BJT LNA design for 2.4 GHz: 13.2 dB gain and 1.89 dB NF from a SPICE model card

Biased at 5.39 mA from 3 V, this Gummel-Poon NPN stage gives 13.23 dB of gain and 1.89 dB NF at 2.45 GHz, with S11 at -13.3 dB and S22 at -18.6 dB. Every small-signal value comes from the transistor's .model card at the operating point the DC run solves, so pasting another card moves the bias, the gain, the noise and K together.

Open in the simulator

Change this: short R2, the 10 ohm in series with the collector feed. The lowest K from 0.1 to 6 GHz falls from 1.22 to 0.23 at 150 MHz, and with K under 1 some passive source or load makes the stage oscillate.

-20020123456GHzdB
S21S11S22Circles mark the table's numbers.
QuantityValueRead at
S21, gain13.23 dB2.45 GHz
S11, input match-13.31 dB2.45 GHz
S22, output match-18.62 dB2.45 GHz
Noise figure from 50 ohm1.89 dB2.45 GHz
Lowest K, Rollett stability factor1.2150.1 to 6 GHz, lowest at 6 GHz
Lowest K with R2 shorted0.234lowest at 150 MHz
Collector current5.39 mADC
Collector-emitter voltage2.946 VDC

Every number here is the simulator's own, computed when the site was built from the same circuit the button opens. How the engine is checked.

The circuit

P1Z=50C1C=4.7pL1L=1.5nR1R=33kQ1BF=110L3L=0.68nGND1GND2V1V=3GND3L2L=4.7nR2R=10C2C=1pC3C=0.33pGND4P2Z=50

How the bias reaches S and NF

The DC run solves the operating point first: 3 V through R1, 33 kohm, sets Ic at 5.39 mA with Vce at 2.946 V. The S sweep then runs Q1 linearized at that point: gm, the base and collector capacitances and the substrate capacitance CJS come from the card at that bias, and the shot noise from the solved currents. Q1's substrate pin goes to ground, as a three-node Q in a vendor macro ties it.

The card

Illustrative values written by IndieRF, not a real part. Paste it, or your vendor's .model card, into Q1 in the inspector. A vendor .subckt with a four-node Q goes in through File, Import SPICE subcircuit.

* Illustrative RF NPN, hand-written by IndieRF. Not a real part.
.model RFNPN_DEMO NPN(IS=1.5e-16 BF=110 NF=1 VAF=35 IKF=0.12 NKF=0.5
+ ISE=1.2e-14 NE=1.8 BR=6 NR=1 VAR=3 IKR=0.02 ISC=4e-15 NC=1.7
+ RB=6 RBM=2.5 RE=0.45 RC=3.2
+ CJE=0.55p VJE=0.85 MJE=0.42 TF=5.5p XTF=6 VTF=0.9 ITF=0.06 PTF=15
+ CJC=0.12p VJC=0.62 MJC=0.33 XCJC=0.35 TR=0.8n FC=0.85
+ CJS=0.28p VJS=0.55 MJS=0.3)

The card is IndieRF's own illustrative values, not a real part, and every L and C is ideal, so the loss of a real input inductor adds directly to the NF of a built stage. The transistor runs at 27 °C with noise referenced to 290 K, and its substrate junction carries no DC current.

Other solved examples

Question about this circuit, or a correction?