How to fix “Chain only find 0 ASIC” on Bitmain Antminer S19 Pro ASICs

“Chain only find 0 ASIC” means the Antminer S19 Pro’s control board ran its boot-time chain scan on one hashboard and not a single ASIC answered. Firmware then powers that board off — you drop from three boards to two and lose roughly a third of your hashrate. A healthy S19 Pro board enumerates 114 BM1398 chips; “find 0 asic” is the worst case, and it points to a break at or before the first chip in the daisy chain, or total loss of power/signal to the board — not a scattered chip failure. This guide walks the fault from log to bench so you fix the right thing instead of blindly reballing chips.

What the error actually reports

Each S19 Pro hashboard is a daisy chain: the control board pushes a clock (CLK), a reset (RST), and commands (CI) forward through all 114 chips, and responses (RO) travel backward to RX. At boot the firmware resets the chain, broadcasts a SET_ADDR, and counts the replies. That count is what the kernel log prints:

• Chain[1] find 114 asic — healthy board.
• Chain[1]: find 29 asic — a partial count; the break is at the chip immediately after the last one detected.
• Chain[1]: find 0 asic then will power off hash board 1 — nothing responded at all.

Zero is diagnostically different from partial. Partial is a mid-chain break; zero means the controller could not even reach chip 1 — no working power/signal to the board, a dead first chip, or a severed response path at the head of the chain. Chains are numbered by hashboard (0/1/2), so the log tells you which board to pull.

Likely causes of a 0-ASIC chain

• Ribbon cable / connector fault. The 18-pin signal cable carries 3.3 V logic power, CLK, CI/RO, RST, and the I2C lines. One broken conductor or a crooked plug kills the whole board’s comms.
• Dead control-board channel. The connector/level-shifter for that slot has failed, so no working board plugged into it will scan.
• Board power gone. If the boost circuit (~12.6 V in → ~20 V out) or the PIC that gates power has failed, the chips never come up and the chain reads empty.
• PIC / EEPROM handshake failure. The board’s PIC microcontroller (U6) and EEPROM (U10) must answer over I2C before firmware trusts the board; a dead PIC or corrupt EEPROM can stop it enumerating.
• First chip dead or shorted. If chip 1 opens, it stops forwarding CLK/CI/RST and its RO never returns — every chip behind it goes invisible. A shorted chip pulls the shared signal lines down and the whole chain reads zero.

Diagnose before you desolder anything

Most 0-ASIC calls are cable, connector, or control-board faults — not silicon. Isolate with swaps first:

1. Confirm the chain in the kernel log. One board or all three? All three reading 0 usually means a control-board, PSU, or firmware problem, not three dead boards.
2. Swap the ribbon cable between the faulty board and a known-good one. If the fault follows the cable, replace the cable and you’re done.
3. Swap board position. Move the suspect board to a different slot/channel (or a known-good control board). If it now scans fine, the channel was at fault; if the fault follows the board, it’s board-level.
4. Reseat and inspect connectors at both ends for bent pins, corrosion, or dust — a common, free fix on used-market units.
5. Power off, wait for discharge, and inspect the board on an anti-static mat under good light: burn marks, lifted/cracked chips, corrosion, or water staining all point to where to probe.

Electrical measurements

Safety: the low-voltage DC domains on the hashboard (single-digit to ~20 V) are not a shock hazard. The PSU is: the S19 Pro’s APW12 is a 200–240 V-only unit and its mains-side/PFC capacitors hold ~400 V after unplugging — never probe the PSU internals until they’ve discharged. Bench-test on a fixture, not the live miner, when you can.

1. Confirm board input and boost. On a fixture, verify the input rail is present and the boost stage is producing its ~20 V output. No boost = no chip power = 0 ASIC.
2. Walk the domain voltages against a known-good board. The S19 Pro splits its 114 chips into 38 domains of 3 chips, each on its own LDO. Black lead on board ground (never the heatsink), red lead on each domain test point. A domain near zero (shorted chip) or unusually high (open chip/LDO) flags the failure — and an abnormal domain voltage stalls the whole board.
3. Check signal integrity into chip 1. Confirm the control board is delivering the 25 MHz CLK, the RST pulse, and CI commands (these lines swing 0–1.8 V; RI/RO ~0.3–1.8 V). Missing clock or reset at the head of the chain explains a 0 count on its own.
4. Trace to the break. If power and input signals are good, probe the RO/RI handoff chip-to-chip from chip 1. Where the response disappears is your dead or non-forwarding chip.

Board-level repair

Once the fault is localized to a chip or cold joint, reflow the suspect CLK/CI/RST/RO joints first — a forwarding failure is often just a cracked ball on a live chip. If reflow doesn’t restore it, replace the chip. The S19 Pro uses the BM1398; fit a matched BM1398 from a donor board or new stock with a proper stencil, reball, and the correct hot-air/preheat profile. Firmware re-enumerates on the next boot, so don’t force address or frequency. Any liquid-cooled S19-family board is fanless and serviced differently — don’t apply air-unit airflow assumptions to a hydro board.

Verify the fix

1. Reassemble, reconnect power and the control board per Bitmain’s connection order, and boot.
2. Watch the kernel log: the repaired chain should now report find 114 asic, not 0.
3. Let it run 15–30 minutes and confirm the board holds its share of hashrate with no power off hash board re-trip and normal per-board temperatures. Only trust the fix after a full cold boot reaches steady-state.

Common mistakes

• Reballing before swapping the cable. Cable and connector faults mimic dead silicon and cost nothing to rule out.
• Grounding a probe on the heatsink. Use a real board-ground point; the heatsink can float and short something.
• Treating a partial count like a 0 count. “Find 29 asic” is a mid-chain break; “find 0 asic” is a head-of-chain or whole-board failure — different repairs.
• Opening the PSU without discharging. Its mains-side capacitors are the only lethal part of this machine.

Related

If the board scans a partial count or a different fault, work it through the ASIC fault finder. Source BM1398 chips, ribbon cables, and rework consumables from ASIC repair parts. If board-level rework is out of scope, start a repair and let D-Central’s bench diagnose and fix it — see the full ASIC repair service.