Skip to content

We're upgrading our operations to serve you better. Orders ship as usual from Laval, QC. Questions? Contact us

Bitcoin accepted at checkout  |  Ships from Laval, QC, Canada  |  Expert support since 2016

PSU_ERR Critical

Antminer L7 – Power Cable Overheating

Power Cable Overheating — C19/C20 cord or APW12-1417 PSU inlet runs above ambient+25 °C under L7 hash load, with discoloration, scorch, or plastic-softening damage visible at the connector contact surface.

Critical — Immediate action required

Affected Models: Antminer L7, L7 9.3 Gh, L7 9.5 Gh, L7 9.16 Gh, L7 8.8 Gh, L7 8.5 Gh, L7 8.3 Gh (all SKUs with APW12-1417 PSU)

Symptoms

  • Cord body uncomfortably hot to grip (>55 °C) 10-20 cm from either connector under full hash
  • Visible discoloration, blistering, or scorch rings on the C19 shell at the PSU end
  • Brown or black marks around the C20 pin holes at the APW12-1417 PSU inlet
  • Plastic or 'burning electronics' smell near the miner during steady-state mining
  • Intermittent shutdowns with `ERROR_POWER_LOST` or `get power type version failed` in `kern.log`
  • Breaker trips on load but holds when the L7 is idle — classic loose-contact thermal signature
  • Inlet voltage at the miner sags 5-12 V below panel reading when the L7 ramps to full hash
  • PDU outlet face deformed or locking tab feels loose and wiggly
  • GFCI / AFCI breaker has tripped once already without an obvious cause
  • Cord in use is a generic 10 A kettle lead, a C13 mislabeled as C19, or a 16 A S9-era cord running an L7
  • Cord jacket feels soft or over-flexible at the connector — the PVC has lost its set
  • IR-thermometer reads ambient + 25 °C or more at any contact in the power chain after 30 min hashing

Step-by-Step Fix

1

Kill power at the breaker immediately — not the miner switch, not by tugging the plug. A hot cord is an active fire hazard and re-plugging a live hot contact pits it further. Let the cord sit 15 minutes before handling. Photograph every damaged surface before you touch anything: the images are your insurance evidence and your D-Central diagnostic packet if you end up shipping hardware in. Never re-energize a miner whose cord has overheated without completing at least the Tier 2 inspection, regardless of how mild the damage looks on the outside.

2

Inspect the full power chain, not just the cord. With the breaker off, walk from panel breaker to wire run to receptacle or PDU outlet to cord to PSU inlet. Look for any discoloration, smell, deformation, or loose screw. The cord is where the symptom shows; the cause is usually one connection upstream. Document every connection with phone photos, mark anything suspect with painter's tape so you don't accidentally re-install a compromised component, and note the state of every locking tab.

3

Read the cord label. If the cord says `10 A`, `13 A`, `125 V`, `14 AWG`, `H05VV-F 1.0 mm²`, or anything other than `16 A` minimum / `12 AWG` / `105 °C`, you have found your problem. Any S9-era cord used on an L7 is almost certainly undersized. Do not re-plug this cord — bag it, tag it, put it in the trash, and source a correct cord before re-energizing. Saving CAD $50 on a cord replacement is negative expected value on a $2500-$4000 L7.

4

Verify circuit ampacity at the panel. An L7 at 220-240 V wants its own dedicated 20 A or 30 A circuit — nothing else on the line. Shared with lighting, a space heater, or a second miner, cumulative current can push either the cord or the breaker past continuous-duty rating. Read the breaker amperage stamp. A 15 A breaker feeding an L7 is marginal with a perfect cord. Canadian 240 V split-phase: licensed-electrician L6-20 or L6-30 install, on its own breaker, with 12 AWG (20 A) or 10 AWG (30 A) conductor.

