IceRiver KS5/KS5L Fan Not Spinning on Startup

POWER OFF AT THE REAR ROCKER NOW. This is not a power-cycle-and-see-what-happens situation. The KS5 / KS5L / KS5M run hot enough that one stuck fan plus continued mining equals burnt `1004LV100` chips in 60-180 seconds. Drop the rocker, pull the IEC cord, walk back to your laptop. Wait 60 seconds for bulk capacitors on the controller and PSU to bleed before opening the chassis.

Screenshot the IceRiver web UI before powering off if you can do it in 30 seconds: exact error code (`120` vs `121` vs both), `Temp1`, `Temp2`, per-board hashrate, firmware version, and front-panel LED status. If you escalate to D-Central, that screenshot saves diagnostic time and money. If the miner is already too hot to leave running while you screenshot - skip it and power off.

Open the top cover. Six Phillips #2 screws on a standard KS5; the KS5L and KS5M add two on the rear bracket. Set screws in a labeled cup. With cover off you'll see two `12038` axial fans bolted with four `M4` machine screws each, connected to the controller via 4-pin pigtails. Identify which fan corresponds to the error code - front for `120`, rear for `121`.

Manual impeller test (fan unpowered). With the miner still off, try to spin the suspect fan's impeller with one finger, light flick. A healthy `12038` rotates 5-15 revolutions before friction stops it. Bearing failure: locks in under one revolution, catches mid-rotation, or refuses to start. If it spins freely it's probably a wire or controller fault, not the fan - skip to Step 7 swap test. If stuck or catches, continue to Step 5.

Visual + foreign-object inspection. Pull the suspect fan out of its mount (four `M4` screws). Inspect both sides of the impeller for hair, lint, debris wrapped around the hub, loose zip-tie ends, dropped screws from prior service, mouse-nest debris, or anything obviously not supposed to be there. Look at blades for cracks, missing chunks, stress fractures. Compressed air both sides of the bearing hub; carefully tease out debris with tweezers. Try the finger-spin again.

Reseat the fan connector at the controller with dielectric grease. Locate the 4-pin headers (silkscreened typically `FAN_F` for front and `FAN_R` for rear). Press the connector latch and pull straight up on the connector body, never the wires. Apply a trace of dielectric grease to the pin contacts - vibration immunity for the next 12-24 months. Reseat firmly, listen for the click. Smoke-test with cover off: power back on, watch the suspect fan during the ~30-second boot self-test - it should ramp from zero to ~6,000 RPM smoothly.

Front-rear swap test - the most decisive zero-cost diagnostic in the IceRiver fan-fault tree. Power off, unplug both fans at the controller, swap their headers (front fan to rear header, rear fan to front header). Power on, wait 2 minutes, watch the dashboard. If error code follows the fan (`120` becomes `121`), the fan is dead - replace. If the error stays on the same header regardless of which fan is plugged in, controller-side fault - new fan won't fix it, drop to Tier 3 / 4.

Probe the fan rail with a multimeter. Plug the suspect fan back into the controller, IEC cable in, leave cover off, stay clear of fan blades. Multimeter on DC volts, probe `V+` to `GND` on the suspect fan's connector while the miner is commanding the fan during boot self-test. PWM pinout: `BLACK = GND`, `RED = +12 V`, `YELLOW = TACH`, `BLUE = PWM` - verify against fan data sheet, not assumption. Healthy: 11.8-12.2 V steady. Dead: below 11 V, zero, or pulsing erratically = control-board rail fault.

Order the right replacement `12038` fan(s). Min spec: `120 x 120 x 38 mm`, `12 V` DC, 4-pin PWM, dual ball bearing, RPM `>= 3,000` (target ~6,000), CFM `>= 140`, dB(A) < 60. Recommended: Delta `AFC1212DE` (~6,000 RPM, ~250 CFM, premium overspec), Sanyo Denki `9G1212H4D03` (San Ace 120 series, premium dual ball), Nidec UltraFlo (verify 38 mm thickness SKU), or OEM-equivalent `DF1203812B2UN` (~6,000 RPM, ~140 CFM). Order in pairs - the surviving fan ate the same dust and thermal cycles.

Power off, unplug IEC cable, wait 60 seconds for capacitors to bleed. Disconnect the suspect fan's pigtail at the controller (press latch, pull connector body straight up - never yank wires). Unscrew the fan from chassis: four `M4` Phillips screws (Torx T10 on some KS5L revisions), two often captive in rubber vibration grommets. Slide fan out, route pigtail through chassis grommet without snagging.

