What actually is RF?
Why can the same 30 mm connection be harmless in one circuit and important in another?
RF is not a different kind of electricity. The same voltage, current, electric fields, and magnetic fields are involved. What changes is the scale of the problem.
An electrical change takes time to travel. When a circuit changes slowly, the signal may change negligibly during that travel time, so we safely ignore it. As the changes become faster, the size and shape of traces, cables, component leads, connectors, and even enclosure openings can begin to affect what the circuit does.
That is the useful first idea of RF: time and physical distance can no longer be separated. There is no sharp frequency where ordinary electronics suddenly becomes RF. It depends on the signal, the structure, and the accuracy the design needs.
A short PCB trace may look like a perfect connection when a sensor changes a few times per second. A 2.45 GHz signal changes billions of times per second, so it can change appreciably while an electrical effect is still travelling through an equally ordinary-looking trace, package, or connector. We will calculate that scale shortly.
Think about itCould a signal that never leaves a coax cable still be RF?
Yes. If propagation time and physical dimensions matter to how that guided signal behaves, RF thinking is useful even when the signal is not intended to radiate.
A signal inside a shielded coax cable or along a PCB trace can still be RF. Radiation is one possible behavior of an RF system, not the definition of RF itself.
A signal that repeats
When we say that a 2.45 GHz signal repeats, what exactly is repeating?
Start with a voltage at one point on a cable. It rises, crosses its middle value, falls, and eventually returns to the same value while moving in the same direction. That full repeat is one cycle.
A sinusoid is the simplest single-frequency repeating pattern. On an oscilloscope, the vertical axis might be voltage and the horizontal axis is time. In a spatial snapshot, the vertical axis might represent one component of electric field while the horizontal axis is position. In both cases, the curve is a graph of a measured quantity.
Think about itIn the spatial view, do the field sensors travel with the wave?
No. Each sensor stays in one place. Its reading changes as the field pattern reaches it. What propagates is the pattern of field change, not the sensor and not the plotted line.
One field. Two views.
This model slows the field change down so you can watch the same wave in two useful ways: at one fixed location over time, and across space at one instant.
Watch the electric field change over time
The arrow shows the electric field at this location. Its direction flips between up and down, and its length shows the field strength.
Compare the electric field along the wave
Think of each arrow as a sensor fixed at one point in space. The sensors do not move. As the wave travels, each sensor measures the electric field changing in direction and strength, one after another along the row.
Arrow bases remain fixed. The changing pattern of arrow direction and length is what propagates.
At this setting, the electric field at a fixed location completes 1 cycle each second. Watch how the fixed-point reading and the pattern across space change together.
It is not. The wave can propagate along a straight cable or through space while the plotted voltage or field value rises and falls. The curve shows value versus time or position; it does not show a particle or a packet of energy snaking along that route.
Go deeperA compact mathematical description
After the picture is clear, a sinusoid can be written as. The symbols name ideas we will build one at a time: amplitude A, frequency f, time t, and phase φ, with the angle written in radians.
Frequency & period
How can we describe the repetition without saying “very fast”?
Frequency, f, counts how many complete cycles occur each second. The unit is hertz (Hz): 1 Hz means one cycle per second. A 100 MHz signal repeats 100 million times per second. A 2.45 GHz signal repeats 2.45 billion times per second.
Period, T, asks the inverse question: how long does one cycle take? Frequency is cycles per second; period is seconds per cycle.
Think about itIf frequency doubles, what happens to the period?
It halves. Twice as many cycles must fit into the same second, so each cycle gets half as much time. Try 1, 2, and 4 Hz below and watch the marked period shrink.
One repeat takes time.
This scale is always one second long. At 1 Hz, that second contains one cycle.
One periodT = 1 s
The blue dimension line marks one complete cycle, from one crest to the next.
At 100 MHz, one cycle takes 10 ns. At 2.45 GHz, one cycle takes about 0.408 ns, or 408 ps. Nothing new has been added to the signal: frequency and period are two ways to describe the same repetition.
Imagine a cable in which a signal travels at about 2 × 108 m/s. One metre adds about 5 ns of delay. That is half of a 100 MHz cycle, but more than twelve complete 2.45 GHz cycles. An ideal cable may simply delay a sine wave without changing its shape, but the delay is no longer insignificant.
