IceRiver Temp1/Temp2 Readings Explained (Optimal 50-60C)

Hard reboot from the rear rocker, not the web UI reboot button. Some IceRiver firmware versions latch a corrupt sensor reading on a soft reboot; a hard rocker reboot clears the I2C state machine on the controller and re-initializes the sensors at boot. Power off at the rocker, wait 60 seconds for caps to bleed, power back on. Wait 90 seconds for full boot, log into the web UI, re-read `Temp1` and `Temp2`. About one in ten spurious high-temp readings clear here and don't return.

Check ambient at the intake grille with an IR thermometer. Not the room middle, not the hallway, not your forehead. Five centimeters in front of the front grille, where the miner is actually pulling air from. This is the number that matters versus IceRiver's published mode ceilings (`<= 35 C` Normal Mode, `<= 30 C` Performance Mode). If you're above the ceiling, fix the ambient first - drop the room temp, duct the exhaust, relocate, or swap to Normal Mode if you were on Performance Mode. The miner will pull `Temp2` back into the green zone the moment ambient is in spec.

Verify `15 cm` clearance front and rear. IceRiver KS-class miners need at least 15 cm clearance front and rear for unobstructed airflow. Curtains, shelves, walls, other miners exhausting into your intake, server racks without fan trays - all of these elevate `Temp2` by reducing CFM through the heatsinks. Move whatever's in the way and re-measure after the miner has 30 minutes to re-equilibrate. This step alone resolves a non-trivial fraction of perceived high-temp situations.

Cross-check against the thermal table. `Temp1` steady-state load: `28-45 C` green, `45-55 C` yellow, `55-65 C` red. `Temp2` steady-state load: `50-60 C` green (this is normal and optimal), `60-75 C` yellow, `75-85 C` red, above `85-90 C` is the firmware hard-trip threshold (codes `350` / `351` / `352` fire). `Temp2 - Temp1` delta: `15-25 C` healthy. If your numbers are in the green zone, your miner is fine - print this table, tape it next to the miner, and stop worrying. If your numbers are yellow or red, continue to Tier 2.

Watch the trend over hours and days. Healthy miners hold steady-state temps for hours. A `Temp2` reading that creeps up `1-2 C` over a few hours is normal (room heating up from miner exhaust, HVAC schedule, sun on the room). A `Temp2` reading that climbs `5+ C` over a few hours and never settles is a flag - usually airflow blockage, paste pump-out starting, or fan RPM degrading. Log the steady-state reading once a week and watch for month-over-month creep - that's how you catch dust loading and paste failure before the firmware throws an actual trip.

Open the chassis and inspect the heatsink fins. Power off, pop the corner screws (Phillips #2 standard, Torx T10 on some KS5L revisions). Look at the heatsink fins on each hashboard. Dust mat, pet hair, drywall film, or a felt-textured layer between the fins is the most common cause of `Temp2` creep over time. Shop-vac the fins gently, hold the impellers stationary with a plastic probe while doing it, blow with compressed air or canned duster to clear deeper dust the vacuum missed. Wipe the front intake grille. Reassemble. Run for 30 minutes and re-read `Temp2`. A dusty heatsink can elevate steady-state `Temp2` by `8-15 C`.

Verify both 12038 fans are at nameplate RPM. Web UI Status page shows fan RPM. KS-class fans (`12038 6000 RPM` axial, typically `DF1203812B2UN` on retail units) should hit `5500-6500 RPM` under full load. If you see one fan at `4000 RPM` or below while the controller is commanding `100%` PWM, the fan is dying. Replace per the 12038 fan DIY replacement guide. A dying fan is the second-most-common cause - after dust - of `Temp2` climbing over months without an actual error code firing.

Check the exhaust path. If your miner exhausts into a closed cabinet, an unventilated closet, or directly back into its own intake (worst case), `Temp2` will climb because the miner is breathing its own waste heat. Re-route the exhaust. Even a cheap dryer-vent style duct out a window cuts `Temp2` by `5-10 C` in a marginal install. In residential setups, dual-purpose mining setups (using miner exhaust as supplemental heat) work great in winter but require a damper or seasonal duct switch in summer.

