ratio.
• Isotropic probe: a three-axis sensor whose reading doesn't depend on how it's held.
• Max hold: a display mode that keeps the highest value seen at each frequency over many sweeps.
• Near field: the region close to an antenna where the simple far-field relationships don't hold.
• Noise floor: the level of background noise below which an instrument can't see signals.
• Polarisation: the direction in which a wave's electric field points.
• RBW: resolution bandwidth, the width of the filter a spectrum analyzer looks through at each point.
• RMS: root mean square, the detector that corresponds to true average power.
• SDR: software-defined radio, a receiver whose processing is done in software.
• TDD: time-division duplex, where a base station and its users take turns on one channel.
• VBW: video bandwidth, a smoothing filter applied after detection.

Common questions

Can an RF meter tell me where a signal comes from?

Not really. A broadband RF meter adds up everything it receives across a wide frequency range into one number, so it can't tell a cell site from a Wi-Fi router or your own phone. A spectrum analyzer shows each signal at its own frequency, which usually identifies the type of transmitter.

Are cheap RF meters accurate?

Consumer broadband meters are useful for comparing places and finding hot spots, but their absolute readings can vary widely with frequency, signal type, detector and how you hold them. For a reading you can compare with the FCC limits, use a calibrated, isotropic field-strength meter.

Why do two RF meters give different readings in the same spot?

They may use different detectors (peak or average), have different frequency responses, or be single-axis meters held at different angles. For bursty signals such as Wi-Fi, 5G or radar, peak and average readings can differ by 20–30 dB while both meters work correctly.

What is the difference between V/m and µW/cm²?

V/m measures electric field strength and µW/cm² measures power density. In the far field they are linked by S = E² / 377 (S in W/m²), so 1 V/m is about 0.265 µW/cm². Close to an antenna that formula doesn't hold.

Is a HackRF a spectrum analyzer?

A HackRF is a software-defined radio. With sweep software such as Consentiva it works as a spectrum analyzer from 1 MHz to 6 GHz. It isn't calibrated out of the box, so it is best at relative measurements: what is present and how it changes.

Can I convert HackRF readings to V/m?

Only with a calibrated receiver and an antenna whose antenna factor you know at each frequency, measured in the far field. The supplied antenna isn't calibrated, so a conversion would carry an uncertainty of several dB or more.

When do I need an RF engineer instead of my own measurements?

When you need a figure to compare with the FCC limits, when measuring close to transmitting antennas or on rooftops, or when the result may be challenged. Your own survey is still useful for showing where, when and at which frequencies to look.