A radio architecture is the arrangement of filters, amplifiers, oscillators, mixers and detectors that carries a signal between the antenna and the information it holds. In a receiver, the RF front end has three jobs: select the wanted signal, raise it above the noise, and recover the modulation. This opening part of the Radio Architectures series starts with the simplest receiver that does all three, the AM detector receiver, because its weaknesses explain why every later architecture exists.
The series follows the signal inward. Parts 1 to 5 cover the receive path from antenna to data converter, and Part 6 turns around and follows the transmit path outward. Each part works one block or one system question through with formulas and numbers, and the wider RF design guide collects the supporting topics.
What an RF Front End Does
An antenna does not know which signal the user wants. It delivers everything in its bandwidth at once: the wanted channel, its neighbours, strong stations in other services, and noise. The front end has to turn that mixture into one clean signal at a level the demodulator can use. Three operations do the work, though their order and location vary from one architecture to another.
- Filtering removes energy outside the wanted channel. It sets selectivity, the ability to separate one signal from its neighbours.
- Amplification raises the signal so that later stages can process it. Where the gain is placed decides how much noise the receiver adds.
- Demodulation recovers the information from the carrier, whether the modulation is amplitude, frequency, phase or a combination of them.
Two measures summarize how well a receiver performs these jobs. Sensitivity is the weakest signal that still gives acceptable quality, usually stated as a required signal-to-noise ratio or error rate. Selectivity is how well the receiver rejects signals at other frequencies. Part 4 treats sensitivity in numbers, and Part 3 examines what happens when strong unwanted signals get past the filters.
Radio System Design Starts with Requirements
Radio system design begins with a short list of constraints, and these have changed far less over the decades than the hardware that meets them. The designer needs to know the carrier frequency range and channel bandwidth, the modulation and the signal quality it requires, the link budget available between transmitter and receiver, and the limits on power, cost and size. A link budget sets the weakest signal the receiver must handle, and the modulation sets the signal-to-noise ratio that signal must reach.
These constraints then drive the choice of RF architecture. A battery-free crystal set and a 5G handset both filter, amplify and demodulate, yet they share almost no hardware, because one receives a single strong analog carrier and the other receives wideband data at very low levels across many bands. Better semiconductor processes and cheap digital signal processing have moved much of the work out of analog circuits, but they have not removed the underlying requirements.
| Architecture | How the channel is selected | Main limitation | Covered in |
|---|---|---|---|
| AM detector | One tuned circuit at RF | Poor selectivity and sensitivity | This part |
| Tuned radio frequency | Several tuned RF stages | Stages must track; bandwidth varies with tuning | Part 2 |
| Superheterodyne | Fixed filter at an intermediate frequency | Image response, more parts | Part 2 |
| Direct conversion | Low-pass filters at baseband | DC offset, I/Q errors | Parts 2 and 12 |
| Direct RF sampling | Digital filters after a fast ADC | Clock jitter, converter power | Parts 5 and 13 |
The AM Detector Receiver
The AM detector receiver, of which the crystal radio is the classic form, has only the blocks shown in Figure 1. The antenna feeds a tuned circuit that selects one station. A diode rectifies the result, a small RC network smooths away the RF and leaves the audio envelope, and an optional amplifier drives a speaker. No oscillator is present and no frequency translation takes place, so everything happens at the received frequency.
At medium-wave frequencies a practical wire antenna is much shorter than a quarter wavelength and looks capacitive, so a series inductance is used to bring it to resonance and increase the current it delivers. Early sets used a metal point pressed against a galena crystal as the rectifier. Today the same job goes to a semiconductor diode, with Schottky types preferred for their low forward voltage and small junction capacitance.
How the diode envelope detector works
On each positive peak of the carrier the diode conducts and charges the capacitor. Between peaks the diode is off and the capacitor discharges slowly through the load resistor, so the output follows the peaks. The time constant has two limits. It must be long compared with the carrier period so that the RF ripple is small, and short enough that the output can fall as fast as the envelope falls. For sinusoidal modulation at frequency fm and modulation index m, the usual limit to avoid diagonal clipping is:
RC ≤ √(1 − m²) / (2π fm m)
For a 5 kHz audio tone at m = 0.8, RC must be no more than 23.9 µs. A 1000 kHz carrier has a period of 1.0 µs, so a choice of about 12 µs sits comfortably between the limits, as the simulated detector in Figure 2 shows. The output tracks the envelope with only a small ripple.