5

Order the correct cord before re-energizing. 16 A C19-to-plug assembly (20 A preferred for headroom), 12 AWG copper, 105 °C jacket, with locking tab if your PDU has a matching latch. Do not improvise. Do not buy the CAD $12 Amazon special; a cord rated for continuous 16-20 A at 240 V with 105 °C jacket costs CAD $35-80 legitimately and is cheap insurance. Bitmain OEM cords are fine, as are compliant assemblies from StayOnline, Tripp Lite, or equivalent.

6

Measure cold contact resistance at the receptacle. With the breaker off and the miner unplugged, measure pin-to-pin resistance on the PDU socket or wall receptacle with a test cord inserted. A healthy socket reads under 50 mΩ. Anything above 100 mΩ is scrap — the contact surface is oxidized or pitted and will run hot under load. Measure the APW12-1417 C20 inlet the same way. Pitted contacts on the PSU inlet mean the PSU itself needs service or replacement, not just the cord.

7

Torque-check every screw terminal in the power chain. Breaker lug, panel neutral bar, receptacle back-wire screws, PDU terminals. Use a torque screwdriver to manufacturer spec — typically 1.6-2.3 Nm for 12 AWG residential lugs (check the breaker's own torque sticker). Any screw that turned more than a quarter-revolution at spec was loose. Loose = hot = eventually melted. Do not guess torque by feel; copper creep under thermal cycling is real and screws that felt tight at install are almost certainly looser six months later under continuous L7 load.

8

Thermal-image the full run under load. After replacing the cord and verifying connections, power the miner up and run at full hash for 30 minutes. IR-thermometer or thermal camera on the breaker body, both sides of the receptacle, both connectors of the cord, and the APW12-1417 inlet. Ambient+15 °C is acceptable. Ambient+25 °C or higher at any contact is a deferred failure — find the bad connection before it melts again. Thermal audits routinely catch loose PDU contacts that cold-resistance testing missed.

9

Load-test the new cord at rated current for 4 hours. Before leaving a newly-installed cord unattended overnight or on a production rig, run the L7 for 4 hours and recheck every contact temperature once per hour. Cords that will fail early usually show thermal drift inside the first 2 hours. A cord stable at hour 4 will almost always stay stable for its design life. This is the single cheapest diagnostic you can do and the one most operators skip.

10

Upgrade the PDU if in doubt. A CAD $60 generic IEC PDU rated 10-13 A per outlet used on a $3000 L7 is terrible economics. Source a 20 A or 30 A metered rack PDU (APC AP7921B, Tripp Lite PDUMV30HV, or equivalent) — the metering catches drift before it becomes damage, and the 20 A+ rating gives the C19 contacts actual current headroom on a 16 A cord running near its limit.

11

Swap in a known-good cord from a working miner as a control test. If you have a second miner on the same rack with a known-good cord, swap cords and see whether the overheating follows the cord or stays with the miner. Fault follows the cord = the cord was bad. Fault stays with the miner = the PSU inlet or PDU outlet is the culprit, not the cord itself. This takes five minutes and saves hours of wrong-direction diagnostic.

12

APW12-1417 inlet replacement on the PSU. If Step 6 found a pitted or damaged C20 inlet, the receptacle module inside the PSU is field-serviceable by a qualified tech. Discharge all primary-side capacitors with a 10 kΩ / 10 W bleed resistor before opening the PSU case; APW12-family PSUs hold mains-voltage charge for 30+ seconds after unplug and have killed people who skipped this step. Desolder the inlet module, clean the PCB, install the replacement, verify ground continuity with a milliohm meter (< 30 mΩ earth path) before closing the case. If any of that made you uncomfortable, ship the PSU to D-Central instead.

13

Hard-wire a dedicated whip. For serious home setups or small farms, eliminate the PDU-and-cord chain entirely with a 12 AWG THHN/THWN whip in flexible metallic conduit, straight from a dedicated 20 A or 30 A breaker to the miner — no detachable cord. Terminate at the PSU end with a genuine C19 inlet pigtail, heat-shrinked and strain-relieved. Eliminates every connector-based failure mode at the cost of a fixed install. Canadian residential: permit and licensed-electrician sign-off required by CEC. Quebec requires a master electrician's seal on any 240 V install.

