Good question, Andrew — short answer: yes, up to a point, a larger corner reflector will generally improve hydrogen line observing, but there are trade-offs.

Here’s the breakdown:

---

1. Collecting Area (Aperture Size)

• Sensitivity in radio astronomy scales with the effective collecting area of your antenna.
• A larger corner reflector provides a bigger aperture, capturing more of the faint 1420 MHz hydrogen line emissions.
• Doubling linear size increases the effective aperture by about four times (area ∝ size²), which can significantly improve signal-to-noise ratio (SNR).

---

2. Gain and Beamwidth

• A larger reflector narrows the beamwidth, concentrating reception in one direction.
• That’s good for rejecting terrestrial interference and increasing gain (better SNR).
• But: a very narrow beam means the sky coverage is limited, so you’ll need precise pointing if you want to track sources (e.g., Galactic plane).

---

3. Impedance Matching & Feed Position

• The corner reflector must be properly scaled to λ = 21 cm:
  • Reflector spacing typically ~0.25λ to 0.5λ behind the driven element (e.g., a dipole).
  • Corner angle (usually 90°) should be maintained for best reflection efficiency.
• If the geometry isn’t scaled well, larger size won’t necessarily improve things — you could get mismatched lobes or reduced efficiency.

---

4. Practical Considerations

• Bigger = heavier + more wind loading. At 1.42 GHz, even a ~2–3 m wide corner reflector is already physically large and challenging to mount.
• Mechanical precision isn’t as critical as at microwave frequencies, but alignment still matters.
• Beyond a certain point, it may be more practical to switch to a parabolic dish or an array (e.g., Yagis, or phased dipoles).

---

✅ Bottom line:
A larger corner reflector does improve hydrogen line observing by boosting gain and reducing noise pickup. But if you go much beyond a few wavelengths in size (say 2–3 m across at 21 cm), the mechanical challenges outweigh the performance gains, and a parabolic dish or phased array may be a better investment.

---

Do you want me to run some quick gain and beamwidth estimates for corner reflectors of, say, 1 m, 2 m, and 3 m aperture at 1420 MHz so you can see the trade-offs numerically?

Great — here are ballpark estimates for a 90° corner reflector at 1420 MHz (λ ≈ 0.21 m), assuming a square mouth of width D and a modest aperture efficiency η ≈ 0.5. These are useful first-order planning numbers:

Mouth width D	Gain (dBi)	3 dB Beamwidth (°)
1.0 m (~4.8 λ)	~21.5 dBi	10.6°
2.0 m (~9.5 λ)	~27.6 dBi	5.3°
3.0 m (~14.3 λ)	~31.1 dBi	3.5°

How these were estimated

• Gain: ( G \approx 10\log_{10}!\left(\dfrac{4\pi,\eta,A_\text{proj}}{\lambda^2}\right) ), with (A_\text{proj} \approx D^2) and η = 0.5 (typical for a practical build).
• Beamwidth (HPBW): ( \text{HPBW} \approx 50.6^\circ \dfrac{\lambda}{D} ) (square aperture, uniform-ish illumination).

Quick design tips @ 1420 MHz

• Driven element: start with a ½-wave dipole ≈ 0.105 m tip-to-tip (each leg ~52.5 mm) and trim for match with your feed + balun.
• Feed position from the corner apex (along the bisector): 0.25–0.35 λ (≈ 5–7 cm) is a good starting range; tweak for best SWR and pattern.
• Corner angle: keep it at 90°.
• F/B and sidelobes: expect >20 dB F/B if the geometry and feed are tidy; absorptive edge treatment can tame sidelobes.
• Mechanical reality: going beyond ~2–3 m brings serious wind load and alignment hassles; at that point, dishes or phased arrays often win on practicality.

If you want, I can also translate those gains into expected SNR improvements for H I drift scans given your RSPduo/GNU Radio setup (assuming bandwidth & integration time), and suggest a feed-to-SDR matching network to keep system temperature low.

Awesome — here’s a practical, numbers-first view tailored to your RSPduo + GNU Radio setup at 1420.405 MHz.

