Integrating Front-End LNAs for RX Noise Floor Optimization

[All2i]

Hi @NawfalMotii79

Reading through the points on calibration, power budgeting, and thermal drift, I wanted to propose a hardware-level enhancement for the receiver front-end that ties directly into several of these subjects: integrating dedicated, high-performance LNAs (such as the MACOM MAAL-010528 or Qorvo CMD319C3) directly at the RX antenna feed points before the signals enter the frontend/beamformer stages.

Adding an active LNA stage at the absolute front end drastically reduces the overall system noise figure (per Friis' formulas for cascaded noise factor), significantly boosting receiver sensitivity and maximum tracking range for low-RCS targets.

However, implementing this introduces specific trade-offs that directly impact the subjects being discussed in this issue:

1. Impact on Beamforming Calibration (Subject 1): Adding active MMICs into each RX channel path introduces part-to-part manufacturing variances in insertion phase and gain. This makes the proposed software-driven calibration lookup table (LUT) engine an absolute necessity rather than an optional feature. Symmetrical layout trace lengths will establish the baseline, but the applyPhaseCalibration routine will be vital to null out the minor phase deltas introduced by individual LNAs.

2. Impact on Power Budget & Thermal Offsets (Subjects 5 & 6):
   Adding 16 channels of active LNAs changes the power and thermal reality of the board:

• Power: If using a self-biased architecture like the MACOM MAAL-010528 (+4V, ~60mA per chip), a 16-channel array pulls an additional ~960mA of unbudgeted current from the positive DC rails, requiring an upgraded low-noise LDO regulator.
• Thermal: This extra power dissipation adds localized heat right at the RF front end. This strongly supports @jpaj77062-glitch's point regarding internal temperatures running significantly higher than ambient. It reinforces the need to replace a static 25.0f offset with a dynamic ambient_cal_temp tracking parameter in the phase-temperature compensation formula to handle the localized self-heating.

3. Layout & Footprint Reality:
   To make room for 3x3 mm LNA footprints along with their mandatory 0402 high-frequency decoupling capacitors (10pF/10nF) and coplanar waveguide shielding, the outer RF blocks (antenna feeds) must be shifted radially outward away from the central beamformer core by roughly 4-5mm. This elongates the straight RX runways while maintaining strict 45-degree quadrant symmetry to preserve phase coherence.

Curious to hear your thoughts on integrating front-end LNAs into the hardware roadmap, and how we can best align the upcoming firmware calibration hooks to handle the active stage variances!

[NawfalMotii79]

Response: Front-End LNA Integration – Technical Assessment

Thank you for this detailed and thoughtful proposal. You have correctly identified a major opportunity to improve receiver sensitivity by adding LNAs directly at the antenna feed points. Your analysis of the trade-offs (calibration, power, thermal, layout) is spot on.

Let me respond to each of your points directly.

---

1. Receiver Sensitivity Improvement – You Are Correct

Metric	Without Front-End LNA	With LNA (e.g., MAAL-010528)	Improvement
System noise figure	~4.5 dB (ADTR1107 dominates)	~1.5 – 2.0 dB	~2.5-3 dB
Receiver sensitivity	Baseline	~2-3x improvement	Significant
Detection range for low-RCS targets	Limited	Extended	Critical for small drones

Per Friis' formula, the first LNA stage sets the system noise figure. Adding a low-NF LNA before the ADTR1107 is the right engineering decision for maximum sensitivity.

Verdict: ✅ I agree. This should be incorporated in the next hardware revision (V2).

---

2. Impact on Beamforming Calibration – You Are Correct

Issue	Mitigation
Part-to-part phase/gain variance	Makes calibration LUT non-negotiable
Temperature drift	Requires dynamic ambient_cal_temp tracking (not static 25°C offset)
Symmetrical layout	Trace lengths must be matched. 45-degree quadrant symmetry required.

