Mean Well RSP Simultaneous-Start Inrush AC Breaker Trip

Switch on AC side one PSU at a time. If you have two or more Mean Well RSPs on the same wall outlet or PDU, never gang-switch. Power up one PSU, wait `≥ 5 seconds`, power up the next. Eliminates AC-side inrush stack-up. Cheapest possible fix: zero hardware, operator discipline only. If you can't trust yourself or a household member to remember the sequence, jump to Tier 2 and automate. This is the same discipline Mean Well's own application notes specify for paralleling RSP-3000 units — discrete AC-on per unit, never ganged.

Bring miners up one-at-a-time after PSU is steady. Plug each Bitaxe / NerdQAxe / NerdOctaxe barrel jack or XT30 in sequentially, `~3 seconds` between connections. The steady-state DC bus is a friendlier place to drop a hot load onto than a powering-up bus. Works as a stop-gap while you source a stagger relay; not a long-term answer if the rig power-cycles after a power outage and no one's home to manually re-plug.

Confirm one PSU output stud per miner. Trace cabling from RSP output to each miner. If two or more miners share a single output bar via Y-splitter, the simultaneous-start inrush at that one stud sums and saturates that pin/wire. Fan out to dedicated studs on the breakout board, one cable per miner, no daisy chains. Same discipline as the rail-sag prevention page — DC side wants engineering, not 'good enough'.

Verify PSU is sized at `1.5×` continuous load minimum. If you're running `1500 W` of miners on an RSP-1500, you're at `1.0×` — outside Mean Well's `0.7×` continuous-duty spec, with zero headroom for inrush. Plan to upsize to RSP-2000 or RSP-3000 for the same load. The math is unforgiving: a PSU at `1.0×` continuous duty is rated for short-term cycle, not 24/7 hashing. RSP-1500 = `125 A` continuous on the 12 V rail, RSP-2000 = `167 A`, RSP-3000 = `250 A` — match accordingly.

Check ambient and let the PSU pre-heat on no-load AC for `30 s` before plugging miners. In a cold garage, the NTC inrush limiter starts colder and limits more aggressively for the first few seconds. Powering the PSU AC-side, waiting `~30 s` for the NTC to self-heat down, then plugging in miner DC loads gives you a softer DC ramp because the secondary stage is already in regulation. Eliminates the 'worked all summer, started tripping in November' failure mode.

Install a 12 V automotive timer relay per load with cascading delays. Bosch-style 5-pin SPDT relays are `$5–10 CAD` each on Amazon; pair with a `0.1–10 s` adjustable timer module (`~$8–15 CAD` each). Wire each relay's load contact in series with the miner's DC enable / barrel-jack hot leg. Set timer 1 = `2 s`, timer 2 = `4 s`, timer 3 = `6 s`, all triggered off PSU-good (the RSP's `12V_aux` or `PV` pin works as the trigger source). Relays close in sequence, each miner draws its inrush in its own dedicated time slot, secondary OCP never sees a stacked spike. Canonical fix for this failure mode.

Use a Mean Well AC soft-start module (AS-040D family or equivalent) on the AC side. In-line modules ramp the AC voltage to the PSU input over `200–500 ms` instead of slamming full mains at `t=0`. Collapses AC-side inrush peak to roughly `1/3` of uncontrolled value. Useful when the wall breaker is the trip source (Step 1's 'lights out' scenario). Cost: `~$40–80 CAD` per module. Recommended in addition to the DC-side stagger relay, not instead of.

Upgrade the wall breaker from Type-B to Type-D. Type-B trips at `3–5×` rated current on transients; Type-C at `5–10×`; Type-D at `10–20×`. For a mining outlet feeding multiple Mean Well RSPs, Type-D tolerates real inrush without nuisance-tripping while still protecting against a sustained fault. Cost: `$30–60 CAD` for the breaker, `$100–250 CAD` for an electrician install. Worth every dollar for a rig you cycle on/off frequently.

Move to a 240 V circuit and a single larger PSU instead of 120 V and multiple PSUs. AC inrush peak in `VA` is identical, but `60 A @ 230 V` is the same energy as `120 A @ 115 V` — half the peak current at every point in the curve, including inrush. If you're on 120 V with two RSP-1500s in parallel, consolidate to one RSP-3000 on 240 V — half the connector count, half the breaker trip risk, half the noise floor.