Wavelength
Period is one cycle viewed in time. What does that same cycle look like across distance?
Freeze a travelling sinusoid for an instant. The distance from one point in the cycle to the next identical point is its wavelength, written λ. Crest to crest is convenient, but any matching pair of points works.
Think about itIf frequency doubles while propagation velocity stays constant, what happens to wavelength?
It halves. The wave completes twice as many cycles while travelling the same distance in one second, so each spatial cycle occupies half the distance. Use the model below to compare frequency first, then change the propagation medium.
One repeat takes up distance.
This scale is always 3 metres long. At 100 MHz in free space, those 3 metres contain 1.00 wavelength.
One wavelengthλ = 3.00 mPhase velocity: c
The blue line marks one wavelength. The cable uses an illustrative velocity factor of 0.66; the PCB line uses an illustrative effective permittivity of 3. Actual guided values depend on cable construction, PCB material and stack-up, trace geometry, and frequency.
In free space, use vp = c, where c = 299,792,458 m/s. A 100 MHz wave is therefore about 3 m long. At 2.45 GHz, it is only about 122 mm long.
The source normally keeps the frequency unchanged when the wave enters a cable or PCB structure. What changes is the propagation velocity. In ordinary dielectric-guided structures the wave is slower, so its wavelength becomes shorter. The exact PCB value depends on material, geometry, and field distribution; the board's bulk dielectric constant alone is not enough for every structure.
Frequency says how often the signal repeats, not how fast it propagates. In vacuum, electromagnetic waves of every frequency travel at c. In a given simple medium, higher frequency usually means a shorter wavelength, not a higher speed.
c/f is the free-space wavelength. In a 0.66 velocity-factor cable, the 2.45 GHz wavelength is about 80.8 mm. A PCB line has its own effective velocity, so the same frequency can have a different guided wavelength again.
Why wavelength matters in RF
When does a trace stop behaving like an ideal connection?
Electrical length compares a structure's physical length with the wavelength travelling through it. It answers a practical question: how much of one cycle fits along this structure?
Think about itWhich is electrically longer: the same 30 mm trace at 10 MHz or at 2.45 GHz?
The 2.45 GHz trace. Its physical length is unchanged, but the wavelength is far shorter, so 30 mm occupies a much larger fraction of a cycle. In a typical guided PCB structure, the wavelength is shorter still.
Compare its physical length with the signal's wavelength in that structure. A 30 mm feature is tiny beside the roughly 30 m free-space wavelength at 10 MHz. At 2.45 GHz, 30 mm is almost one quarter of a free-space wavelength, and it can be an even larger fraction of the shorter guided wavelength on a cable or PCB.
The free-space wavelength is about 122.4 mm. One quarter is about 30.6 mm. That does not make every 30.6 mm object a perfect antenna, but it tells an RF engineer to stop treating that size as automatically negligible.
Electrical length can be written as a fraction of wavelength, or as the phase accumulated while the signal travels through the structure.
The lumped approximation treats a component or connection as though its voltage and current are effectively uniform at one instant. That approximation is useful when propagation delay is negligible for the question being asked. When it is not, different positions can have different voltage, current, and phase. The structure has distributed behavior.
A distributed cable or trace delays the signal. If its impedance changes abruptly, part of the wave can return toward the source; that returned wave is a reflection. Later lessons will develop how a line can also change the impedance seen at its input.
Antennas deliberately use wavelength-scale structures to exchange energy between a guided signal and radiated fields. Interconnects can show wavelength-dependent behavior even when radiation was never the designer's goal.
Once a connection is electrically significant, layout becomes part of the circuit. Trace geometry, reference planes, connector transitions, impedance control, termination, and the position of the measurement reference plane can all change the result.
Go deeperThere is no universal length cutoff
Rules such as “shorter than one tenth of a wavelength is lumped” are warning-scale heuristics, not laws. Acceptable error, impedance discontinuities, loss, bandwidth, and measurement goals matter. Fast digital edges also contain frequency components far above the clock rate, so rise time can make a low-clock-rate interconnect an RF problem.
Frequency and wavelength tell us when geometry matters. Next we need a different question: how large is the signal we are sending through that geometry?
Amplitude
Two signals can repeat at the same rate but have very different magnitudes. How do we say exactly how large one is?
