If you work in RF, you’ve been burned by the Friis formula at least once. But once the idea that “the first stage determines everything” clicks, a lot of things suddenly make sense. When I first started doing board-level RF, it took me a long time to truly internalize one thing: how does a 0.5 dB degradation in LNA noise figure translate to nearly 1 dB worse overall receiver sensitivity?
Here’s what I eventually figured out:
- The total receiver NF is more than 50% determined by the first-stage LNA — in a typical chain it’s 50–70%, and in high-gain configurations it can exceed 80%
- Higher LNA gain reduces the noise contribution from downstream stages, but don’t think you can just stack gain indefinitely
- The NF and gain of the first stage are the two most critical metrics in the entire chain — period.
- In real-world design, the NF vs. gain trade-off is a problem every RF engineer has to face
1. Starting with the Friis Formula
In 1944, Harold T. Friis published the formula at Bell Labs that became an RF engineer’s bible:
Ftotal = F1 + (F2-1)/G1 + (F3-1)/(G1·G2) + …
Consider a typical receiver chain:
| Stage | Module | NF (dB) | Gain (dB) |
|---|---|---|---|
| Stage 1 | LNA | 0.8 | 18 |
| Stage 2 | Filter + Mixer | 8 | -2 (loss) |
| Stage 3 | IF Amplifier | 6 | 25 |
| Stage 4 | Baseband / ADC | 15 | — |
Plugging into the Friis formula — and remember, you must convert to linear domain first:
| Parameter | dB Value | Linear (×) |
|---|---|---|
| F₁ (LNA) | 0.8 dB | 1.202 |
| G₁ (LNA Gain) | 18 dB | 63.1 |
| F₂ (Filter + Mixer) | 8 dB | 6.31 |
| G₂ (Mixer Loss) | -2 dB | 0.631 |
| F₃ (IFA) | 6 dB | 3.98 |
| G₃ (IFA Gain) | 25 dB | 316.2 |
| F₄ (Baseband / ADC) | 15 dB | 31.62 |
Ftotal(linear) = 1.202 + (6.31-1)/63.1 + (3.98-1)/(63.1×0.631) + (31.62-1)/(63.1×0.631×316.2)
= 1.202 + 0.084 + 0.075 + 0.002 = 1.363
→ Ftotal = 10·log(1.363) ≈ 1.35 dB
Look at that — total NF is only 1.35 dB, and the LNA alone accounts for 0.8 dB, nearly 60% of the total.
What’s even more striking is how fast the downstream contributions decay. The mixer has an NF of 8 dB, but because there’s an LNA with 18 dB gain in front of it (63× linear), its contribution to total NF is crushed below 0.2 dB. By the time you reach the fourth-stage baseband, it’s practically irrelevant.
Why the LNA must sit directly after the antenna: boost the gain before the signal is attenuated or contaminated by additional noise — after that, downstream noise becomes “invisible.”
2. What Happens When the First Stage Degrades?
This is the most interesting part. Let’s run a comparison:
Scenario A (Normal): LNA NF = 0.8 dB, Gain = 18 dB → Total NF ≈ 1.35 dB
Scenario B (LNA NF degraded by 0.5 dB): LNA NF = 1.3 dB, Gain = 18 dB → Total NF ≈ 3.33 dB
Scenario C (LNA gain dropped by 5 dB): LNA NF = 0.8 dB, Gain = 13 dB → Total NF ≈ 2.34 dB
Let these numbers sink in:
- LNA NF degrades by only 0.5 dB → total NF degrades by ~2.0 dB. That’s a 4× amplification, far worse than 1:1.
- LNA gain drops by 5 dB → total NF rises by nearly 1 dB. Why? The “shielding effect” against downstream noise is significantly weakened — the mixer and IFA noise contributions resurface.
And if you eliminate the LNA entirely, feeding the signal straight into the mixer:
Scenario D (No LNA, direct to mixer):
Ftotal = 6.31 + (3.98-1)/0.631 + (31.62-1)/(0.631×316.2) = 6.31 + 4.72 + 0.15 = 11.18 (linear)
→ Total NF ≈ 10.5 dB
Sensitivity goes out the window. This is why you’ll essentially never see a receiver without an LNA — unless your application genuinely doesn’t care about sensitivity (certain high-power near-field communications excepted).
Engineering rule of thumb:
For every 0.1 dB degradation in LNA NF, system total NF degrades ~0.08–0.1 dB.
For every 1 dB degradation in mixer NF, system total NF degrades only ~0.02–0.05 dB.
The return on investment is not even in the same order of magnitude.
3. So Just Crank LNA Gain to Infinity?
Answer: No. Three reasons why.