Verify all DC fanout cabling is 14–16 AWG and short. Inrush events stress cable resistance at peak current. A `1 m` cable at 16 AWG drops `~5 mV per amp` resistive — at a `20 A` inrush spike that's `0.1 V` of cable droop, which steals headroom from the miner's input UVLO. Shorter, fatter cables are friendlier to inrush events. Match cable gauge to the inrush spike (peak), not to the steady-state current. For NerdQAxe and Bitaxe Hex XT30 connections, verify connector quality and contact spring tension.

Add a soft-start NTC + bypass relay on the DC fanout to each miner. For each miner load, wire a `5D-9 NTC thermistor` in series with the 12 V hot leg, with a `12 V` relay's NO contact across it. The NTC limits cap-charging inrush to `~3 A peak` for the first `~100 ms`; the relay then closes (driven by a delay timer triggered off PSU-good) and shorts out the NTC for normal steady-state operation. Net effect: per-load soft-start, eliminates simultaneous-inrush even with all miners powered together. BOM: `$5 NTC + $5 relay + $8 timer + connectors` per load. Build time: `30 min per load`.

Replace aged input NTC inrush limiter on the PSU. Open the PSU (warranty considerations apply), locate the input-side NTC (typically `5D-15` or `10D-20` marking near the AC entry), desolder, replace with same-spec part (Ametherm SL15 series, EPCOS B57364S series). New NTC = restored cold-resistance characteristic = restored inrush limiting. Recommended for any RSP that's seen 2+ years of daily power-cycling. **Caution:** input filter caps stay charged at high voltage after AC removal — discharge with `10 kΩ 5 W` resistor for `60 s`, verify `< 10 V` on a DMM before touching.

Re-cap the PSU output stage for tightened transient response. RSP-1500/2000/3000 use bulk electrolytic + ceramic in parallel on the secondary. Replace bulks with same-or-larger capacitance, low-ESR, high-life-grade parts (Panasonic FR / FM, Nichicon PW, Chemi-Con KZE / KZH). Add `100 μF` polymer caps (Panasonic POSCAP or equivalent) directly across the output bus near the terminals to lower the high-frequency impedance — softens the OCP-latch tendency on transients. EE-grade mod, not a beginner job. Ship to D-Central if your hot-air rework isn't there yet.

Build a programmable Arduino / ESP32-based sequencer for 4+ load fanouts. Hand-wired timer relays get unwieldy past 3 loads. An ESP32 with a multi-channel relay shield and 30 lines of code gives you a full programmable startup table — load 1 at `t=2s`, load 2 at `t=4s`, load 3 at `t=6s`, plus telemetry to MQTT or AxeOS API for boot health monitoring. Adafruit, SparkFun, AliExpress all sell modules. Build cost: `$25–50 CAD` for an 8-channel sequencer.

Install an AC-side current-monitoring relay with auto-cycle on overcurrent. A current-sensing relay (CR Magnetics CR4380 family or similar) on the PSU's AC input, set to trip on inrush spike beyond a threshold and auto-reset after a programmable cool-down, prevents repeated AC-side trips from carbonizing the NTC further. Data-centre grade discipline brought home to a residential mining build.

Stop DIY when: multiple failed re-caps, repeat NTC failures within months, breakout-board PCB scorching, wall breaker that won't reset cleanly, or a PSU that hiccups OCP even with a single load on a healthy circuit. At that point you need a programmable bench load with calibrated transient injection and a scope to characterize the inrush curve under controlled conditions. That's bench gear most home miners don't have. Book a slot at https://d-central.tech/services/asic-repair/ for the PSU diagnostic and re-cap, or for a complete rig power-train review.

Ship clean: PSU + suspect cables + breakout board, anti-static bag for any boards, double-box with `≥ 5 cm` of foam on every side. Include observed symptom (which fails first — breaker / PSU LED / miners), miner inventory and steady-state load math, AC line voltage if you've measured it, ambient temperature at failure, photographs of the actual install. Diagnostic time saved at the bench equals bench cost saved equals your repair bill saved.