Mean Well HLG Constant-Current Mode Wrong for Bitaxe

Power off at the wall before touching anything on the HLG. The brick stores residual charge in its bulk caps for several seconds after wall power is removed. The HLG-series uses sealed M4 screw terminals or pre-attached output leads — do not probe live. Remove conductive jewelry. Use insulated tools throughout. The IP67 housing has small gaps where bare conductor is exposed inside; respect the brick's sealed nature.

Read the full HLG model number off the white label and photograph it. The trailing suffix (`-A`, `-AB`, `-B`, `-BC`, none) determines whether the unit has a CV/CC switch and trim pots. Plain `HLG-Hxxx-12` (no suffix) has no slide switch — jump to Tier 2 measurement steps. `-A`/`-AB`/`-BC` SKUs have the switch and pots accessible under the side cover.

Locate the CV/CC slide switch under the side cover or sticker. Peel back the rubber/silicone cover or sticker on one side of the housing. This voids the IP67 seal — only do it if the brick is in indoor service or you plan to reseal with electronics-grade silicone. Photograph the switch position before touching it. Labelling varies by production run — typical positions are `CV`/`Vo`, `CC`/`Io`, or `Vo+Io`. For a Bitaxe load, `CV` (or `Vo`) is the only correct setting.

Slide the switch to `CV` using a small non-metallic tool — plastic spudger or wooden toothpick. The throw is short (a few mm). Confirm the switch has fully clicked into the new position. Power on. Connect Bitaxe. Watch AxeOS dashboard hashrate over `5 minutes`. If it climbs to expected nameplate, the diagnosis is confirmed and the fix is locked in.

Re-seal the side cover with a thin bead of clear electronics-grade silicone (`Dow 738` or `Permatex 81158`) if the brick is in outdoor or wet-environment service. RTV-1 silicone restores IP67-equivalent protection. Allow `24 hours` cure before exposing to wet conditions. Skip this step entirely if the HLG lives indoors. Verify the switch can't be inadvertently bumped.

Multimeter on DC volts at the HLG output terminals, no load: expect `12.0 ± 0.6 V`. Connect Bitaxe, let stabilize at full hash for `2 minutes`, re-measure at the *miner end of the cable* (not at the PSU). Healthy under-load reading: `11.7-12.2 V` for Gamma, `11.5-12.0 V` for Hex. Anything below `11.5 V` indicates current-limit, undersizing, or cable resistance.

On `-A`/`-AB` SKUs, check both trim pots (`Vo` and `Io`) located near the slide switch under the side cover. Each is a multi-turn potentiometer (`~15` turns end-to-end). Crank the `Io` pot fully clockwise (maximum current ceiling) using a small flathead — for any Bitaxe load, there is no scenario where you want CC limiting active. Document the new pot position in your build log.

Adjust the `Vo` pot to compensate for cable drop. With Bitaxe running at full hash, measure rail voltage at the *miner* end. If reading is `11.6-11.8 V`, turn `Vo` clockwise in `1/4-turn` increments until miner-end voltage reads `12.0 V`. Stay within the brick's ±10% trim range or you will trip OVP. This compensates for `~1 m` of `18 AWG` lead resistance under `12 A` load.

Inspect cable connectors. XT30, XT60, barrel jack, screw terminal — all show degraded contact resistance after `100+` plug cycles. A `30 mΩ` connector dropping `0.4 V` at `12 A` looks identical to a sagging PSU at the miner end. Replace any connector re-mated more than twice this year. Apply DeoxIT D5 to spring contacts before re-mating.

Verify wall voltage at the HLG input. Mean Well rates HLG for `90-305 V AC` universal input. Below `100 V` the brick begins derating. Brownout from `120 V` to `105 V` during peak residential load reduces input headroom and can amplify a marginal sizing issue into a visible failure. Run a dedicated 15 A circuit if heavy concurrent loads share the same breaker.

Scope the rail during boot. Trigger on the rising edge of the `12 V` line. Healthy HLG soft-start: clean ramp from `0 V` to `12 V` over `300-500 ms`, no overshoot, no oscillation, no collapse. CC-mode HLG under inrush-heavy load: voltage rises to `~10.5-11.4 V` and pins there, no cycling. This scope capture is definitive proof of CC operation when the side switch is hard to read or sealed.

Re-seal the IP67 cover with proper electronics-grade silicone (`Dow 738` or `Permatex 81158`) if the HLG is in outdoor service and you have broken the seal. Apply a thin bead around the cover, press home, allow `24 hours` cure. Do not use bathroom-tile silicone — it contains acetic acid and corrodes copper traces over time.

Substitute a `Mean Well LRS-Hxxx` unit if the build is continuous indoor service. The LRS-series at the same wattage is cheaper, simpler (fixed CV mode on most variants, no IP67), and the right answer for a rack on a shelf. `LRS-350-12` for a Bitaxe Hex, `LRS-200-12` for Gamma + Ultra, `LRS-600-12` for multi-miner. The HLG is overkill indoors and inherits the CC-mode trap for nothing.

Substitute a `Mean Well RSP-1000-12` for serious multi-Bitaxe / Hex / NerdQAxe builds. `RSP-1000-12` (`1000 W × 80 A`) is fixed-CV with proper soft-start and far more headroom than any `HLG-240H`/`-320H` for serious multi-miner work. Drops in with the same wiring conventions, `~$160 CAD`. The HLG can be retired to its intended LED-driver duty.

Document trim-pot positions and switch state in your build log. Record turns from fully-CCW for both `Vo` and `Io`, slide-switch position, no-load rail voltage, full-load rail voltage. This is `30 seconds` of writing that saves hours of re-diagnosis when the brick is reshelved or redeployed.

Stop DIY when slide switch is in `CV`, `Io` pot is at maximum, sizing is correct (continuous ≥ `1.5× peak`), and the rail still sags under load. The HLG is defective. RMA under Mean Well's `7-year` warranty (preserved if you did not break the IP67 seal). For Bitaxe-side faults that masquerade as PSU sag — TPS546 partial failure, BM-series silicon issue, NVS corruption affecting frequency control — book a D-Central diagnostic.

What D-Central does at the bench: test fixture with a programmable load + precision PSU lets us replicate Bitaxe inrush and steady-state behaviour on demand. We characterize TPS546 input current under controlled rail conditions, verify BM1370 / BM1368 / BM1366 silicon health independent of the user's PSU, and produce a written diagnosis with scope captures. From there the path is TPS546 replacement, BM-chip replacement (we stock spares), or a clean bill of health with a PSU-side recommendation.

Ship safely. Bitaxes ship in anti-static bags, double-boxed with `≥ 5 cm` foam on every side. Include the HLG model number, the rail voltage you measured under load, the AxeOS firmware version, and photographs of the slide-switch position. The more context you ship with the unit, the less diagnostic time we have to spend re-discovering what you already know.