ASIC Repair Basics: How to Replace Fans, PSUs, and Hashboards Yourself

An ASIC miner is, mechanically speaking, three things bolted into a metal box: a pair of fans, a power supply, and a stack of hashboards. When a unit dies, it is almost always one of those three components — and almost never all of them at once. The bad news is that Australian repair options are thin on the ground: most warranty work routes through China, turnaround is measured in months, and out-of-warranty repairs are often not worth the freight. The good news is that the most common failures are also the most fixable at the kitchen table, provided you are comfortable with a screwdriver and willing to read a multimeter.

This guide covers the three repairs that account for the overwhelming majority of dead miners in Australian homes: fan replacement, PSU swaps, and hashboard diagnostics. None of it requires soldering. All of it assumes your unit is already out of warranty — if it is not, see our guide on mining hardware warranties in Australia before you pick up a screwdriver, because opening the chassis voids most of them.

Before you start: diagnosis comes first

The single biggest mistake Aussie miners make is replacing the wrong part. A unit that won't power on might be the PSU — or it might be a tripped circuit, a dead C19 cable, or a hashboard short pulling the PSU into protection. A unit that powers on but reports zero hashrate might be a dead hashboard — or it might be a control board that lost its firmware after a brownout.

Start with the miner's own log. Every modern ASIC exposes a web dashboard that records temperatures, hashrate per board, and fan RPMs. If you are not already familiar with reading those logs, our guide on how to read your miner's stats walks through every field that matters. Look for:

• Fan RPM = 0 on one or more fans → fan failure (or fan header failure on the control board)
• "PSU error" or unit doesn't power on at all → power supply
• Hashrate drops to 2/3 or 1/3 of nameplate, or one board shows "X" in the dashboard → hashboard fault
• Hashrate flickers in and out, high reject rate → could be any of the above, or a network/firmware issue

If you are not confident the unit is the problem at all, run through the basics first: different outlet, different C19 cable, different Ethernet port. More than one "dead" miner has turned out to be a $4 power cord.

Tools you actually need

You do not need a workshop. The bare minimum:

• A Phillips #2 screwdriver and a small flat-head
• A digital multimeter (any $40 Jaycar unit will do)
• An anti-static wrist strap, or at minimum the habit of touching the bare metal chassis before handling boards
• Thermal paste (Arctic MX-4 or similar) if you are pulling heatsinks
• A clean, well-lit bench — not the lounge room carpet
• Compressed air and a soft brush for cleaning, which solves more problems than people admit

Optional but useful: a spare PSU of known-good condition (an APW12 or similar) for swap-testing, and a hashboard tester if you plan to do this more than once.

Fan replacement

Fans are the most common failure on any ASIC, and the easiest fix. They are also the most preventable — the majority of fan deaths in Australia trace back to dust ingress and heat, both of which our thermal management guide for the Australian summer covers in detail. A dead intake fan kills hashrate within minutes as the boards throttle; a dead exhaust fan can cook a hashboard before the firmware shuts the unit down.

Symptoms

• Dashboard reports 0 RPM on one fan
• Unit shuts down within minutes of starting under "high temperature" error
• Audibly louder than usual (remaining fans spinning at 100%)
• Visible damage, cracked blades, or a fan that won't spin freely by hand

The replacement

1. Power down and unplug everything. Wait two minutes for capacitors to discharge.
2. Identify the fan. Most modern Antminer-class units use 12038 fans (120 × 120 × 38mm). WhatsMiner M30S/M31S+ units use 14038 fans. Older units use 12038 or 12025. Read the sticker on the dead fan before you order — pin count (usually 4-pin PWM) and connector type matter.
3. Unscrew the fan shroud. Four screws on the chassis side, usually Phillips. The fan will lift out as an assembly.
4. Unplug the fan header from the control board. Note the orientation — the connector is keyed but cables can get pinched.
5. Fit the new fan in the same orientation. Airflow direction matters: intake fans push air in, exhaust fans pull air out. The arrow on the fan housing tells you which way air moves.
6. Reconnect the header, screw the shroud back down, power up. Check the dashboard — the new fan should report RPM within 30 seconds.

Cost: $15–$40 per fan depending on brand. Buy genuine replacements rated for at least 6000 RPM and IP55 if your unit lives anywhere remotely dusty.

PSU replacement

The power supply is the second most common failure, especially in regions with unstable mains voltage or frequent storms. PSU electrolytic capacitors age out, fans inside the PSU itself die and let the unit overheat, and surges from the grid kill input-side MOSFETs. Most Bitmain PSUs (APW12, APW17) are sealed boxes — you don't repair them, you replace them.