14

Audit the breaker and panel. Pull the panel cover (power off, please), thermal-image the breaker body after a 30-minute load, reconnect, torque the lug to spec (typically 2.8 Nm for 20 A breakers — check your specific breaker's sticker). Aged breakers — 15+ years — can lose trip calibration. Replace any breaker that runs warm to the touch under nameplate load, even if it isn't tripping. On 240 V split-phase verify phase balance — single-pole feeds on a dual-phase panel cause neutral current that loads the neutral conductor above its rating.

15

Flash DCENT_OS for real-time power telemetry. DCENT_OS — D-Central's own open-source Antminer firmware, with the full per-chip HW%, tuning, autotuning, and stratum v2 feature set of Braiins OS+ / LuxOS / Vnish, open-source and maintained publicly — surfaces power-draw telemetry, rail voltages, and PSU temperature the way stock Bitmain firmware does not. Pair DCENT_OS with a smart PDU for minute-by-minute resolution on whether the cord-and-connector chain is drifting resistive. Alternatives: Braiins OS+, LuxOS, Vnish.

16

Derate the L7 if the circuit is permanently marginal. If you're stuck on a 15 A 120 V residential circuit and can't run a new 240 V line yet, drop the L7 power profile in DCENT_OS (or Braiins OS+ / LuxOS) to 75-80% until the electrical is sorted. You'll lose 700-900 W of hashrate, but you'll stop running the cord at 200% of its rated current. A derated L7 at 75% still hashes at 6.9-7.1 Gh/s, which beats a melting cord every time. Schedule the 240 V install, then un-derate.

17

Ship the PSU in if the inlet or primary-side is damaged. APW12-1417 inlet-module replacement, primary-side capacitor inspection, input-fuse replacement, MOV surge-protection sanity check, and full dielectric + ground-continuity validation is bench work. D-Central stocks APW12 inlet modules, fuses, MOV surge components, and compliant cords. Book an ASIC Repair slot at d-central.tech/services/asic-repair. Canada-wide shipping, 5-10 business-day turnaround, US and international welcomed.

18

Ship the miner in if the internal harness overheated. Once heat reaches the miner side of the C20 inlet, it can propagate into the internal DC harness and the control-board power rail. Control-board-side damage needs bench diagnostic — rail voltages, continuity, sometimes connector replacement on the PCH side — and the BM1489 hashboard chips are sensitive to rail excursions during the fault window. Field-repair is not reliable here; the miner will either cascade-fail within weeks or hide latent damage that shortens its life by years.

19

Professional electrical audit on the circuit itself. If the diagnostic trail leads back to the breaker, panel, or utility feed, you are outside any DIY envelope. Licensed electrician, full load audit, panel inspection. D-Central can refer electricians in Quebec, Ontario, and Alberta with mining-farm experience; contact the repair desk for a referral if you're Canadian. This is not optional cost — home-mining fires are not insurable in most Canadian provinces if the install wasn't permitted and inspected.

20

Pack and ship safely. APW12-1417 PSU packed in anti-static bag, double-boxed with at least 5 cm of foam on every side. Include a note with observed symptoms, firmware version in use, cord label photo, and contact info — saves diagnostic time at the bench, which saves you money on the quote. For a full L7, pack the hashboards separately in anti-static bags; the stock L7 chassis vibration isolation is not enough for freight handling.

When to Seek Professional Repair

If the steps above do not resolve the issue, or if you are not comfortable performing these repairs yourself, professional service is recommended. Attempting advanced repairs without proper equipment can cause further damage.

Related Error Codes

Still Having Issues?

Our team of Bitcoin Mining Hackers has been repairing ASIC miners since 2016. We have seen it all and fixed it all. Get a professional diagnosis.