Think about itWhich has more power: a 1 V signal or a 0.5 V signal?
There is not enough information. Ask: 1 V peak, peak-to-peak, or RMS? Across what impedance? What waveform? Only then can the powers be compared.
Amplitude is the magnitude of a stated quantity relative to a stated reference. We will begin with voltage relative to 0 V, but RF engineers also measure current, electric field, magnetic field, and power. “Amplitude” is incomplete until the quantity and unit are clear.
Same timing. Different height.
This view always shows a zero-offset 1 kHz sine wave. Changing the peak voltage changes how far the signal rises above and below 0 V; it does not change the timing.
- Peak voltage · Vpk
- 1 V
- Peak-to-peak · Vpp
- 2 V
- RMS voltage · Vrms
- 0.707 V
The blue line measures peak voltage from the 0 V reference to the crest. Waveform limit: these readouts assume a zero-offset sine. Other waveform shapes have different RMS-to-peak relationships, and a DC offset changes the total RMS value.
For a sine wave centered on 0 V, peak voltage, Vpk, runs from 0 V to one crest. Peak-to-peak voltage, Vpp, runs from the negative peak to the positive peak. RMS voltage, Vrms, is the equivalent DC voltage that would produce the same average power in a resistor.
Consider a resistive 50 Ω load. A 1 Vpp sine wave has only 0.3536 Vrms and delivers 2.5 mW. A 1 Vrms sine wave into that same load delivers 20 mW. Both may be casually called “one volt,” yet their powers differ by a factor of eight.
A voltage amplitude alone is not power. For a resistive load,, so waveform, impedance, and where the voltage is measured all matter.
It might mean peak, peak-to-peak, RMS, or even a source setting specified for a particular load. A trustworthy RF measurement states the qualifier, waveform, impedance, and reference plane: for example, “1 Vpp sine measured across 50 Ω.”
Amplitude tells us how large a signal is. To understand what happens when signals meet, we also need to know where they are within their cycles.
Phase
If two sine waves have the same frequency and amplitude, can they still behave differently?
Yes. They can be at different positions within their cycles. Phase describes that relative position as an angle: one complete cycle is 360°. To make a phase statement, choose a reference signal and a reference plane.
Think about itWhat happens when two equal, coherent signals are 180 degrees apart?
They cancel ideally at that observation point. Set the phase explorer below to 180°: the two input traces oppose each other and the resultant becomes zero.
What happens when two equal signals meet?
Change their phase offset. The input amplitudes stay equal, while the sum trace shows their voltages adding point by point. Then change frequency to see why the same phase angle does not always mean the same time shift.
The addition is partly constructive. Here, A is the peak amplitude of either input signal. At 180°, lead and lag are equivalent modulo one cycle. A fixed phase relationship requires equal-frequency, coherent signals.
- At 0°, equal signals rise and fall together. Their sum has amplitude 2A.
- At 90°, one signal is one quarter-cycle ahead or behind. Their sum has amplitude √2A.
- At 180°, one signal is at its positive peak when the other is at its negative peak. Ideal equal signals cancel.
Phase and time delay are connected. A delay occupies some fraction of a cycle, and that fraction becomes a phase lag. The same physical delay occupies more cycles as frequency rises.
A fixed 100 ps delay is only 3.6° at 100 MHz, 36° at 1 GHz, and about 88.2° at 2.45 GHz. The delay did not change. Its size relative to the cycle did.
Cable length, PCB routing, filters, antenna spacing, and measurement fixtures all create phase differences. Those differences can produce reinforcement, cancellation, beam steering, or a misleading measurement, depending on where and how signals combine.
Phase is meaningful relative to another coherent signal, a defined time origin, and a stated reference plane. Move the reference plane along a cable and the measured phase changes.
It does not. A 90° lag is 2.5 ns at 100 MHz, 250 ps at 1 GHz, and about 102 ps at 2.45 GHz. Always state the frequency with a phase-to-delay conversion.
Go deeperThe sign of delay
With the convention sin(2πft + φ), a delayed signal x(t - τ) has phase change, modulo 360°. The minus sign means lag. A phase display wrapped to one turn cannot reveal how many whole cycles of delay occurred, so wideband or multi-frequency measurements are needed to remove that ambiguity.
Why do we keep using sine waves?
