Published Nov 13, 2026 · Last updated Nov 13, 2026 · 4 min · IndieRF
Bias Tee Design for GPS L1: 39 nH Choke, 46 dB Isolation
Lumped GPS bias tee at 1.575 GHz, solved: the choke and its SRF set the DC port isolation.
A bias tee for a GPS L1 active antenna built from a 27 pF DC block, a 39 nH choke with a 2.6 GHz SRF and a 100 pF bypass loses 0.046 dB at 1.575 GHz and holds the DC port 46.4 dB below the RF. From a 3.3 V supply it feeds 14.97 mA up the coax, and the antenna's amplifier sees 3.294 V after 6 mV across the choke's 0.4 Ω.
The sheet opens solved, with S21, S11 and S31 from 0.5 GHz to 3 GHz and the DC operating point. The parts are illustrative values, the same ones the bias tee example page uses, and not one vendor's part. Type your choke's L, Q and SRF from its datasheet into L1 and the plots follow. Change L alone, as the note on the sheet says, and 19.5 nH reads −40.5 dB, because the 2.6 GHz SRF stays where it was.
The choke and the bypass divide the RF
Isolation is a voltage divider. The choke runs from the RF line to the feed node, and the 100 pF bypass holds that node close to ground. At 1.575 GHz the 39 nH choke is 386 Ω of reactance, and its winding capacitance raises that to about 610 Ω, since the band sits at 0.61 of the SRF:
The bypass is past its own series resonance at 796 MHz (100 pF with 0.4 nH of ESL), so at 1.575 GHz it is 2.95 Ω, mostly inductive. The divider gives dB, within 0.2 dB of the engine's −46.4 dB. Double the choke's impedance or halve the bypass's and S31 drops about 6 dB.
The DC block sits on the other side of its resonance. 27 pF with 0.4 nH of ESL is series resonant at 1.53 GHz, so C1 is 0.37 Ω in the 50 Ω line at the band, and S11 reads −27.4 dB.
Pick the choke by its SRF
Swap L1 and the numbers below come out of the same engine. Bigger chokes resonate lower, so each row keeps the 39 nH part's winding capacitance and moves the SRF with it.
Scroll sideways for more columns
| L1 | SRF | S31 at 1.575 GHz | S21 at 1.575 GHz | S31 at 2.4 GHz | S21 at 2.4 GHz |
|---|---|---|---|---|---|
| 10 nH | 5.13 GHz | −31.9 dB | −0.295 dB | −31.6 dB | −0.173 dB |
| 19.5 nH | 3.68 GHz | −38.3 dB | −0.101 dB | −39.8 dB | −0.073 dB |
| 27 nH | 3.12 GHz | −41.9 dB | −0.066 dB | −45.6 dB | −0.053 dB |
| 39 nH | 2.60 GHz | −46.4 dB | −0.046 dB | −57.5 dB | −0.041 dB |
| 56 nH | 2.17 GHz | −52.1 dB | −0.037 dB | −57.0 dB | −0.035 dB |
| 82 nH | 1.79 GHz | −61.8 dB | −0.032 dB | −49.3 dB | −0.031 dB |
| 120 nH | 1.48 GHz | −69.6 dB | −0.030 dB | −46.4 dB | −0.029 dB |
At 1.575 GHz more inductance buys more isolation down the whole table, even at 120 nH, where the SRF sits below the band: a choke's impedance peaks at self-resonance and falls slowly on both sides, and 0.096 pF is still about 1 kΩ at 1.575 GHz. At 2.4 GHz the order changes. The 39 nH and 56 nH chokes, resonant at 2.60 GHz and 2.17 GHz on either side of 2.4 GHz, give −57.5 dB and −57.0 dB, while the 120 nH choke gives −46.4 dB. If the same tee has to keep 2.4 GHz out of the supply, pick the choke whose SRF lands near 2.4 GHz rather than the biggest one.
Insertion loss barely moves: every choke from 19.5 nH up loses 0.101 dB or less at both frequencies. The 10 nH choke costs 0.295 dB and gives only −31.9 dB of isolation at the band.
The table holds Q at 50 and the winding capacitance at 0.096 pF for every value. Parts in one real series differ in both, so read the SRF column of your choke's datasheet and type it in. The trend carries over and the exact dB does not.
The DC side is Ohm's law
The antenna's amplifier sits on the sheet as 220 Ω, about 15 mA at 3.3 V. The supply sees 0.4 Ω + 220 Ω, so 3.3 V drives 14.97 mA and the choke's Rdc drops 6 mV. The sheet has no cable between the tee and the antenna, so for a long run add the coax's loop resistance to Rdc.
The cable is also the next RF loss after the tee: Coax Loss on 5 m of RG-174 at 1575 MHz. Belden prints RG-174 attenuation to 1 GHz, so the tool marks 1575 MHz as extrapolated.
References
- D. M. Pozar, Microwave Engineering, 4th ed., section 4.3: the scattering matrix of a network from its admittances. The note's numbers are the schematic engine's, checked against a numpy nodal solve of the same three nodes and against ngspice 47 at 1.575 GHz.
- Analog Devices, AN-2061: Wideband Bias Tee Design Using 0402, SMD Components.
Related
Questions? Contact
Discussion
No comments yet. Start the thread with a measurement, a correction, or a worked example.