Symptoms

• Unit completely dead, no lights, no fan twitch on power-on
• Unit powers on briefly then shuts off (PSU going into over-current protection — could also be a shorted hashboard pulling it down)
• Burnt smell or visible scorching on the PSU connectors
• PSU's own fan is dead and the unit overheats within 10 minutes

The replacement

1. Confirm the diagnosis first. Disconnect the PSU's output cables from the hashboards entirely, then power the PSU on its own. If it still won't turn on (no fan, no output), it's the PSU. If it powers on fine with the hashboards disconnected, the fault is downstream — likely a hashboard short pulling the PSU into protection.
2. Match the replacement to the unit. An S19 needs an APW12; an S21 needs an APW17 or equivalent. Mismatched PSUs either won't have enough wattage or won't have the right output voltage profile for the hashboards. Wattage headroom of 10–15% above the miner's rated draw is sensible.
3. Disconnect everything from the old PSU — typically a set of three high-current DC cables to the hashboards plus a control signal cable to the control board.
4. Unscrew the PSU from the chassis. On most Antminers the PSU sits on top in its own compartment and lifts straight out.
5. Drop the new unit in, reconnect the cables in the same order. The DC output cables are colour-coded and keyed; the signal cable is usually a small JST connector. Don't force anything.
6. Plug into the wall on a known-good circuit and power up.

If you are sourcing a replacement, a quality 2000W+ unit like the 2000W full-modular ATX PSU we stock works for some setups, but note that ATX PSUs are not a direct substitute for an APW-series Bitmain PSU on an S19/S21 — those miners expect a specific signal handshake. ATX PSUs are appropriate for GPU rigs and for some open-source/Bitaxe builds, not for stock Antminers. When in doubt, match like-for-like.

Hashboard diagnostics and reseating

This is the hardest of the three and the one most likely to end with "post it overseas." Most ASIC chip failures cannot be home-repaired — they require reballing the BGA chips with a hot air station, which is a specialist job. But a surprising number of "dead" hashboards are actually fine boards with a bad data cable, a loose connector, or a single thermal pad that has gone hard. Worth checking before you write the board off.

Symptoms

• Dashboard shows hashrate at exactly 2/3 (one board dead) or 1/3 (two boards dead) of nameplate
• "X" or "fault" indicator on a specific board in the web UI
• Asic chip count shows fewer chips than nameplate (e.g., "126/195" on an S21 Pro)
• Board temperature reads as 0°C, -127°C, or some other obviously wrong value (usually a dead temperature sensor, sometimes the whole board)

What you can fix at home

1. Power down, unplug, ground yourself. Hashboards are static-sensitive.
2. Remove the chassis lid.
3. Identify the faulty board from the dashboard log. Boards are usually numbered 0, 1, 2 from one side.
4. Unplug the board's data ribbon cable from the control board. These ribbons are a major failure point — bent pins, corrosion, or a cable that has worked loose causes a lot of "dead" boards. Inspect the ribbon and both connectors for damage. Replace the ribbon if it looks suspect ($5 part).
5. Reseat all power connectors on the board. The high-current copper bars or screw lugs that feed the board sometimes loosen with thermal cycling — finger-tighten plus a quarter turn with a spanner.
6. Inspect the board visually. Look for: burnt chips (black scorching), bulged capacitors, lifted solder pads, or thermal paste that has completely dried out. A burnt chip is game over for home repair. Dried paste is fixable.
7. If you are pulling the heatsink to repaste, work slowly. The heatsink is held to the chip array by thermal adhesive on older boards and by clamps on newer ones. Forcing it off can lift a chip. Warm the board gently with a hairdryer first to soften old paste.
8. Reassemble, power up, watch the dashboard. If the board now reports correctly, you've won. If chip count is still low, the board has died at the silicon level and needs professional repair or replacement.

Replacement hashboards are available for popular models (S19, S19j Pro, S21) on the secondary market in Australia, but pricing is volatile. Sometimes a full second-hand miner is cheaper than three replacement boards — see our notes on new vs second-hand ASIC miners for context, and mining hardware resale value for what a dead-board unit is realistically worth as a parts donor.

When to stop and send it out

Be honest about what you can and can't do. The following are not home repairs:

• Reballing or replacing individual ASIC chips on a hashboard
• Replacing surface-mount capacitors or voltage regulators
• Re-flashing a control board that has lost its bootloader
• Repairing a PSU's internal PCB

Each of those needs hot air, a microscope, and someone who does it every day. In Australia, dedicated ASIC repair shops exist in Sydney, Melbourne, and Brisbane — expect $80–$200 per board for chip-level repair, plus freight. Whether that's worth it depends on the age of the miner and the going hashprice. If the unit is a generation old and out of warranty, parts harvesting plus replacement is often the better economic call. Our guide on when to upgrade your miner covers the maths.

Preventive maintenance is cheaper than repair

Almost every fan failure, PSU failure, and hashboard failure on a home miner traces back to one of three things: dust, heat, or dirty power. Quarterly compressed-air cleaning, a half-decent surge protector, and keeping intake temperatures under 35°C will extend the life of an ASIC by years. We cover the cooling side in our thermal management guide, and the firmware side — including auto-tune profiles that reduce thermal stress — in our firmware guide and safe overclocking guide.

If you need replacement parts, fans, cables, or a known-good PSU to swap-test with, browse our rig accessories collection, or have a look at our full ASIC miners catalogue if you have decided the repair isn't worth it. We ship Australia-wide from Perth, and if you are not sure which part you need, email hello@minerhub.com.au with your miner model and the dashboard error — happy to point you the right way before you spend money on the wrong component.