Mount the new fan and verify airflow direction. Same `M4` screws, same orientation as original. Critical: every axial fan has an airflow-direction arrow on the side of the frame. The front-intake fan must blow air INTO the chassis; the rear-exhaust fan must blow air OUT. Installing backwards halves your effective airflow and pushes hot exhaust back across hashboards - `Temp2` will climb instead of fall. Reuse rubber vibration grommets to cut audible noise 2-4 dB(A) and reduce bearing wear.

Connect the new fan's pigtail to the controller. If the new fan came with a matching 4-pin connector, plug it straight in, listen for the click. If bare wires, crimp or solder a matching 4-pin housing. Standard 4-pin PWM pinout: `BLACK = GND`, `RED = +12 V`, `YELLOW = TACH`, `BLUE = PWM`. VERIFY against the new fan's data sheet - industrial fans sometimes use yellow/green/white. Wrong polarity on `+12 V` and `GND` instantly destroys the new fan's driver IC and may pop the IceRiver controller's fan-rail fuse.

Smoke-test before closing the chassis. Cover off, IEC cable in, hit the rocker. Watch the new fan during the ~30-second boot self-test. New fan should ramp from zero to ~6,000 RPM smoothly, no vibration, no rattle, no audible bearing growl. Dashboard should clear `Error 120` or `121` within 60 seconds of boot completing. If the dashboard still throws the error after the swap, you didn't fix a fan fault - back to Step 7 swap test before closing up.

Run at full hashrate for 15 minutes with cover still off. Watch dashboard `Temp1` and `Temp2` trend lines. Both should stabilize in spec (typically 30-60 C depending on ambient and per-board load). New fan RPM should track thermal demand cleanly. If `Temp2` is climbing instead of stable, the new fan is installed backwards - power off, flip it, retry. This 15-minute burn-in is the verification window most operators skip and the one that catches subtle problems (intermittent TACH wire, partly-seated connector, fan installed backwards).

Reinstall the cover, re-torque all chassis screws (Phillips #2 or Torx where applicable), snug not gorilla-tight. Reinstall rear-bracket screws on KS5L / KS5M. Verify nothing is pinched between cover and chassis - fan pigtails, harnesses, ribbon cables. Run a 24-hour burn-in. If new symptoms appear within 24 hours (rattle that wasn't there day-one, recurring `Error 120/121`, asymmetric `Temp1/Temp2`), you have either a DOA replacement fan or a controller-side issue the fan was hiding.

Tier 3 controller-side rework if Step 7 isolated a controller fault. Tools: low-magnification stereo microscope or 10x loupe, multimeter with continuity beep, fine-tip soldering iron at 350 C, replacement SMD fuses. Visually inspect the IceRiver controller around the suspect fan rail for: blackened or popped SMD fuses on `+12 V` rail (typically `F1`/`F2` near the fan headers, varies per KS5 revision), discolored or blistered solder mask, cracked MLCCs near the PWM driver MOSFET, SOT-23 / SOT-89 packages with thermal damage.

Replace blown SMD fuse if found. KS5 fan rails typically use `2A` or `3A` SMT0805 fast-blow. Match the rating EXACTLY - wrong rating either pops again immediately or fails to protect on the next stall event. Hot-air rework or fine-tip iron at 350 C, flux generously, swap the part, verify continuity at <1 ohm between pads. If rail still dead after fuse, the PWM driver MOSFET may be dead - bench territory, drop to Tier 4.

Stop DIY and book a D-Central ASIC Repair slot when any of these are true: (a) new fan installed correctly with verified polarity, connector seating, and airflow direction - and `Error 120/121` returns within 24 hours, (b) both fan rails read dead at the controller, (c) visible heat damage / blistered solder mask / burnt-component odor on the controller, (d) replacement fan exhibits the same fault as the old fan within first boot or first hour (controller PWM driver glitching or DOA from a bad batch), (e) second fan failure within 90 days on the same miner, (f) fan failure cascaded with `Temp2` above 80 C for hours and silicon damage is possible. Book at https://d-central.tech/services/asic-repair/ - D-Central is the Western English-language IceRiver bench authority, Quebec-based, Canada / US / international shipping welcomed, typical turnaround 5-10 business days.