Skip to content

We're upgrading our operations to serve you better. Orders ship as usual from Laval, QC. Questions? Contact us

Bitcoin accepted at checkout  |  Ships from Laval, QC, Canada  |  Expert support since 2016

BITAXE_OVERHEAT_PASTE Warning

Bitaxe – Overheats Despite Cooling (Thermal Paste Degraded)

Bitaxe ASIC sustains 80-95 C while ambient is OK and the fan is at 100%. Heatsink fins read only mildly warm to the touch. Classic die-vs-heatsink thermal gradient: factory thermal paste has dried, pumped out, or was applied poorly, so heat is not crossing the TIM layer to the heatsink. Hashrate drops 5-25% as AxeOS throttles to keep the chip below the 75 C self-protect latch.

Warning — Should be addressed soon

Affected Models: Bitaxe Supra (BM1368), Bitaxe Ultra (BM1366), Bitaxe Gamma 601/602 (BM1370), Bitaxe Hex (6x BM1368) - any variant where factory TIM has dried, pumped out, or was applied poorly.

Symptoms

  • AxeOS System Info -> Temperature reads sustained at 80-95 C while the miner is hashing
  • Fan duty is at or near 100% (commanded, audibly maxed) but ASIC will not cool below 78 C
  • Ambient measured at the fan intake with an IR thermometer is <= 25 C - room is NOT the problem
  • Heatsink fins feel only mildly warm to the touch (finger comfortable for 5+ seconds) while ASIC reports > 80 C - classic TIM-gradient signature
  • Hashrate has dropped 5-25% below variant nameplate (Ultra ~500 GH/s, Supra ~700 GH/s, Gamma ~1.2 TH/s, Hex ~3 TH/s)
  • AxeOS occasionally trips the 75 C self-protect latch (Device Overheat banner) under load
  • Frequency drops or Vcore cuts intermittently while AxeOS auto-throttles to stay below 75 C
  • Miner is 12+ months old and has been running continuously - factory paste at end-of-life window
  • Miner was recently re-pasted (by you or a previous owner) and overheating started or worsened after the work
  • Heatsink was bumped, dropped, or remounted recently - mount pressure may have shifted
  • On Bitaxe Hex: one of six chips runs 5-10 C hotter than the others in per-chip telemetry - localized TIM problem on that chip
  • Power draw is at expected nominal (12-25 W variant-dependent) - chip is producing the heat it should, so not an electrical issue
  • After cleaning heatsink + fan of dust, verifying ambient, and confirming fan RPM matches nameplate, temperature still will not drop

Step-by-Step Fix

1

Confirm the gradient. Boot to steady state for 20 minutes. Read AxeOS ASIC temperature - note the value. Carefully touch a heatsink fin (NOT the heat-pipe area or the chip directly). If ASIC reports > 80 C and the heatsink fin is comfortable to hold a finger on for 5+ seconds, you have confirmed a TIM gradient: heat is generated but not transferred to the heatsink. If both die and heatsink are hot, the problem is airflow, not paste.

2

Verify ambient at the intake. IR thermometer pointed at the air entering the fan. Target <= 25 C. If ambient is > 30 C, no paste job in the world will keep the chip cool - relocate the miner or fix room cooling first. Bitaxe in a south-facing office in July is the room talking, not the chip.

3

Compressed-air the heatsink fins from the exhaust side. A year of dust accumulation looks identical to TIM failure in AxeOS readouts. Three seconds of compressed air can save you a paste job. Push dust OUT through the exhaust, never IN through the intake. Re-test after dusting.

4

Set fan to manual 100% in AxeOS and re-test. Settings -> Fan -> Manual -> 100%. Forces maximum airflow regardless of auto-curve lazy ramping. If ASIC drops 5+ C from auto-mode, your fan curve was lagging the heat - paste is probably fine. If no change, paste is the problem.

5

Update AxeOS to the latest release before tearing into hardware. Several thermal-reporting bugs have been fixed across releases - make sure your readings are accurate before you trust them. Bitaxe Web Flasher (https://bitaxeorg.github.io/bitaxe-web-flasher/) handles this in two minutes.

6

Gather supplies. Isopropyl alcohol 99% (NOT 70% rubbing alcohol - must be 99% for clean evaporation), lint-free wipes (microfibre or paper, not cotton balls), Phillips #1 or T8 driver depending on variant, plastic spudger for adhesive-bonded heatsinks, and your chosen TIM: Arctic MX-6, Thermal Grizzly Kryonaut, or Honeywell PTM7950 pad. Work in a clean, well-lit, dust-free area - paste contamination is a real failure mode.

7

Power off, disconnect, wait 60 seconds. Discharge any residual capacitance. Move the miner to your workspace. Photograph the assembly from multiple angles before disassembly - reference for reassembly orientation.

8

Remove heatsink mechanism. On most Bitaxe variants this is 2-4 push-pins or screws-with-springs from the underside of the PCB. On thermal-adhesive bonded variants, warm the heatsink to ~60 C with a hairdryer for 30 seconds to soften adhesive, then gently pry with a plastic spudger. Never metal, never force - if it does not come off easily, stop and ship to D-Central. A broken die or lifted BGA pad costs 10x what a heatsink swap costs.

9

Clean both surfaces thoroughly. Apply IPA 99% to a lint-free wipe and clean every trace of old paste from the die top and the heatsink baseplate. Repeat with fresh wipes until they come away clean. Inspect under good light for any residue. Both surfaces should look mirror-clean with no visible paste streaks. Old paste residue under new paste creates an air gap that defeats the whole job - this step is non-negotiable.