Worked Example: AM Power and Bandwidth
Amplitude modulation places a copy of the audio on each side of the carrier. A station at 1000 kHz sending a 5 kHz tone produces lines at 995, 1000 and 1005 kHz, so a signal with 5 kHz of audio bandwidth occupies 10 kHz of spectrum. With a carrier power Pc, each sideband carries Pcm²/4 and the total is Pc(1 + m²/2).
| Modulation index m | Carrier share of total power | Each sideband share of total power |
|---|---|---|
| 1.0 | 66.7% | 16.7% |
| 0.3 | 95.7% | 2.2% |
Even at full modulation, only 16.7% of the transmitted power sits in each sideband. The two sidebands carry the same information, so a receiver that could use one sideband alone would still need just that 16.7%, which is 7.8 dB below the full transmitted power. Most programme material runs well below full modulation, and at m = 0.3 the carrier takes 95.7% of the power. The carrier does serve a purpose, because it is what allows a simple diode to work without any reference oscillator in the receiver. That simplicity at the receiving end is paid for in transmitter power and in occupied spectrum.
The basics of amplitude modulation are covered separately. Other detector circuits replace the diode to recover frequency or phase modulation, and phase-shift keying and QAM dominate modern data links, but those signals need more receiver hardware than a single rectifier.
Selectivity and Q in a Tuned Circuit
The tuned circuit is the only filter in a detector receiver, so its Q factor sets the selectivity. To tune 1000 kHz with a 250 pF capacitor, the inductance must be L = 1/((2πf)²C) = 101.3 µH. The 3 dB bandwidth of the loaded circuit is f0/Q, and the response at an offset Δf is approximately:
Attenuation (dB) = 10 log10(1 + (2QΔf / f0)²)
AM broadcast stations in North America sit on a 10 kHz raster. The table applies the formula to the adjacent channel, and Figure 3 plots both curves.
| Loaded Q | 3 dB bandwidth | Rejection at 10 kHz offset |
|---|---|---|
| 50 | 20.0 kHz | 3.0 dB |
| 150 | 6.7 kHz | 10.0 dB |
These numbers expose the problem. A Q of 50 is respectable for a loaded circuit, and it rejects the neighbouring station by only 3.0 dB. Raising Q to 150 reaches 10.0 dB, but the 3 dB bandwidth of 6.7 kHz now cuts into the 10 kHz signal and dulls the audio. Reaching 30 dB of rejection at 10 kHz offset would need a Q near 1580, far beyond what a tunable LC circuit can provide, and the audio would be lost anyway. A single resonator cannot give both a flat passband and steep skirts. Getting that shape requires several coupled resonators, and making several resonators tune together is the problem that shapes Part 2.
Sensitivity of a Detector Receiver
The second limit is signal level. The thermal noise in a 10 kHz bandwidth at 290 K is −134.0 dBm, as explained in the page on thermal noise and kTB. If noise were the only limit, a receiver could work near that level. A diode with no gain ahead of it cannot get close, because a rectifier needs a usable voltage to overcome its own forward characteristic. At low drive the diode falls into a square-law region where the output is proportional to the square of the input, so the audio fades much faster than the RF does. As a rough guide, an unbiased diode detector needs a signal on the order of −50 dBm, which is about 84 dB above the thermal noise in the channel.
That gap of more than 80 dB represents performance a detector receiver simply leaves unused. A crystal set works only because broadcast transmitters are powerful and the listener is often close. Audio gain after the diode does not help, since the damage has already been done at the detector. The gain has to come before the detector, at RF. Adding it brings noise, distortion and stability into the design, and those are the topics of the next three parts. Readers who want to refresh the units used here can see dB and dBm explained.
Why Radio Architecture Moved On
The detector receiver shows the two pressures behind every later design. Selectivity improves when filtering takes place at one fixed, convenient frequency, and sensitivity improves when low-noise gain sits ahead of everything else. Moving the signal to a fixed frequency needs an oscillator and a mixer, and the oscillator and mixer add problems of their own. Each architecture in this series can be read as a different answer to the same question: where should the signal be translated, filtered and digitized? The receiver architecture comparison sets the modern answers side by side.
Today the phrase radio front end usually means everything between the antenna and the data converter: band filters, switches, the low-noise amplifier, mixers and baseband or IF amplifiers. In a phone, much of that hardware sits inside integrated front-end modules, but each block still performs one of the three jobs listed at the start of this page.
Next in the Series
Part 2, Heterodyne and Homodyne Receivers, LOs and Mixers, adds RF gain and frequency translation: the tuned radio frequency receiver, direct conversion, the superheterodyne, and the local oscillators and mixers that make translation possible.