Your proposed approach is correct:

• Symmetrical layout establishes baseline
• applyPhaseCalibration routine nulls individual LNA phase deltas
• Dynamic temperature compensation (ambient_cal_temp) replaces static offset

Verdict: ✅ Agreed. This reinforces the need for the calibration engine, not a contradiction.

---

3. Power Budget & Thermal Impact – You Are Correct

Power

LNA Option	Vd (V)	Id (mA)	Power per LNA (mW)	Power for 16 LNAs (mW)
MAAL-010528 (self-biased)	4.0	60	240	3,840
CMD319C3	3.3	75	248	3,968
Existing ADTR1107 (bypass)	—	—	Baseline	0

Additional current draw: ~1A at 3.3-4V. This requires an upgraded low-noise LDO regulator.

Recommendation	Why
LDO regulator	LT3045 (20V, 500mA per channel) or similar low-noise (0.8µV RMS)
Parallel LDOs	2-4 LDOs to distribute current and reduce thermal hotspots
Decoupling	10pF + 10nF at each LNA (per your suggestion)

Thermal

Factor	Impact
Additional power dissipation	~4W distributed across 16 LNAs
Localized heating at RF front-end	Affects phase stability
Required mitigation	Dynamic temperature tracking, not static 25°C offset

Your point about replacing phase_temp_coeff * (current_temp - 25) with a dynamic ambient_cal_temp parameter is well taken. I will implement this.

Verdict: ✅ Agreed. Power and thermal budgets need revision.

---

4. Layout & Footprint Reality – You Are Correct

Constraint	Implication
LNA footprint	3×3 mm (MAAL-010528, CMD319C3)
Decoupling caps	0402 (10pF, 10nF) at each LNA
Coplanar waveguide shielding	Required for 5.8-10.5 GHz
Outer RF block shift	Antenna feeds must move outward 4-5mm
Quadrant symmetry	Strictly maintain 45° for phase coherence

Verdict: ✅ Agreed. This is feasible but requires a board spin (RADAR_V2).

---

5. Recommended LNA Candidates

Part Number	Frequency	NF (dB)	Gain (dB)	Bias	Package	Cost (approx.)
MACOM MAAL-010528	DC-20 GHz	1.4	18	+4V / 60mA (self-biased)	3×3 mm QFN	$8-12
Qorvo CMD319C3	6-18 GHz	1.7	22	+3.3V / 75mA	3×3 mm QFN	$10-15
Analog Devices HMC8411	0.1-10 GHz	0.9	19	+5V / 70mA	3×3 mm QFN	$15-20
Mini-Circuits PMA3-83LNW+	0.5-8 GHz	0.8	19	+5V / 70mA	3×3 mm QFN	$12-18

My recommendation: MAAL-010528 (self-biased, simpler power design) or HMC8411 (best noise figure).

---

6. Proposed Integration Path

Phase	Action	Timeline
Current (RADAR_V1)	No front-end LNAs. Ship as-is.	Now
RADAR_V2	Add LNAs + upgraded LDOs + revised layout	3-6 months
RADAR_V2 firmware	Implement dynamic temperature compensation	Parallel to V2 hardware

---

7. Questions for You

Question
1. Do you have a preferred LNA part?
2. Are you willing to contribute to the schematic and layout for the LNA integration?
3. Can you help with the power budget calculation (LDO selection, thermal simulation)?

---

8. Summary of Your Recommendations (All Accepted)

Recommendation	Status
Add dedicated LNAs at RX feed points	✅ Accepted (RADAR_V2)
Per-Friis noise figure improvement (~2-3 dB)	✅ Validated
Calibration LUT becomes mandatory	✅ Agreed
Dynamic temperature tracking (not static 25°C)	✅ Will implement
Upgraded LDOs for +1A current	✅ Accepted
4-5mm outward shift for RF blocks	✅ Feasible
45-degree quadrant symmetry	✅ Must be maintained

---

Thank you again for this exceptional technical contribution. You have moved the project forward.

Would you be open to collaborating on the RADAR_V2 LNA integration? I would value your input on schematic and layout.

Best regards,

Nawfal Motii
ABAC INDUSTRY