Bluetooth and Wi-Fi carry data. Their real waveforms are not one endless, perfect sine wave, so why do RF engineers keep drawing sinusoids?
Because a complicated signal can be represented by sinusoidal components with different frequencies, amplitudes, and phases. Add the right components together and they reproduce the original waveform. This is the intuition behind Fourier analysis; we do not need the transform mathematics yet to use the idea.
Think about itWhat happens when we add smaller sinusoids at three and five times a starting frequency?
Their sum develops flatter tops and steeper transitions. It begins to resemble a square wave, even though every ingredient is a smooth sinusoid. Build the approximation below one component at a time.
Can smooth sine waves build a square-like shape?
Start with a 1 Hz fundamental. Then add smaller odd harmonics. The light traces show the individual sinusoids; the blue trace is their exact point-by-point sum.
- Fundamental · 1 Hz · amplitude 1
- Sum
The 1 Hz fundamental sets the repetition rate. By itself, it remains smoothly rounded everywhere.
This is a truncated approximation of a zero-mean, symmetric square wave with a 50% duty cycle—not a perfect square wave. That symmetry is why this example uses only odd harmonics; other duty cycles can introduce even harmonics, and a DC offset adds a zero-frequency term. The usual overall 4 / π scale factor is omitted, so the fundamental has normalized amplitude 1. Compare the shape and relative amplitudes, not an absolute voltage.
| Time-domain signal | Frequency-domain view |
|---|---|
| An infinite-duration ideal sine wave | One ideal frequency component |
| A repeating non-sinusoidal waveform | Components at integer multiples of its repetition frequency, plus DC if its average is not zero |
| A modulated RF packet | A band of nearby components around its operating frequency |
| A short pulse or fast edge | A broad range of frequency components |
Sine waves are especially useful because many RF networks behave predictably for small signals: feed in a steady sine and the output remains at the same frequency, with a changed amplitude and phase. Engineers can use a sine as a clean probe, analyze how a filter, cable, amplifier, or antenna treats each component, then combine the results to understand the full signal.
A spectrum analyzer uses this second view. Instead of drawing voltage versus time, it shows how signal energy is distributed across frequency. That bridge leads directly to later lessons on signals, bandwidth, and modulation.
Only an ideal sine wave that exists forever produces a perfect spectral line. Real packets start and stop. Modulation intentionally occupies bandwidth, while oscillator phase noise, nonlinear products, and spurious signals add further components.
Go deeperA compact Fourier idea
A periodic waveform can be represented schematically as, where for repetition frequency . Some component amplitudes may be zero. This is a representation of one waveform, not a claim that tiny physical sine waves are independently travelling inside it.
The formal network property used above is called linear and time-invariant: scaled inputs produce scaled outputs, sums can be analyzed component by component, and the behavior does not change merely because the experiment starts later. Real RF hardware only approximates this model over a stated signal range and bandwidth.
RF spectrum
Where do kHz, MHz, and GHz sit, and where do familiar radio systems fit?
The electromagnetic spectrum is the full family that includes radio, infrared, visible light, X-rays, and more. Within its radio-frequency region, engineers use decimal prefixes to keep large numbers readable:
- 1 kHz = 1,000 Hz
- 1 MHz = 1,000,000 Hz
- 1 GHz = 1,000,000,000 Hz
Think about itWhich is higher in frequency: 100 MHz FM or the 1.575 GHz GPS L1 signal?
GPS L1. Convert to the same unit first: 1.575 GHz is 1,575 MHz, about 15.75 times 100 MHz. Prefixes make the numbers readable, but comparisons still require matching units.
| Technology | Typical scale | Unit | What the scale suggests |
|---|---|---|---|
| AM broadcast | roughly 500-1,700 kHz | kHz | Compact receiving antennas are often electrically short |
| FM broadcast | around 100 MHz | MHz | About 3 m free-space wavelength at 100 MHz |
| Cellular | hundreds of MHz to several GHz | MHz / GHz | Many bands; the exact range depends on system and region |
| GPS L1 | 1.57542 GHz | GHz | A precise satellite-navigation carrier frequency |
| Bluetooth and 2.4 GHz Wi-Fi | around 2.4 GHz | GHz | The recurring scale used in this lesson |
| 5 and 6 GHz Wi-Fi | several GHz | GHz | The same path delay creates a larger carrier-phase shift |
These are orientation examples, not an allocation chart. Actual bands and channel rules depend on the technology, standard, and jurisdiction. When compliance or measurement matters, state the exact frequency range rather than relying only on a familiar label.