Reason 1: Linearity
Higher gain means larger signal amplitude at the LNA output. As soon as the desired signal or an interferer gets strong enough, the LNA enters compression or saturation before any other stage in the chain. Once in compression, nonlinear distortion products (IMD3, IMD2) get amplified by every subsequent stage. This is far worse than noise.
Concrete numbers: for a typical 2.4 GHz Wi-Fi LNA, going from 15 dB to 22 dB gain typically drops OIP3 by 3–6 dBm. Blocker tolerance takes a direct hit. In urban environments saturated with Wi-Fi and Bluetooth signals, one strong blocker means dropped packets and terrible user experience.
Reason 2: Stability
High gain means stronger positive feedback risk. Parasitic coupling between LNA input and output — bond wires, package pins, PCB traces — is far more likely to cause oscillation at high gain. This is especially problematic for wideband LNAs covering multiple octaves, where the high-frequency K-factor can drop below 1.
I’ve debugged this exact scenario: LNA designed for 20 dB gain, but at 5 GHz K = 0.87 — a low-level oscillation. The fix? Sacrificed 3 dB of gain, added an RC feedback network for stability compensation.
Reason 3: Power and Area
Higher gain typically means more transistor stages or higher bias current. In a mobile phone RFIC, the LNA power budget is typically only 5–15 mW — there’s no headroom for unlimited gain increases.
| Design Constraint | Low Gain (12–15 dB) | Mid Gain (16–20 dB) | High Gain (22–25 dB) |
|---|---|---|---|
| NF (typical) | 0.6–1.0 dB | 0.8–1.5 dB | 1.2–2.0 dB |
| OIP3 | Higher | Moderate | Lower |
| Stability | Easy | Needs attention | Needs careful compensation |
| Power | 3–8 mW | 8–15 mW | 15–30 mW |
| Best for | Strong interferers / wide dynamic range | General-purpose (most cases) | Weak-signal priority / power-insensitive |
4. What’s the Actual Relationship Between Sensitivity and NF?
At this point someone’s probably wondering: how exactly does total NF translate to sensitivity?
Sensitivity = -174 dBm/Hz + NFtotal + 10·log(BW) + SNRmin
- -174 dBm/Hz: Thermal noise floor (kT, the theoretical minimum at 290 K room temperature)
- NFtotal: Receiver total noise figure (dB)
- 10·log(BW): Noise power contribution from bandwidth
- SNRmin: Minimum SNR required for demodulation (modulation-dependent)
Typical SNRmin values by standard:
- GPS (BPSK): ~6–8 dB
- GSM / GPRS: ~9 dB
- Wi-Fi 6 OFDM: ~4–6 dB
- LTE QPSK: ~2–4 dB
- LoRa (SF=12): ~-20 dB (negative! The spreading gain is that powerful)
Now let’s plug in LTE 10 MHz bandwidth:
| Parameter | Scenario A (Good LNA) | Scenario B (Poor LNA) | Delta |
|---|---|---|---|
| NFtotal | 2.0 dB | 4.0 dB | 2 dB |
| BW | 10 MHz (→ 70 dB) | ||
| SNRmin | 3 dB | ||
| Sensitivity | -99 dBm | -97 dBm | 2 dB |
Two dB doesn’t sound like much? In free-space path loss terms, it translates to roughly 20% shorter cell coverage radius. For base station planning, that means you either build ~20% more sites to fill coverage gaps or accept degraded edge-of-cell user experience. In a handset conformance test under the same signal conditions, the poor-LNA phone will show one fewer bar of signal.
This is why semiconductor companies pour enormous R&D resources into LNA design — it directly determines the “sensitivity” number on the datasheet. And that number is one of the metrics customers care about most.
5. Practical PCB-Level Design Tips
At the board level, here are a few hard-won lessons worth writing down:
-
Don’t sacrifice too much matching for NF
Many engineers know the LNA input should be conjugate-matched for minimum NF (Γopt), but Γopt rarely sits at 50 Ω. If you force a match to Γopt, your return loss will be poor, potentially causing standing-wave oscillation between the LNA and the upstream SAW filter or duplexer.
My approach: strike a compromise between NF and S11. I typically accept 0.1–0.2 dB above the optimum NF in exchange for S11 better than -10 dB. Measured results confirm this trade-off delivers the best value.
-
Don’t overlook the insertion loss in front of the LNA
Many system designers account for antenna switch and duplexer insertion loss separately, but these losses stack directly onto total NF. A typical 4G duplexer Tx/Rx isolation path has ~0.8–1.5 dB insertion loss — meaning your carefully designed 0.8 dB NF LNA suddenly becomes 1.6–2.3 dB in the system.
