Bitaxe – Overheating

Power-cycle the Bitaxe from the wall - not a soft reboot in AxeOS, a full unplug for 30 seconds. This clears any wedged I2C state or latched overheat flag that a warm reboot will not reset. Plug back in, wait 2 minutes for boot, then re-check temps in AxeOS. If the display was stuck on an overheat warning but temperatures were actually normal, this alone often clears it.

Open AxeOS, navigate to Settings and Preset, and set the preset to default (stock). Save, then reboot. Watch the ASIC Temp reading for 15 minutes. A cool drop of 10 C or more confirms your overclock was the cause - rebuild tuning more conservatively using the slow-add method in step 9. If no change, your thermal issue is mechanical or environmental, not firmware tuning.

In AxeOS, set fan to manual mode at 100%. Confirm Fan RPM reads at the rated speed for your hardware (approximately 6,000 RPM for stock 40 mm, approximately 4,500 RPM for 60 mm Hex fans). If RPM reads 0 or far below rated, the fan is dead, unplugged, or its connector is oxidized - proceed to Tier 2 fan replacement. If RPM is correct but temps are still high, the problem is downstream of the fan.

Check AxeOS, System, Firmware version. If you are on v2.7.1, v2.8.0, v2.8.1, v2.9.0, v2.11.0, or any v2.14.0-beta, these are known-problem builds for thermal sensing. Download the latest stable release from github.com/bitaxeorg/ESP-Miner/releases, flash both esp-miner.bin and www.bin via the AxeOS web uploader, and power-cycle. Patched v2.9.1+ resolves the EMC2101 I2C race that causes -1 C ghost readings.

Remove the Bitaxe from any enclosure, sealed case, or dense stack. Place it on an open bench with a minimum 10 cm clearance on every side. Measure intake air temp with an IR thermometer directly at the fan grille. If it is above 28 C on a Gamma/Ultra or above 32 C on a Hex, your ambient is outside the thermal envelope - relocate, add room ventilation, or wait until the ambient drops before making other changes. Environmental fixes are free; paste jobs on a cooked chip are not.

Power off and unplug the Bitaxe completely. With a Phillips #00 or #0 screwdriver (Ultra/Supra/Gamma) or the Hex's hex-key retention, remove the heatsink hold-down screws. Lift the heatsink straight up - do not twist, which can lift the ASIC's BGA pads. Inspect the thermal paste or phase-change pad: it should be a uniform thin film. Dried, cracked, squeezed-out, or separated paste means the thermal bond is broken. This is the single most common cause of BX_TEMP on Bitaxes over 12 months old.

Clean both the ASIC die and the heatsink contact surface with isopropyl alcohol 99% and a lint-free wipe. Do not use rubbing alcohol from a drug store - it contains water and oils that leave residue. Let both surfaces dry for 60 seconds. Apply a single rice-grain-sized drop of Arctic MX-6 (or MX-4, or Thermal Grizzly Kryonaut) to the center of the ASIC die. Do not spread - let the screw pressure spread it. Reseat the heatsink with even, gradual pressure on all screws (tighten in a diagonal pattern).

Power on and observe ASIC Temp for 20 minutes. Expect a drop of 8-15 C at stock if paste failure was the cause. If temps have dropped but you are still above safe limits at stock, your fan may be underperforming - inspect and replace per step 10. If temps did not change, the thermal paste was not the issue; proceed to step 9.

If you pushed an overclock before the thermal event, rebuild tuning from scratch - do not reuse the old preset. Starting at stock (confirmed stable with a baseline of 30 minutes of clean hashing), add 25 MHz at a time with a 10 minute stability window between each step. Watch both ASIC Temp and HW Err %. Stop at the last step BEFORE either metric moves outside its window (below 70 C and below 2%). That step is this specific chip's silicon-lottery ceiling. It varies per unit - do not copy someone else's preset.

If the fan is physically dead or under-RPM, order a replacement. Stock 40 mm PWM fans for Ultra/Supra/Gamma are approximately $12 CAD; stock 60 mm Hex fans are approximately $18 CAD - both available in the D-Central store. Disconnect from the fan header, match the connector orientation (keyed), and reinstall. Verify RPM in AxeOS before declaring the fix successful. Generic fans from Amazon often have wrong PWM curves - stick with Bitaxe-spec replacements.

If you suspect the EMC2101 is reading garbage despite current firmware, use an external IR thermometer or a K-type thermocouple to measure the heatsink temperature independently. Compare against what AxeOS reports. A discrepancy greater than 10 C - with firmware at v2.9.1+ - indicates a hardware sensor or trace fault. At this point the board needs bench diagnostic work; book a D-Central repair slot.

Upgrade the heatsink to a D-Central high-performance unit. We developed the first aftermarket heatsinks for both Bitaxe and Bitaxe Hex - larger fin surface area, better fan coupling, and a pre-cut thermal pad sized to the exact chip. On a Gamma running at a 550 MHz / 1.20 V preset, our heatsink typically drops ASIC Temp by 6-10 C vs the stock cooler. On a Hex running all 4 chips, the Hex heatsink is more or less mandatory at any meaningful overclock.

For users running a Bitaxe inside a 3D-printed case or stand: verify that the case is the D-Central-designed Mesh Stand (we pioneered the original) or a derivative with equivalent open-mesh sides. Solid-wall cases trap heat. If you are running a closed case, either swap to a mesh design or add an external 80-120 mm fan blowing across the Bitaxe at low RPM to provide forced convection through the case vents.

For Hex owners: all 4 BM1370 chips must have good thermal contact independently. A common failure mode is one chip running 15 C hotter than its neighbours because that chip's portion of the heatsink lost paste contact during assembly or shipping. Disassemble the heatsink completely, inspect each chip contact individually, and apply fresh paste to all four before reassembly. A Hex with one thermally-starved chip still hashes - but its HW error rate on that chip position will destroy your effective hashrate.

If you operate a Bitaxe at sustained temperatures above 65 C in a warm climate, consider under-volting instead of overclocking. Dropping Voltage by 25-40 mV while keeping Frequency at stock typically cuts power draw 15-25% and temperature 8-12 C - at the cost of 3-6% hashrate. On a solo miner rolling lottery dice, longevity is worth more than marginal hashrate. Bitaxes running cool and slow for three years hit more blocks than Bitaxes running hot for one.

Stop DIY when temps remain above 75 C at stock after fresh paste, a known-good fan, and firmware v2.9.1+; when ASIC Temp readings are inconsistent with an external IR probe and firmware is current; when you see visible damage (chip discoloration, burnt smell, melted solder mask, or a bulged capacitor near the VRM); or when the Bitaxe has suffered a physical drop, liquid spill, or reverse-polarity power incident. Any of these equals bench work, not desk work.

D-Central bench process for thermal-damaged Bitaxes: full visual inspection under a scope, I2C bus continuity test to the EMC2101, VRM output voltage measurement under load, ASIC junction-temp-vs-frequency sweep on our test fixture, thermal paste refresh, and - if the chip itself is damaged - salvaged-grade BM1366/BM1368/BM1370 chip replacement with hot-air rework. Because we stock every Bitaxe variant and carry the full parts catalog, turnaround is typically 3-7 business days from receipt. Ship Canada-wide or US/international.

Pack the Bitaxe for shipment: remove from any case or stand, wrap the PCB in an anti-static bag, place inside a padded box with at least 3 cm of foam or bubble wrap on every side. Do not ship with the PSU - we have them in stock. Include a note with your observed symptoms, firmware version, AxeOS preset settings at time of failure, and your contact info. This shortens our diagnostic and keeps your repair bill smaller.