What is best type of power supply to use for bias-tees that power LNAs at dish feed in hydrogen line systems?

Short answer: use the quietest DC you can. In practice that means:

• Best: a battery (LiFePO₄ 12 V) feeding an ultra-low-noise LDO mounted right by the bias-tee/LNA (e.g., LT3042/LT3045/ADM7150 set to the LNA’s voltage).
• Second best: a mains linear (transformer) supply → LC/π filtering → the same ultra-low-noise LDO near the feed.
• Only if you must: a switch-mode supply followed by aggressive filtering and a post-reg LDO at the mast.

Why (and how)

Hydrogen-line work is extremely sensitive to conducted and radiated hash. Switching PSUs produce combs that fold into your passband; LNAs happily conduct this back onto the coax. The cure is low source noise and point-of-load regulation.

A practical, robust stack

• Source: 12–13.8 V LiFePO₄ (or a transformer-based linear bench supply). Size it to ≥3× the LNA current (typ. 50–200 mA per LNA).
• At shack (before the bias-tee):
  • Common-mode choke on the DC lead.
  • π-filter: 10–22 µH → 47–100 µF low-ESR // 0.1 µF → 10–22 µH.
  • Metal enclosure, feedthrough capacitors for the DC pass-through.
• At the mast/feed (as close to the LNA as possible):
  • LDO module (LT3042/3045, ADM7150/1) set to the LNA voltage (5 V is common; some pro LNAs want 12–15 V).
  • 10 µF + 0.1 µF right on the LNA supply pins; add a small series RC snubber if you still see spikes.
• Bias-tee DC leg: large-SRF inductors or multi-section choke so the DC path is high-Z at 1420 MHz; RF leg with low-loss RF capacitors (NP0/C0G).
• Protection: resettable polyfuse sized ~2× LNA current, and a TVS diode on the DC line at the mast.
• Grounding: star ground in the shack; choke the coax (ferrite sleeves) to kill common-mode.

If you must use an SMPS

• Choose an isolated, medical-grade unit.
• Put it in a screened box; add two common-mode chokes, a π-filter, then the LDO (with ≥0.5–1 V headroom).
• Verify on your waterfall: a “comb” at the switching fundamental and its harmonics means you need more filtering or a different supply.

Targets & checks

• Aim for <1 mVpp ripple at the LNA pins (10 Hz–100 kHz) after the LDO.
• Watch your SDR FFT near 1420.405 MHz for wandering spurs when you swap supplies.
• Run 12–15 V up the coax and drop locally with the LDO; avoid DC-DC at the mast unless it’s followed by a very good LDO and filtering.

Conclusion – optimal choice:
For hydrogen-line LNAs powered through a bias-tee, battery or linear → π-filter → ultra-low-noise LDO at the feed is the gold standard. It’s quiet, simple, and dramatically reduces the chance of PSU-borne lines in your spectra.