IceRiver KS3 vs KS3L vs KS3M Error Behavior Compared

Hard power-cycle the unit. Pull the `C19` cord (not the soft-power button), wait 30 seconds for PSU rails to bleed down, plug back in. Watch the dashboard for 5 minutes. Stratum hangs and ARP table glitches that present as zero-hashrate clear here roughly 60% of the time. Applies to KS3 / KS3L / KS3M identically.

Confirm the variant label against the chassis sticker, the box, and the web UI's `Status` model string. Mismatch = open a seller ticket before doing anything else. KS3-class second-hand market sees re-labelling more often than any other IceRiver tier; the three variant names are visually similar and easy to mis-state.

Re-enter pool URL, port, and worker without the `stratum+tcp://` prefix. IceRiver firmware does not accept the prefix. Save, click `Restart`. Same on all three variants but statistically dominant on KS3M because the customer base skews more first-time and copy-pastes from Bitaxe / Antminer guides.

Walk the four-LED block against the shared decoder. `D2` blinking only = healthy. `D1+D3` blinking = fan abnormal. `D1+D2` blinking = network issue. `D2+D3` blinking = high-temperature alarm. LED meanings are identical KS3 / KS3L / KS3M — train this on muscle memory for at-a-glance diagnosis.

Verify ambient air temperature at the intake with an IR thermometer. Target ≤30 °C for KS3 (hottest-running variant), ≤35 °C for KS3L and KS3M (more thermal margin per unit). If ambient is in band, move on; if ambient is high, the variant doesn't matter — clean filters and improve airflow first before anything else.

Measure wall-power against variant nameplate using a Kill A Watt or equivalent AC meter rated for 12 A continuous. Steady-state at 15 minutes. KS3 healthy band: 2880 W-3520 W. KS3L healthy band: 2880 W-3520 W. KS3M healthy band: 3060 W-3740 W. Out-of-band low on KS3 / KS3M = a hashboard or chain is dark; out-of-band low on KS3L is probably designed-in. Out-of-band high = thermal trip imminent.

Re-seat all three hashboard data and power cables. Power off at the breaker. Inspect connectors for blackening, oxidation, bent pins. Reconnect firmly. Listen for the click. Same procedure across KS3 / KS3L / KS3M — the physical hashboard interconnect is identical.

Swap hashboards between slots to localize a fault. Label the slots, move the suspected-bad board to a known-good slot, observe. Fault follows the board = bad board. Fault stays in slot = bad control path or cable. Same procedure all three variants. Note: cross-variant hashboard swaps are not a viable repair strategy because firmware expects a specific chip-count-per-board for each variant.

Inspect the `C19` cord and PSU `C20` inlet under strong light. Browning, deformation, pin tip discoloration = condemned. Replace with a 16 A-rated, 14 AWG minimum heavy-gauge industrial cord (Tripp Lite / StarTech / Cyberpower). KS3 is the highest-stress variant for cord melt (highest sustained AC current); KS3M close behind because of higher absolute wattage; KS3L lowest-stress because lighter silicon load even on the same AC envelope.

Verify line voltage at the panel under load. 235 V-245 V expected on 240 V split-phase, 202 V-212 V on 208 V commercial. Low line voltage = PSU sag = elevated PSU temperature codes (`233`-`239`). Same threshold all three variants. KS3 and KS3M show PSU sag faster than KS3L because their absolute wattage is higher and the PSU is closer to its thermal envelope.

Cross-reference against the chip count mismatch reference if `Status` log shows 9, 26, or 52 chips on a board. The detection-tier numbers (9, 26, 52) are how IceRiver firmware reports a partial chain detection where some chips along the chain have failed and broken the daisy-chain. Pattern is identical KS3 / KS3L / KS3M. Repair is the same: identify which chip in the chain is dead, replace.

Replace failed ASIC chips using the chip family Zeus catalogs as P38C03 2342, P37P57 2342, P38C39 2339, P3AS24 2345, P3GY31 2347. Random delivery is the norm — Zeus and other parts suppliers ship whatever's in stock from the family, and IceRiver firmware accepts cross-family chips on the same board. Reflow workflow: preheat ~150 °C bottom side + hot-air ~310-330 °C top side for ~30 s. Same procedure all three variants.

Refresh thermal paste and pads on all chips. Arctic MX-6 or Thermal Grizzly Kryonaut. Uniform thin layer, don't glop. KS3 = most critical variant (chips run hottest). KS3M = second priority. KS3L = least critical (lightest silicon load). Refresh interval: 12 months for KS3, 15-18 months for KS3M, 18-24 months for KS3L.

Inspect the PSU module for inlet damage, capacitor bulging, or burnt-component odor. KS3 and KS3M PSUs work harder than KS3L's; their failure timeline is shorter on average. Replacement modules are sourceable through D-Central and Zeus parts. VERIFY exact PSU part number against current IceRiver spec before ordering, especially for KS3M which sometimes ships with a different PSU revision than KS3.

Migrate to a 240 V dedicated circuit if the unit is running on 120 V. KS3 and KS3M on 120 V pull >26 A peak — over the breaker rating of any standard residential 15 A-20 A circuit. KS3L on 120 V pulls slightly less but is still over a 15 A circuit. All three variants effectively require 240 V for reliable operation; the 120 V install is a leading cause of cord melt, breaker trip, and PSU stress codes.

When to ship to D-Central: more than two chips dead on the same hashboard; PSU module showing physical damage; chassis ground continuity failure (>5 Ω); multiple cord/inlet melt events on the same install (root cause is upstream of cord); or any electrical event you're not certain you fully diagnosed. Past those points the failure mode is upstream of where DIY can isolate. Same threshold across all three KS3 variants.

D-Central bench process: test fixture with programmable load, per-board isolation, chip replacement with graded P38C03 / P37P57 / P38C39 / P3AS24 / P3GY31 parts (or salvaged from condemned KS3-class boards), full reflow and reseal, post-repair 24-hour burn-in at the variant's nameplate hashrate, signed-off documentation of the AC path's safety state. Western retail repair authority for IceRiver — Zeus (China) is the only non-Western alternative.

Ship safely. Pack the entire input cord with the unit (we need to inspect what failed). Anti-static bag any removed PSU module. Double-box with 5 cm of foam every side. Include a written description: confirmed variant (KS3 / KS3L / KS3M), serial, install voltage, run length to the panel, observed symptoms, firmware version, and a photo of any burn pattern. The narrative saves bench time which saves repair cost.