Simulation and verification results
Link budget (requirement R2)
f-3dB = 138.2 MHz (R1 needs >= 50 MHz: PASS)
Best sensitivity in sweep: -46.60 dBm at M = 200 (sweep edge region; sensitivity is very
best sensitivity (-45.62 dBm vs best -46.60 dBm). Higher M buys <1 dB while moving
3) SENSITIVITY AT THE DESIGN POINT
Chosen design: Rf = 3300 ohm, Cf = 0.45 pF, M = 100, f-3dB = 138.2 MHz
Margin vs R2: 11.6 dB (sensitivity is that far below the -34 dBm requirement).
post gain x1: sensitivity = -45.62 dBm (delta vs x1 = +0.000 dB)
post gain x5: sensitivity = -45.66 dBm (delta vs x1 = -0.046 dB)
the x5 stage is comparator overdrive: at sensitivity the
Rf Cf(pF) f-3dB(MHz) Mopt sens@1e-6(dBm) Q@-34dBm verdict
for margin this application does not need (11+ dB already).
certain, not marginal. That is normal for a receiver with
~12 dB of sensitivity margin below -34 dBm and a 3.3k TIA on
pk-pk -- marginally over 2 ns. That number rests entirely on
worst-case sensitivity loss of 0.118 dB, vs a 1-sigma noise fraction
of 3.6 % at the R2 point and an R2 margin of 11.6 dB. Ripple-induced AM is well below
Same sensitivity coefficient, for temperature: at fixed bias
HV RIPPLE VERDICT: PASS (at design M, with temp compensation fitted)
BARE: A_eff = 0.196 mm^2 -> on-axis range to sensitivity = 19 cm.
i.e. 8.4 dB BELOW sensitivity: bare mode does not cover a
range to sensitivity = 2.84 m.
sensitivity (1/d^2 only; emitter off-axis falloff not modelled).
Chosen design: Rf = 3300 ohm, Cf = 0.45 pF (total, incl. parasitics), M = 100, f-3dB = 138.2 MHz
R1 bandwidth: f-3dB 138.2 MHz >= 50 MHz -- PASS
R2 VERDICT: PASS (locked Rf = 3.3 kohm design passes as originally specified)
TIA ngspice
f-3dB = 153.1 MHz peaking = 0.00 dB integrated output noise = 0.567 mV rms (link budget 0.569) PASS f-3dB >= 50 MHz (R1) PASS peaking < 3 dB PASS step settles to 5% within 1 UI (20 ns) PASS noise within 3 dB of link budget 0.569 mV VERDICT: PASS
HV flyback
switching: Vout=408 V open-loop, APD ripple=2.89 mVpp PASS switching Vout in 300..600 V band at design Ipk (open-loop power delivery class) PASS APD-node ripple < 20 mVpp (R9) PASS averaged loop: 150 V pre-step within 5% PASS averaged loop: settles within 2% of 420 V PASS averaged loop: overshoot < 5% VERDICT: PASS
Digital simulation
PASS: (a) zero-noise codec, 1e6 bits PASS: (b) BER monotonic non-increasing with SNR PASS: (c) PER sim vs analytic within 4-sigma bound PASS: (d) BER < 1e-6 at Q=6 SNR over 1e6 bits PASS: digital_sim -- all checks passed
Schematic netlist verification
netlist check: 630 connected pins, 213 nets, 530 expected connections PASS: netlist matches circuit.py exactly; structural checks pass
Link budget — BER vs incident power
TIA AC response (ngspice)
TIA AC response render not generated — run scripts/make_fab.sh.
Digital simulation
