Pulling a hashboard is the first physical step in almost every ASIC repair: chip-chain faults, a dead voltage domain, a chain that won’t enumerate, or a board that needs to go on the bench for a standalone hash test. The core procedure is much the same across the common Antminer families. On the S17, S19 and S21 generations, three hashboards (Chain 0, Chain 1, Chain 2) bolt to a copper bus bar for power and connect to the control board with an 18-pin flat ribbon cable. The older S9 differs — it feeds hashboard power through per-board connectors rather than a bolted copper bus bar — but the same removal, ESD and handling discipline applies. This guide walks the removal end to end, with the safety and handling steps that keep a good board from becoming a dead one.
Before you touch anything: fully power down the miner, unplug the PSU from the wall, and disconnect the PSU output leads from the board. The boost circuit on each hashboard steps the 12–15 V PSU rail up to ~19–20 V and holds charge in bulk capacitors. Let the board’s bulk capacitors bleed down (a few seconds) before working on the boards.
Tools you’ll need
- Anti-static wrist strap and grounded mat — ESD kills ASIC chips silently (see the ASIC repair parts bench-supply list)
- Phillips #2 screwdriver (control-board cover and fan baffle screws)
- Nut driver / socket for the copper bus bolts (commonly 8 mm or 10 mm depending on model)
- Small flat pick or spudger to release the ribbon-cable latch
- A tray or anti-static bag to lay the board on chip-side up
- Your phone — photograph each connector and screw location before you remove it
Step-by-step: removing the hashboard
1. Remove the control-board cover
With the miner powered down and unplugged, take out the cover screws (locations vary by model — reference the per-model repair manuals). On tube-style Antminers there is a release detent or round button on the top cover; press it and lift the upper cover straight up to clear it. Set the screws aside grouped by where they came from.
2. Detach the rear fan assembly
Remove the screws holding the rear fan baffle and lift the fan assembly clear of the frame. The fan cable is still plugged into the control board — do not yank the baffle. Support the fan and either unplug its connector at the control board or leave enough slack so the cable is not stressed. Note the fan’s airflow orientation (intake vs. exhaust arrow on the housing) so it goes back the same way.
3. Disconnect the signal cable and free the power bus
Locate the 18-pin flat ribbon (signal) cable running from the hashboard to the control board. Release its latch or clip and pull the connector straight off — pull on the connector body, never on the ribbon, and don’t rock it side to side, as this bends the header pins. Then remove the fixing screws that hold the hashboard to the frame and the bolts securing the copper bus bar. The copper bus carries the full PSU current to the domains; keep track of any washers and note the stack-up order for reassembly.
4. Slide the hashboard out
With the signal cable off, the bus bolts out and the frame screws removed, gently work the board free and slide it out of the chassis. Handle it only by the PCB edges — do not touch the ASIC chips, the heatsinks, or the boost inductors, and never flex the board. The aluminum heatsinks are heavy and can lever the board if you let it hang; support both ends. Lay the removed board chip-side up on your anti-static mat and label which chain position it came from.
How to confirm you did it right
- No cables are pinched, stretched, or still attached to the removed board.
- The ribbon-cable header on the control board is undamaged — pins straight, none bent or pushed in.
- All screws, bolts and washers are accounted for and grouped by origin.
- The board’s heatsinks are intact and none of the aluminum fins were bent during extraction.
Common mistakes to avoid
- Skipping the discharge wait. Working on a board with charged bulk capacitors risks a short and a slipped tool across live copper.
- No ESD protection. A static discharge you can’t even feel can degrade or kill a chip in the daisy chain, turning a simple removal into a dead-chain fault.
- Pulling the ribbon by the cable. This is the fastest way to tear conductors or bend the header — a new fault you introduced yourself.
- Mixing up chain positions. If you’re troubleshooting, always note whether the board was Chain 0, 1 or 2 so the miner’s error log matches the physical board.
- Over-torquing on reassembly. Copper-bus bolts and hashboard screws should be snug, not gorilla-tight — over-torque cracks pads and warps the bus contact.
When to escalate
If the board came out because a chain won’t enumerate, is missing chips, or shows a dead voltage domain, removal is only the first step — the real diagnosis happens on the bench with a multimeter and a standalone test setup. Cross-reference the exact fault string your firmware reported in the ASIC fault finder before you start desoldering anything. BGA-level chip replacement needs a hot-air rework station, reballing tooling and a controlled profile; if you don’t have that bench, don’t guess.
Related
Reinstall is the reverse of this procedure: seat the board, hand-snug the copper bus and frame screws, reconnect the 18-pin signal cable straight and fully, then the fans — and never power on with a board only partially seated. For model-specific screw maps and torque callouts see our ASIC repair manuals, look up error codes in the ASIC fault finder, source replacement heatsinks, ribbon cables and hardware from ASIC repair parts, or hand the job to our bench at start a repair.
