In discussions with members of SARA, some of them have discussed that my mechanical Tohtsu CX-520D will degrade the signal too much, if I use it in front of the LNA to switch a reference signal in and out. Is this correct?

Short answer: you’re right to be cautious — the CX-520D does add loss, but “degrade the signal too much” depends on how fussy your system is. The datasheet/specs show the CX-520D insertion loss is very small at L-band (≈0.15–0.25 dB around 1–1.5 GHz), so the noise-figure penalty is roughly equal to that loss (i.e. ≈0.2 dB NF penalty). That is small in absolute terms but not negligible if you’re chasing the lowest possible system temperature. (funkelektronik.eu)

Below I give the numbers, how they were calculated, and practical comments so you can judge whether it’s “too much” for your setup.

---

1) What the CX-520D actually specifies

Manufacturer / reseller data show:

• Insertion loss: ~0.15 dB @150 MHz, ~0.18 dB @500 MHz, ~0.20 dB @1 GHz and ≈0.20–0.25 dB up to ~1.5–2 GHz (so at 1.420 GHz expect ≈0.2 dB).
• Isolation and VSWR are good (isolation tens of dB; VSWR ≈1.03–1.1 depending on freq).
• Coil: 12 V DC, coil current ≈160 mA, life ≈1e6 operations. (wimo.com)

(links above are the CX-520D datasheets / product pages.)

---

2) How that loss affects noise figure / sensitivity (numbers)

Rule of thumb and basic theory: a passive loss L (dB) placed before the first amplifier produces a noise-figure penalty roughly equal to that loss (in dB). That follows from Friis / passive-component NF = loss rule. (microwaves101.com)

I ran three representative cases (typical LNA NFs are used):

• insertion loss (switch) = 0.2 dB (realistic for CX-520D at 1.42 GHz)
• also show 0.5 dB and 1.0 dB for comparison (solid-state or poorer relays / long coax etc.)

Calculated results (ΔNF is how much extra NF the switch adds):

Switch loss (dB)	LNA NF (dB)	System NF after switch (dB)	ΔNF (dB)
0.2	0.5	0.70	+0.20
0.2	1.0	1.20	+0.20
0.2	2.0	2.20	+0.20
0.5	0.5	1.00	+0.50
1.0	0.5	1.50	+1.00

(Computed using Friis/passive-loss rules; passive loss multiplies the LNA’s noise factor so NF increase ≈ the passive loss in dB.) (Engineering LibreTexts)

Converted to equivalent noise temperature (useful for radio astronomy): with T0=290 K, a 0.2 dB system NF ~0.7 dB corresponds to an effective receiver noise temperature of roughly 50–95 K depending on the LNA NF you started with (see table details in the calculations). In short: 0.2 dB ≈ a modest rise in system temperature; 1 dB would be a substantial hit. (See references above for the conversion formulas.) (reeve.com)

---

3) Practical interpretation for hydrogen-line work (your situation)

• If your LNA is already very low NF (e.g., 0.3–0.7 dB) and you’re trying to detect very small antenna temperature changes (mK–K level), an added 0.2 dB penalty is measurable and matters.
• If your whole system NF is a few dB already (e.g., modest LNA, long feed cable, SDR front end), then +0.2 dB is small in percentage terms and usually acceptable.
• The dominant penalty equals the switch loss — so CX-520D’s ~0.2 dB is small compared with many other real-world losses (long coax, connectors, balun losses). (wimo.com)

---

4) Other non-NF effects worth considering

• VSWR / matching: the CX-520D shows good VSWR at L-band; poor matching can cause ripple in your spectrum or standing-wave problems. Specs show VSWR ≈1.03–1.1 in the band. (wimo.com)
• Isolation when in load position: the relay’s isolation (tens of dB) is usually fine for an antenna↔load switch, but check the datasheet at your exact frequency if you need very high isolation for your calibration scheme. (wimo.com)
• Contact aging, moisture, microphonics: mechanical relays can develop higher loss/mismatch if contacts corrode or if the relay gets wet. Weatherproofing and occasional verification are important.
• Switch location: putting the relay at the antenna feed (mast head) keeps cable losses common-mode — that’s usually best for calibration accuracy despite the small NF penalty. Putting an attenuator or poor switch between antenna and LNA at lower point typically hurts more because of extra cable loss. (on4sh.be)

---

5) Recommendations / options

1. If you want the simplest & lowest ongoing risk: keep the CX-520D if the ~0.2 dB penalty is acceptable for your sensitivity goals — it’s a reasonable, low-loss mechanical relay for mast mounting. (funkelektronik.eu)
2. If you need absolute lowest noise: consider calibration alternatives that do not add loss before the LNA:
   • noise-diode injection via a directional coupler (inject a calibrated noise source into the line after a very small coupling loss), or
   • a custom front-end (integrated radiometer module) that implements switching and loads inside a carefully engineered low-loss front end (Spacek Labs / research front ends). These options avoid putting an appreciable passive loss directly in front of the LNA. (eBay UK)
3. If you stick with a mechanical relay: minimise additional cable/connector loss — mount the relay at the mast head, use N-type weatherproof connectors and a short jumper to the LNA. Verify the relay’s insertion loss and VSWR at 1.42 GHz with a VNA if you require firm numbers. (wimo.com)

---

6) Final bottom line (TL;DR)

• The CX-520D will not ruin your signal: it typically adds ≈0.2 dB of loss at L-band, which produces ≈+0.2 dB noise-figure penalty. That is small but measurable.
• If your system aims for the lowest possible NF (mK-level radiometry), use a different calibration method (noise-diode injection / integrated radiometer front-end) or accept the small hit and ensure excellent weatherproofing and minimal extra cable. (wimo.com)

---