This is why high-end solutions push the LNA as close to the antenna port as possible. Some architectures even adopt an “LNA first” layout, placing the LNA directly before the antenna switch.
-
Temperature effects on NF genuinely cannot be ignored
LNA NF degrades with rising temperature. The rough rule: every 10°C increase adds ~0.05–0.15 dB to NF. For automotive applications (-40°C to +105°C), high-temperature NF can be 0.8–1.2 dB worse than room temperature. If you only budget a 0.5 dB design margin, sensitivity will fail spec after the car sits in the summer sun for a few hours.
Tip: When running PVT simulations, don’t just check NF at the tt corner — always verify the ss corner + high temperature combination.
-
Leave enough headroom on gain compression point
As discussed earlier, gain and linearity trade off against each other. When selecting components, my rule of thumb: the LNA’s P1dB should be at least 10–15 dB above the maximum expected input signal level. For cellular bands, P1dB should ideally land in the -10 to 0 dBm range (depending on whether a front-end filter strips out-of-band blockers first).
6. Summary
Back to the original question: “Why is the first stage of the RF front-end so important?”
Answer: The Friis formula shows that the first stage’s NF is passed to the entire system almost unattenuated, while the first stage’s gain determines how much downstream noise gets “shielded” from the total.
So the RF front-end design priority is always:
LNA NF > LNA Gain > Everything else
It’s not that other stages don’t matter. It’s that when resources are finite — power, die area, BOM cost — you put your money where it makes the biggest difference.
For engineers working on UAV communication systems, software-defined radios, or industrial RF links, the same LNA-first principles apply. At Aomway, our high-performance FPV and long-range telemetry systems are designed with this exact Friis-first philosophy — every fraction of a dB in the LNA stage translates directly to usable range and link reliability in the field. If you’re developing RF front-ends for drone or industrial wireless applications and want to discuss LNA selection, matching network optimization, or system NF budgeting, feel free to reach out at [email protected].
Have questions about this article? Feel free to contact us at [email protected] — we’re happy to help!
Frequently Asked Questions
1. Why does the LNA noise figure dominate the total receiver NF?
Per the Friis formula, the noise contribution of each downstream stage is divided by the cumulative gain of all preceding stages. The LNA has no preceding gain to divide its NF, so its contribution passes through at full strength. Downstream contributions are divided by the LNA’s gain (63× in our example), reducing them to near-negligible levels. This is why a 0.5 dB degradation in LNA NF can translate to a ~2 dB total NF degradation — the LNA’s NF gets amplified by the chain, not attenuated.
2. What’s the practical difference between NF and noise temperature?
Noise Figure (NF) and noise temperature (Te) describe the same physical phenomenon in different units. The conversion is Te = T0 × (F – 1), where T0 = 290 K. A 1 dB NF (F = 1.259) corresponds to Te ≈ 75 K. Noise temperature is more commonly used in satellite communications and radio astronomy where the antenna looks at a very cold sky; NF is more common in terrestrial communications. Both follow the same Friis cascading logic.
3. Can I use multiple LNA stages in series for ultra-low NF?
Yes, but with rapidly diminishing returns. A two-stage LNA where the first stage provides 12 dB gain with 0.6 dB NF and the second provides 10 dB more gain with 1.5 dB NF will yield a total LNA NF of ~0.65 dB — the second stage’s contribution is negligible because it’s already divided by 15.8×. The real constraint is usually not the NF of additional stages but the linearity, stability, and power budget as discussed in Section 3. Multi-stage LNAs are common in ultra-low-NF applications like radio telescopes, but they come with significant OIP3 and stability challenges.
4. How does antenna mismatch affect the effective LNA NF?
An impedance mismatch at the LNA input changes the noise matching condition. The LNA’s datasheet NF is measured at Γopt (the source reflection coefficient for minimum NF). If the antenna presents a different impedance, the effective NF increases according to: NF(Γs) = NFmin + 4Rn|Γs – Γopt|² / (Z0|1+Γopt|²(1-|Γs|²)). This is why antenna-LNA co-design matters — a poor match can easily add 0.3–0.8 dB to the effective NF.
5. For FPV drone video links at 5.8 GHz, how much does LNA selection matter?
It matters enormously. A typical 5.8 GHz analog FPV receiver without an LNA might have a system NF of 8–10 dB, yielding sensitivity around -85 dBm at 6 MHz bandwidth. Adding a good LNA with 1.0 dB NF and 15 dB gain can bring system NF down to ~2.5 dB, improving sensitivity to roughly -92 dBm — a 7 dB improvement that can easily translate to 50%+ range extension. This is why premium FPV goggles and ground stations from manufacturers like Aomway include carefully tuned LNA front-ends: the LNA is literally the difference between a flyable video link and static at the edge of range.
