The Antminer B3 is a legacy Blake2b (Siacoin) ASIC, so it runs its own dedicated firmware — SHA-256 images built for an S9 or S17 will not load and will not work. Reflashing a B3 is a two-stage job: you upload a first firmware image to fix the fan-speed reporting, boot the miner and confirm it is healthy, then upload a second image that raises the hash rate. Both files are flashed the same way, through the miner’s built-in web interface. The single rule that never bends: do not cut power or pull the network cable while an image is writing. Interrupting a firmware write mid-flash is the fastest way to brick a control board.
Before you start
- Put the B3 on a wired Ethernet connection to the same LAN as your computer. Firmware flashing over anything flaky is asking for a half-written image.
- Feed the miner from a healthy PSU and a circuit that will not trip. A power blip during the write is exactly the failure you are trying to avoid.
- Have both firmware files downloaded to your computer before you begin, so nothing stalls mid-process:
ANTMINER-B3-201805232020-750M.tar.gz— corrects the fan-speed display problem. Flash this one first.update_1000.tar.gz— improves the B3’s hash rate. Flash this one second, only after the miner is confirmed running.
- Know the miner’s IP address. If you do not, use the miner’s IP-report button together with Bitmain’s IP Reporter tool, or check your router’s DHCP client list for the new device.
Stage 1: Flash the fan-fix firmware
Do this stage first and let it finish completely before touching the second file.
- Open a browser and enter the B3’s IP address to reach its web interface. Log in with the miner’s credentials (default is
root/rooton this generation unless you changed it). - Find the firmware upgrade section of the interface — on this classic Antminer UI it lives under the System menu. Labels vary slightly between builds; look for the option that lets you upload a firmware image file.
- Choose the
ANTMINER-B3-201805232020-750M.tar.gzfile. If the interface offers a keep-settings option, leaving it checked preserves your pool configuration; unchecking it returns the miner to defaults. For a clean reload, an un-checked (full reset) flash is the more reliable path. - Start the upgrade and walk away from the power switch. A normal flash finishes in about five minutes. If it has not completed after roughly ten minutes, treat the process as stalled, power-cycle, and re-attempt the flash rather than waiting indefinitely.
- The miner reboots on its own when the write completes. Wait for it to come back on the network.
Stage 2: Bring the miner up and verify
- Reacquire the IP address if it changed on reboot, and log back into the web interface.
- Point the miner at the algorithm-appropriate pool (a Blake2b / Siacoin pool — the B3 cannot mine Bitcoin) and let it run for several minutes.
- Open the miner’s status page and confirm every hashboard is reporting. All ASICs should show a live count. If any chip position shows
xor??, that board is not fully working — do not proceed to the second firmware flash until you understand why, because a bad board will not be fixed by a firmware image. - Confirm the fans are spinning and their RPM now reads correctly — that is the specific defect the first image corrects.
Stage 3: Flash the hash-rate firmware
- With the miner confirmed healthy and logged in, return to the firmware upgrade section.
- Select
update_1000.tar.gzand start the upgrade. Same discipline as before: no power interruption, no unplugging. - On success the interface reports “System Upgrade Success.” The miner reboots, and after it rejoins the pool you should see the higher hash rate reflected on the status page over the following minutes as the shares settle.
Common mistakes
- Flashing the wrong-order or wrong-algorithm image. The B3 is Blake2b hardware; only B3 firmware belongs on it. Flash the fan-fix image before the hash-rate image, not the reverse.
- Killing power because it “looks stuck.” A write can appear frozen while it is finishing. Give it the full ten-minute window before intervening.
- Flashing a board that is already faulty. If a hashboard shows
??before you start, firmware will not repair it. That is a hardware fault, not a software one. - Ignoring a failed upload. If the browser reports an upload or verify error, re-flash on a stable wired link before rebooting — do not leave the miner running on a partial image.
If it will not come back
If a flash is interrupted and the miner no longer serves its web interface, the control board can usually be recovered by writing a recovery image to a microSD card and booting the board from it — the standard fallback for this Antminer generation. If recovery fails, or if hashboards keep reporting ?? and dead ASIC counts after a clean reflash, the fault is on the hardware side. At that point the board needs bench diagnosis and component-level repair, not another firmware attempt.
Related: Work through the model-specific procedures in our ASIC miner manuals, isolate a dead board with the ASIC fault finder, and when a hashboard is genuinely failed, start a repair to send it in for bench-level service.