Gerber'd quasi-Yagi — SMA on PCB (rear tab or emitter) / pigtail (stand-mount)

Interactive 3D model — auto-rotates, drag to orbit

Statistical resolution limit (SRL) — closest two targets tellable apart vs SNR · hover a point (session 014z DF model)

gerber_yagi_sma_gerber_rg402.zip gerber_yagi_sma_rg402.step gerber_yagi_sma_3d_rg402.html gerber_yagi_sma_rg402.json gerber_yagi_pigtail_gerber_rg402.zip gerber_yagi_pigtail_rg402.json gerber_yagi_emitter_sma_gerber.zip gerber_yagi_emitter_sma.json gerber_yagi_sma_offset.json gerber_yagi_sma_feedx.json gerber_yagi_sma_trim.json

Cable routes compared — antenna params per route (FDTD PML_8)

cable route	gain	F/B	res	S11 @915	Zin @915
bare element — no hardware (reference)	8.87 dBi	10.0 dB	902 MHz	−24.3 dB	54+5j Ω
REAR-TAB SMA @ (−23.5, −6) + straight RG-402 jumper	8.94 dBi	9.8 dB	908 MHz	−26.1 dB	53+4j Ω
… + λ/2 cable down the stand (worst case)	8.93 dBi	9.8 dB	908 MHz	−26.2 dB	53+4j Ω
PIGTAIL — RG-402 through the Ø6.5 exit + λ/2 down the stand	8.95 dBi	9.8 dB	908 MHz	−26.4 dB	52+4j Ω
EMITTER direct-solder SMA (retuned l_drv 140)	8.85 dBi	9.8 dB	~915 (double-dip)	−17.9 dB	—
… + λ/2 cable straight down	8.76 dBi	9.6 dB	~915 (double-dip)	−18.5 dB	61−7j Ω

All rows FDTD openEMS PML_8, large air box, port at the emitter as always; hardware modelled as the real-size body + post blades + jumper/cable EXTERIOR (the coax interior is 50 Ω by construction). The two rear-tab routes are CABLE-INDEPENDENT — adding the λ/2 rod moves S11 by ≤0.1 dB — because the exit sits in the reflector's field-quiet shadow and the slit balun holds off common-mode. The emitter-jack build trades that: the body under the feed raises R (matched a shade shallower, −18 dB class) and it stays cable-independent only for the vertical exit — a parallel-to-board run re-tunes −12 MHz and follows cable dressing (see the routing study).

Run-line offset — is y=−6 optimal, and can the run be angled? (FDTD PML_8)

run-line placement (SMA + λ/2 cable in every row)	gain	F/B	res	S11 @915	Zin @915
parallel y=−4 (shield edge 1.2 mm off the slot)	9.03 dBi	10.7 dB	928 MHz	−26.5 dB	54−3j Ω
parallel y=−5	8.98 dBi	10.0 dB	910 MHz	−30.1 dB	52+3j Ω
parallel y=−6 — SHIP	8.93 dBi	9.8 dB	908 MHz	−26.2 dB	53+4j Ω
parallel y=−7	8.95 dBi	9.8 dB	908 MHz	−28.1 dB	52+3j Ω
parallel y=−8 (posts at the pour edge — limit)	8.96 dBi	9.8 dB	910 MHz	−28.7 dB	52+3j Ω
ANGLED, emitter end pulled to y=−4	9.35 dBi	11.1 dB	918 MHz	−20.4 dB	42−4j Ω
ANGLED, emitter end pulled to y=−3	4.26 dBi	0.8 dB	978 MHz	−7.8 dB	119+3j Ω

Is y=−6 optimal / how tolerant? The parallel run is a FLAT optimum: y=−5…−8 are indistinguishable (±0.05 dB, ±2 MHz) — placement is non-critical, ±1 mm changes nothing; y=−6 is simply the middle of the return strip. At y=−4 the shield starts loading the slotline (res +20 MHz — it becomes a tuning knob, though the 915 match survives via the double-dip), and past −8 the flange/posts leave the pour — so the usable window is y=−5…−8. ANGLED run (emitter end pulled toward the slot to shorten the exposed centre jump): end at −3 destroys the antenna in BOTH boundaries (the grounded shield converging on the slotline's open end unbalances the balun; the cable radiates common-mode). End at −4 looks fine in PML but MUR and PML disagree wildly there (9.4 vs 5.7 dBi) — the config sits on a resonant cliff ~1 mm of fab tolerance from catastrophe. Verdict: keep the run PARALLEL; the ~2 mm of exposed centre an angle would save is ~1.5 nH ≈ j9 Ω, absorbed by the match.

Getting resonance exactly to 915 — solder-point sweep + the tip-trim recipe (FDTD PML_8)

Asked: move the centre-wire solder point to put resonance AT 915. Modelling the drive WHERE THE WIRE ACTUALLY CROSSES the slot (a localized 4 mm port instead of the historical full-arm-width port) first shifts the truth: the as-built antenna resonates ~892 MHz, not 908 — the wide port under-reported the drive-point reactance (+18j at 915).

centre-wire solder crossing x (localized 4 mm drive; SMA + λ/2 cable)	res	S11 @915	Zin @915
full-arm-width port — the historical idealization	908 MHz	−26.2 dB	53+4j Ω
x=52	886 MHz	−14.0 dB	50+20j Ω
x=54	888 MHz	−14.7 dB	52+19j Ω
x=56	890 MHz	−15.3 dB	54+18j Ω
x=58 — SHIP (the knob's maximum)	892 MHz	−15.4 dB	55+18j Ω
x=60	892 MHz	−15.3 dB	56+17j Ω
x=62	890 MHz	−14.9 dB	57+18j Ω
x=64 (near the arm trailing edge)	872 MHz	−10.1 dB	57+34j Ω

The crossing position is a WEAK knob: 872–892 MHz over the whole arm, with its maximum exactly at the shipped x=58–60 (the balun-length gain and open-slit-stub loss cancel there). It cannot reach 915. What does reach 915 is the driven LENGTH — still an assembly-time adjustment on a fabbed board: trim equal copper off each arm tip (~5–6 MHz per mm of total length).

driven length (= tip trim) @ solder x=58	trim per tip	S11 @915	Zin @915
l_drv 135.4 — as-drawn	—	−15.4 dB	55+18j Ω
l_drv 132 — PICK: res 915	1.7 mm	−30.5 dB	52+2j Ω
l_drv 130	2.7 mm	−24.6 dB	51−6j Ω
l_drv 128 (overshot)	3.7 mm	−14.4 dB	48−19j Ω

RECIPE: keep the solder crossing at x=58, TRIM 1.7 mm OFF EACH DRIVEN-ARM TIP → res 915.0, S11 −30.5 dB, Zin 52+2j (PML_8, SMA + λ/2 cable; same for the pigtail build). The trim marks are drawn dashed on the feed figure; READMEs in both Gerber zips carry the step. Trim symmetrically and sneak up on it — 0.5 mm of trim ≈ 5 MHz.