Microwave is a name commonly used for the higher-frequency part of the broader RF domain; the exact lower boundary varies by convention. At 2.45 GHz, a signal is in UHF under the usual 300 MHz-3 GHz band nomenclature and is also commonly described as microwave.
Go deeperStandard RF band designators
These decade-wide names help engineers communicate scale. Their wavelengths below are approximate free-space values. They become shorter in ordinary dielectric transmission lines such as the cable and PCB examples used here.
| Band | Frequency | Approx. free-space wavelength |
|---|---|---|
| VLF | 3-30 kHz | 100-10 km |
| LF | 30-300 kHz | 10-1 km |
| MF | 300 kHz-3 MHz | 1 km-100 m |
| HF | 3-30 MHz | 100-10 m |
| VHF | 30-300 MHz | 10-1 m |
| UHF | 300 MHz-3 GHz | 1 m-10 cm |
| SHF | 3-30 GHz | 10-1 cm |
| EHF | 30-300 GHz | 10-1 mm |
Putting it all together — a 2.45 GHz signal
What does one frequency number tell an engineer about time, distance, layout, and measurement?
Follow one 2.45 GHz sinusoidal component from a transceiver, through a PCB feed and cable, toward an antenna. The source establishes 2.45 billion cycles per second. From that one starting point, the earlier ideas connect.
| Concept | At 2.45 GHz | Engineering meaning |
|---|---|---|
| Frequency | 2.45 GHz | 2.45 billion cycles each second |
| Period | 408 ps | Only 0.408 ns for one complete cycle |
| Free-space wavelength | 122.4 mm | One cycle viewed across distance |
| Free-space quarter-wave | 30.6 mm | A useful first physical scale, not a universal antenna length |
| 30 mm in free space | 100 ps / 88.3° | Almost one quarter of a cycle |
| 30 mm in 0.66-VF cable | 152 ps / 133.7° | The slower wave makes the same length electrically longer |
| Spectrum location | UHF; commonly microwave | In the 2.4 GHz region used by Bluetooth and Wi-Fi |
The same 30 mm physical path is almost 90° long in free space, about 134° in the illustrative cable, and about 153° on an illustrative PCB line with effective permittivity εeff = 3, a material-and-geometry value that sets the line's phase velocity. The frequency is unchanged. Velocity, wavelength, delay, and electrical length change with the medium and geometry.
Frequency still tells us nothing about amplitude or power. The designer must separately state voltage or power, its qualifier, the impedance, and the reference plane. And a real Bluetooth or Wi-Fi transmission is not an infinitely thin line at 2.45 GHz: modulation distributes its energy across a finite band of frequency components.
Frequency sets the cycle rate. Period gives the time per cycle. Propagation velocity turns that time into wavelength. Wavelength makes physical dimensions electrically meaningful. Delay appears as phase, phase controls how coherent signals add, and Fourier components explain how a real modulated signal occupies spectrum.
Think about itIf the frequency doubles to 4.90 GHz while the structure and velocity stay unchanged, what changes?
Period and wavelength halve. The physical delay stays the same, but it spans twice as many cycles, so the electrical length and total phase lag double. A layout that was manageable at 2.45 GHz may be much less forgiving at 4.90 GHz.
Check your understanding
Answer each question in your own words, then reveal the model answer.
01A 2.45 GHz sinusoid travels in free space. What are its approximate period, wavelength, and phase change across 30 mm?
Model answerAbout 408 ps, 122.4 mm, and 88.3°. The 30 mm path is already close to one quarter of a cycle.
02The same 2.45 GHz signal enters a cable with velocity factor 0.66. What stays fixed, and what changes?
Model answerFrequency and period stay fixed. Phase velocity and wavelength fall; the wavelength becomes about 80.8 mm, so the same 30 mm path grows to about 133.7° electrically.
03Why does a real 2.45 GHz data packet occupy a band rather than one infinitely narrow spectral line?
Model answerIt starts, stops, and is modulated. Those time-domain changes require a range of sinusoidal frequency components around the operating frequency.