IceRiver PSU Error 233-239 Overtemperature Protection

Hard power-cycle: switch off at the wall outlet for 60 seconds, then power back up. Some 233-239 events are transient - a brief ambient spike, a momentary fan stall, a sensor noise burst - and a clean cold boot resets the warning state. If the code clears and stays clear after 60 minutes of stock-load hashing, no further action needed. If it returns, escalate to step 2.

Drop to Normal Mode if currently in Performance Mode. Normal Mode runs the silicon at lower frequency / voltage, drops dissipation by 15-25%, and pulls the rig out of the danger band almost immediately. Run Normal Mode for 60 minutes, observe. If the code stays clear, you have a Performance-Mode-specific cooling envelope problem - fix airflow before re-enabling Performance Mode.

Move ambient air aggressively. Open windows, run a box fan blowing across the rig's intake, drop the room temperature however you can. Even a 2-3 C drop in ambient often clears 233-239 because the rig was sitting right at the edge. This is a free 10-minute test that tells you whether the cause is ambient or systemic.

Vacuum the intake filter and the rear exhaust grille. Power off at the wall. Use a shop-vac or compressed air. Wipe the intake grille with a clean microfiber. Verify nothing blocks the front 15 cm or rear 15 cm of the chassis. Re-power, observe 30 minutes. Filter loading is responsible for the documented majority of 233-239 fires on rigs older than 6 months - this is the highest-leverage Tier 1 step.

Verify ambient with an IR thermometer or thermometer at the intake grille. Not room-middle, not hallway, not 'feels cool to me'. Measure at the front grille of the miner with the rig running. Target: <=35 C Normal Mode, <=30 C Performance Mode. If you're above target, you've found your cause - fix ambient before any other diagnostic effort.

Replace the suspect fan. Power off at the wall. Open the chassis (typically four corner screws on KS-class). Identify the fan reading below nameplate, or refresh all fans preventively past 18 months. KS5L spec: 12038 form factor, 6000 RPM, DF1203812B2UN or compatible. Disconnect the fan power lead, unscrew, swap, reconnect, secure cabling so it doesn't get sucked into the new fan. Re-power, verify the new fan spins to nameplate via the web UI.

Re-seat the PSU AC inlet and the PSU-to-hashboard power cables. Power off at the wall, unplug from outlet. Inspect the AC inlet on the PSU for any sign of melting, scorching, or loose fit. Disconnect the PSU output cables to the hashboards, inspect contacts for blackening or oxidation, reconnect firmly. Loose power-side connections produce voltage drop, which produces heat at the PSU side - direct cause of 233/234.

Add active forced airflow on the PSU intake (KS5L / KS5M specifically). If you've confirmed codes are clustered in 233/234 (PSU thermal), aim a small 120 mm desk fan or clip-on fan at the PSU's intake side. Documented community workaround for the BP-H-3640's low thermal headroom. Not pretty, but it works - and it's a $15 fan vs a $300+ PSU replacement.

Verify line voltage at the outlet under load. With the rig hashing, multimeter the AC outlet. Expect 108-125 V on US/Canadian 120 V circuits or 220-240 V on 220 V circuits. KS5L published AC range is 180-285 V. Sustained operation near the bottom of that range increases PSU current draw, increases PSU heat, and pushes into the 233/234 warning zone. Time-of-day-correlated warnings often track line voltage sag during peak hours.

Roll back to a known-good firmware version if warnings started immediately after a recent firmware update. Pull the prior version's image from your records or contact IceRiver support via iceriver.io/faq - prior versions aren't always public but support keeps them on file. Apply via web UI or SD-card recovery. Some IceRiver firmware revisions ship thermal-control regressions that fire spurious 233-239 codes.

Thermal paste refresh on all hashboards. Power off, fully disassemble the suspect hashboard from the chassis. Remove heatsink screws, separate heatsink from PCB carefully (gentle even pressure, no prying). Clean chip die and heatsink mating surface with 99% IPA and lint-free wipes. Apply Arctic MX-6 or Thermal Grizzly Kryonaut in a uniform thin layer. Re-mount heatsink with even cross-pattern torque. Reassemble, run rig, watch Temp1/Temp2 settle. Schedule annually thereafter.

Replace the BP-H-3640-class PSU on KS5L / KS5M when 233/234 keeps firing after Tier 1 + 2. The PSU has aged out of its already-marginal thermal envelope. Source the replacement through D-Central (we're building IceRiver parts inventory) or Zeus Mining's parts catalog. Mains-side work - confirm comfort with AC wiring before proceeding, or escalate to Tier 4. Power off, unplug, swap PSU, reconnect output rails, mount, restart.

Reflow voltage-domain regulator on persistent 238 codes. Pull the suspect hashboard. Apply flux to the PMIC / LDO pads and the surrounding decoupling caps. Reflow with hot-air station: preheat PCB to ~120 C from below, then top-side hot air at 300-320 C for ~30 seconds. Let cool naturally, clean residue with 99% IPA. Cracked solder joints under regulator packages after thousands of thermal cycles are a documented 238 cause; reflow restores them.

Inspect for damaged thermal pads on regulators / domain ICs. Adjacent to the main ASIC heatsink, KS-class hashboards have smaller thermal pads bridging voltage-domain ICs to the heatsink (or to a secondary heatsink on KS5L/KS5M). These pads dry out faster than the main paste - 12-18 months is realistic. Replace with high-quality 1.5 mm or 2.0 mm thermal pad - match thickness to the original (a 1.5 mm pad in a 2.0 mm gap is worse than nothing).

Audit any third-party OC firmware (xyys / tswift via rdugan/iceriver-oc). If running OC firmware on KS0-class or KS5L/KS5M, revert to stock IceRiver firmware as a diagnostic. Third-party OC pushes silicon past where the rig's cooling and voltage regulators were designed to live. If stock firmware clears the warnings, the rig isn't safely OC-able in your environment - run stock, period. Don't argue with the silicon.

Stop DIY and ship to D-Central when: Tier 1-3 exhausted and 233-239 keeps firing; PSU shows visible damage (scorch, melt, swelling caps - mains-side hazard); 238 keeps firing after reflow attempt; capacitor bulging, discolored components, or burnt-component smell anywhere; or two or more KS rigs in the same room fire the same warnings (room-level airflow audit needed). Book a D-Central ASIC repair slot at d-central.tech/services/asic-repair/.

D-Central bench process: test fixture probes thermal sensors directly to confirm sensor health independent of the controller, full thermal-paste refresh on all hashboards, regulator reflow / replacement as needed, PSU diagnostic and replacement (BP-H-3640 or model-specific) using bench-tested replacement units, post-repair 24-hour burn-in at nameplate with continuous thermal logging to verify the rig holds healthy Temp1/Temp2 under sustained load. D-Central is positioning as the Western English-language IceRiver repair authority.

Ship safely. Pack the full miner if possible - KS chassis are compact and thermal diagnostics often need the assembled unit. Anti-static bags around exposed boards, double-box with >=5 cm foam every side. Include a printed note with: model + serial, exact firmware version, every error code observed in the 233-239 range with timestamps, what you've already tried (cite the step numbers from this page), room ambient at the time codes fired, and your contact. The note saves diagnostic time, which saves you money.