10

Apply new paste. For MX-6 / Kryonaut: small rice-grain (~2 mm^2) at the centre of the die. Do NOT pre-spread - the heatsink mounting pressure spreads it uniformly. For PTM7950: cut a square of pad slightly smaller than the die top with a clean craft knife, peel one liner, place on die, peel second liner. Avoid bubbles and skin/finger contact - oils contaminate the bondline.

11

Reseat heatsink with even pressure. Lower straight down onto the die - no twisting or sliding (which moves paste off-centre and creates dry spots). Engage push-pins or screws in a cross / star pattern, tightening each a partial turn at a time so pressure builds evenly across the contact patch. Final torque: finger-tight + a quarter turn for screws-with-springs, full click for push-pins. Do not over-tighten - bending the PCB cracks BGA solder joints, which is far worse than warm paste.

12

Reconnect, boot, and observe 30 minutes of steady-state hashing. Record the new ASIC temperature. Expected: 8-18 C lower than before on a miner with degraded TIM. If improvement is < 5 C, your mount pressure or paste application is wrong - redo from Step 9. If you see 15+ C improvement and stable temps < 70 C, you are done.

13

Document your work. Photograph the new temperature reading and note the date in your records. Set a calendar reminder for 36 months from now to re-paste - MX-6 and Kryonaut are good for 3-5 years; PTM7950 for 5+. Tracking this beats guessing when a future overheat returns.

14

Upgrade to a D-Central Bitaxe heatsink if your existing heatsink is warped, undersized, or stamped aluminium. D-Central pioneered the Bitaxe heatsink ecosystem - we manufactured the first heatsinks for both standard Bitaxe variants and the Bitaxe Hex, with flat-lapped baseplates spec'd for BM1366/1368/1370 contact patches. Order at d-central.tech/product-category/bitaxe/, install with the same Tier 2 paste workflow.

15

Honeywell PTM7950 install (best long-term TIM). Phase-change pad: solid at room temp, melts at ~45 C, forms a stable bondline that resists pump-out indefinitely. Cut a square slightly smaller than the die top with a sharp craft knife on a clean cutting mat. Peel one liner, position on die, press gently to seat, peel second liner. Mount heatsink as in Step 11. First boot will show momentarily elevated temps (75-80 C for ~2 minutes) as the pad melts - this is normal phase-change behaviour. After phase change, expect rock-stable temps with no pump-out for 5+ years.

16

Bitaxe Hex per-chip TIM job. On a Hex with 5-10 C per-chip temperature spread, contact between heatsink and individual chips is uneven. Remove heatsink, clean all six die tops AND all six heatsink contact patches separately, apply paste (or PTM7950 pads cut to size) to each die individually, reseat with extra attention to even pressure across all six positions. Star-pattern tightening matters more on Hex than on any other variant. Verify in AxeOS per-chip telemetry that spread is < 3 C - if not, redo the chip(s) that lag.

17

Improve airflow ducting around the heatsink. A 3D-printed or sheet-metal shroud that channels fan air directly across the heatsink fins (rather than letting it spill around the sides) is worth 3-8 C on most Bitaxe setups. D-Central stocks several. The original D-Central Bitaxe Mesh Stand - which we manufactured first, before any clones - was designed specifically with airflow geometry in mind.

18

Tune OC/UV around the new thermal headroom. With re-paste + good heatsink + good airflow, you have earned 10-20 C of margin. Use 5 MHz of that for higher frequency, or use it for lower Vcore (run cooler, run quieter, run longer). Mining Hacker advice: take half the margin as performance, leave the other half as longevity insurance. Underclock-and-undervolt operators see 2-3x the chip lifetime of OC-to-the-edge operators.

19

Stop DIY and ship to D-Central if any of: die shows visible heat damage (discoloration, scorching, lifted edges); heatsink cannot be removed without pulling BGA off PCB (means thermal-adhesive bonded too aggressively at factory and needs a hot-air rework station); temperature stays high after a clean re-paste with a known-flat heatsink (chip itself is degraded - bench evaluation needed); or a Bitaxe Hex shows one chip persistently 15+ C hotter than its neighbours after a clean Tier 3 job (cracked thermal vias under that chip - chip-level rework on the daisy chain).

20

D-Central bench process: microscope inspection of die surface + heatsink baseplate flatness; hot-air station for adhesive removal without damaging BGA pads; die-level temperature mapping with thermocouple + IR camera; selective PTM7950 application on individual Hex positions; chip replacement (BM1366/1368/1370) with new-old-stock or graded recovered chips when die is damaged beyond TIM-level rescue; post-repair 24-hour burn-in with full AxeOS telemetry log. Turnaround 5-10 business days.

21

Ship safely. PSU disconnected, miner in anti-static bag, double-boxed with >= 5 cm of foam on every side. Include: serial number, AxeOS version, screenshots of bad temperature, what TIM you used and when, a paragraph describing the symptoms. Ship to D-Central ASIC Repair (d-central.tech/services/asic-repair/). Canada-wide, US/international welcomed.

When to Seek Professional Repair

If the steps above do not resolve the issue, or if you are not comfortable performing these repairs yourself, professional service is recommended. Attempting advanced repairs without proper equipment can cause further damage.

Related Error Codes

Still Having Issues?

Our team of Bitcoin Mining Hackers has been repairing ASIC miners since 2016. We have seen it all and fixed it all. Get a professional diagnosis.