Refresh thermal paste on the hashboards. If the miner is `> 12 months` old, has been cleaned and fanned and the readings are still elevated, paste pump-out is the next layer. Power off, remove the offending hashboard, lift the heatsink, clean old paste with isopropyl `99%` and a lint-free wipe, apply a thin uniform layer of Arctic MX-6 or Thermal Grizzly Kryonaut on the chip dies, refit the heatsink with the original mounting hardware (don't overtorque - clips are spec'd for a specific pressure). Reassemble, run 30 minutes, re-read. A successful paste refresh typically drops `Temp2` by `5-12 C` on a 12-18 month old miner.

Replace thermal pads on PMIC / voltage-domain ICs. Same procedure as paste, different parts. Each KS-class hashboard has thermal pads on the secondary heat-generating components (PMIC, voltage regulators, VRMs). These compress and lose effectiveness over `18-24 months`. Replace with a `1.5 mm` pad of equivalent thermal conductivity (the original spec is in IceRiver's parts documentation; cross-reference with Zeus Mining's KS-series repair notes if you can't find OEM). Reassemble, run, re-read. Pad refresh on its own usually moves `Temp2` by `2-5 C`.

Consider a fan upgrade. If you've cleaned, fanned, pasted, and padded and `Temp2` is still elevated, the original `12038 6000 RPM` fans may be near end-of-life across the board. A full set of new OEM-equivalent `12038`s (typically `4-6` fans depending on KS variant) brings the cooling stack back to factory spec. Don't substitute consumer PC fans - wrong static-pressure curve. Verify-flag: confirm exact part number against your specific KS-series revision before ordering.

Roll firmware back if sensor readings look glitched after a recent update. If `Temp1` reads `-10 C`, `Temp2` reads `255 C`, or both sensors lock at the same value mid-load, you have a firmware regression rather than a hardware fault. Pull the prior image from https://www.iceriver.io/firmware-download/, match the variant exactly to your model (KS3M / KS3L / KS5L / KS5M - wrong variant bricks the controller), flash via the web UI's firmware upgrade page. If the web UI is too unstable to flash through, escalate to SD-card recovery.

Sensor calibration cross-check. Touch a calibrated thermal probe (Type-K thermocouple with a Fluke 117 or equivalent multimeter) to the heatsink base on the chip side, run the miner to steady state, and compare against the dashboard `Temp2` reading. They should agree within `5-8 C` once you account for the heatsink-to-junction thermal gradient. If the dashboard reads `60 C` and your thermocouple on the heatsink reads `90 C`, the sensor is reading low and the miner is silently overheating - hard stop, ship to D-Central for sensor and silicon inspection.

Stop DIY when any of these are true: `Temp2` sustained above `80 C` (silicon damage starts), `Temp1 / Temp2` cold-state spread exceeds `5 C` (sensor calibration drift), readings did not move after Tier 1 + 2 + 3 (hardware fault), elevated `Temp2` paired with chip-count mismatch (silicon degradation), `Temp2` reading lags ambient changes (sensor or filter-cap failing), or any visible thermal damage on the hashboard. Book a slot at https://d-central.tech/services/asic-repair/.

What D-Central does at the bench. Sensor calibration verification against a known-good reference KS-series, full hashboard thermal service (paste + pads + heatsink reseat), chip-level temperature mapping with thermal-camera capture, `1004LV100` chip replacement if individual chip hotspots are confirmed, fan harness rework, post-service `24-hour` burn-in at nameplate (KS3M ~`6 TH/s` @ ~`3,400 W`; KS5L ~`12 TH/s` @ ~`3,400 W`; KS5M ~`15 TH/s` @ ~`3,400 W`) with `Temp1` / `Temp2` logged every 60 seconds. We ship the unit back with the burn-in log so you can verify the new steady-state reading against this page's table.

Pack and ship. Double-box the chassis. Hashboards in anti-static bags if you're shipping just the boards. Include a printed note with current `Temp1` / `Temp2` readings, ambient temp, firmware version, observed symptoms, and your contact info. A web UI screenshot of the Status page during the elevated-temp event saves bench-reproduction time and reduces the bill. Canadian customers ship to our Quebec bench; turnaround `5-10` business days. US and international welcomed - the repair desk handles international shipping paperwork.