Drift-scan H I (Milky Way) — brightness-temperature view

For diffuse Galactic H I, the line fills your beam, so detectability is governed mainly by system temperature and (channel bandwidth × integration time) — not aperture — once the beam is sky-filled. Use the radiometer equation:

[
\Delta T_\mathrm{rms}=\frac{T_\mathrm{sys}}{\sqrt{B,t}}
]

Typical choices and results (per spectral channel):

• Channel widths (velocity resolution):
  1 kHz → 0.21 km/s; 5 kHz → 1.06 km/s; 10 kHz → 2.11 km/s; 20 kHz → 4.22 km/s.
• Example noise (ΔT_rms) for common (T_\mathrm{sys}) values:

(T_\mathrm{sys})	(t)	(B)	(\Delta T_\mathrm{rms})
80 K (good LNA)	600 s	5 kHz	0.046 K
80 K	3600 s	5 kHz	0.019 K
150 K (OK LNA)	600 s	10 kHz	0.061 K
300 K (no LNA)	600 s	10 kHz	0.122 K

Galactic H I line brightness is typically 5–20 K off-plane and 30–100 K on the plane. Even conservatively, with (T_\mathrm{sys}=150) K, (B=10) kHz, (t=600) s, you get:

• Off-plane 5 K line: SNR ≈ 5 K / 0.061 K ≈ 82
• Plane 30 K line: SNR ≈ 490
  So detection and clean profiles are straightforward. Upsizing the antenna won’t change these SNRs much for beam-filling H I; improving (T_\mathrm{sys}), RFI rejection, and longer (t) do.

Point-source / small-source H I (resolution or extragalactic) — flux view

Here SNR does scale with effective area. A handy figure is SEFD:

[
\mathrm{SEFD}=\frac{2kT_\mathrm{sys}}{A_\mathrm{eff}} ;;(\mathrm{Jy})
\quad\text{with}\quad A_\mathrm{eff}\approx \eta A_\mathrm{geo},;\eta\sim0.5
]

For a 90° corner reflector (square mouth (D)):

(D)	(A_\mathrm{eff}) (η=0.5)	SEFD @ 150 K	Relative SNR vs 1 m
1 m	0.50 m²	~8.3×10⁵ Jy	1.0×
2 m	2.00 m²	~2.1×10⁵ Jy	4.0× (~+6 dB)
3 m	4.50 m²	~9.2×10⁴ Jy	9.0× (~+9.5 dB)

Extragalactic H I peaks are usually ≲0.1–0.5 Jy for bright nearby galaxies at km/s-class resolution — so with SEFD ≳10⁵ Jy you’re still far from single-dish extragalactic work. Arrays/correlation change the game; for single-aperture total-power, stick to Galactic H I unless you massively scale up and cool the front end.

What helps SNR the most (in order)

1. Lower (T_\mathrm{sys}): front-end LNA at the feed (NF ≤ 0.5–0.7 dB, 20–30 dB gain), then a band-pass around 1420 MHz to kill out-of-band RFI.
2. Clean spectral channels: choose (B) to balance velocity resolution and noise (5–20 kHz is a sweet spot for drift scans).
3. Integrate longer: SNR grows as (\sqrt{t}).
4. Stable total-power chain: thermal/mechanical stability, reference switching if possible.

A solid RF chain (feed → RSPduo)

• Driven element: ½-wave dipole (~105 mm tip-to-tip) at the corner-reflector focus.
• Balun & match: A 1:1 current balun at the feed, plus a tiny L-match to bring ~73 Ω dipole to 50 Ω:
  • Series L ≈ 5.5 nH, shunt C ≈ 1.5 pF at 1420 MHz (low-Q, layout-critical).
  • Or skip the L-match and accept ~1.5:1 SWR; the LNA will hardly notice.
• LNA #1 (at the feed): NF ≤ 0.5–0.7 dB, Gain ~25 dB, handles local RFI without compression. Power via bias-tee.
• Band-pass filter: ~1420 ± 5–10 MHz (SAW or cavity) right after LNA #1 to protect downstream stages.
• LNA #2 (optional): +10–20 dB to overcome cable loss; follow with a 6–10 dB pad before the RSPduo to keep gain linear.
• Ferrites / CMC: clamp-on ferrites near the feed & at the shack to kill common-mode.
• Coax: low-loss 50 Ω (e.g., LMR-240/400 class for long runs).

Corner-reflector size: what you “get” in practice

• 1 m (≈ 4.8 λ): ~21.5 dBi, HPBW ≈ 11°, excellent for wide sweeps, very easy HI.
• 2 m: ~27.6 dBi, HPBW ≈ 5.3°, tighter RFI rejection, nicer beam mapping; 4× flux SNR vs 1 m.
• 3 m: ~31.1 dBi, HPBW ≈ 3.5°, wind & mechanics start to bite; 9× flux SNR vs 1 m.

Quick recommendations for your setup

• If your goal is clean Galactic H I spectra and drift maps, prioritize front-end LNA + filter + stability over making the reflector huge. A 1–2 m corner is a great sweet spot.
• If you want narrower beams for mapping structure or rejecting local RFI, go 2–3 m.
• For extragalactic H I with a single aperture, you’ll need much more area (or go to interferometry—which you’re already set up for with the RSPduo).