Everything between the antenna and the converter costs either sensitivity or power, and usually both. The receive chain sets how weak a signal the site can hear; the transmit chain sets what the site costs to run. Neither is improved by the digital design downstream, which is why front-end choices are made first and lived with longest.
A receiver's sensitivity is set by its noise figure, and the cascade arithmetic is unforgiving: the first amplifying stage dominates, because every later stage's noise contribution is divided by the gain ahead of it. Half a decibel lost in front of the low noise amplifier is half a decibel of sensitivity gone, and no amount of processing recovers it.
That is why the LNA sits as close to the antenna as physically possible, and why every component before it — connector, filter, switch, cable — is chosen for insertion loss above all else.
It also explains the tower-mounted amplifier. Put the LNA at the top of the mast and the feeder loss falls after the gain rather than before it, where it costs almost nothing. Put it at the bottom and the same cable is subtracted directly from sensitivity.
| Approach | How it separates | What it costs |
|---|---|---|
| FDD with a duplexer | Transmit and receive occupy different frequencies; a filter pair isolates them | Insertion loss in both paths, physical size, and isolation that must exceed 90 dB or the transmitter deafens its own receiver |
| TDD with a switch | Transmit and receive share the frequency and alternate in time | Lower loss and smaller, but every site in the area must agree on the switching boundary or one network desensitizes another |
| Circulator | A ferrite device routing energy by direction | Non-reciprocal and lossy; used where a duplexer cannot provide the bandwidth |
The duplexer is often the largest and heaviest component in a radio unit, and its isolation requirement is why a small filter cannot simply be substituted.
| What it gives | What it costs | |
|---|---|---|
| LDMOS | Mature, inexpensive, excellent below about 3 GHz | Efficiency and gain fall away at higher frequency; larger die for the same power |
| GaN | Higher power density, higher efficiency, works well above 3 GHz, tolerates heat | More expensive per device, and the thermal design becomes the limiting factor rather than the transistor |
| Class AB | Simple, linear, predictable | Efficiency collapses at backoff, and a modulated signal spends most of its time backed off |
| Doherty | Two amplifiers, one always on and one engaging at peaks; efficiency holds at backoff | Narrower bandwidth, harder to tune, and it depends on digital pre-distortion to meet the spectral mask |
| Envelope tracking | Supply voltage follows the signal envelope, so the device is never over-supplied | A fast, efficient modulator is itself difficult; more common in handsets than base stations |
Modulated signals have a high peak-to-average ratio, so the amplifier operates backed off from saturation almost all the time. Doherty exists because that is where efficiency would otherwise be worst.
Everything between the antenna and the data converters: duplexer or TDD switch, filters, low noise amplifier on receive, power amplifier and driver on transmit, and the matching and protection around them. It sets both the sensitivity and the energy cost of the site.
Because of noise figure cascade. The first amplifying stage dominates the total, since every later stage's noise is divided by the gain ahead of it. Any loss before the LNA — connector, filter, cable — subtracts directly from sensitivity and cannot be recovered downstream.
A filter pair that lets transmit and receive share one antenna at different frequencies. It has to provide isolation typically exceeding 90 dB, because a transmitter of tens of watts sits beside a receiver listening at a fraction of a picowatt. That isolation requires physical volume, which is why it is often the largest component in a radio unit.
Two amplifiers working together: a main device always conducting and a peaking device that engages only on signal peaks. It holds efficiency at backoff, which matters because a modulated signal with a high peak-to-average ratio spends most of its time backed off from saturation. The cost is narrower bandwidth and a dependence on digital pre-distortion.
LDMOS is mature and inexpensive and performs well below about 3 GHz. GaN offers higher power density and efficiency, works well above 3 GHz and tolerates higher junction temperature, at higher device cost — and it shifts the limiting factor to the thermal design rather than the transistor.
Because the amplifier dominates base station power consumption, and that is an operating cost across the network's entire life. A few points of efficiency multiplied by thousands of sites and many years is the return that justifies crest factor reduction, pre-distortion and Doherty architecture in the first place.
Placing the LNA at the antenna rather than at the base of the mast. The feeder loss then occurs after the gain instead of before it, where it costs almost nothing, rather than being subtracted directly from sensitivity.
Written for engineers. Share it with one.
What makes the amplifier legal.
KNOWLEDGEDPD, CFR and the amplifier.
KNOWLEDGEHow isolation fails.
PRODUCTThe unit this sits in.
TECHNOLOGYLink budget and antenna.
PRODUCTWhere thermal limits output.
Tell us the specification, the constraint and the deadline. Programs that cross silicon, radio, embedded and AI are where Faststream is strongest.