“No Hardware Version Found” and a blank “BMMiner” field on the S9/T9 status page mean one thing: the mining daemon (bmminer) never came up, so the control board has nothing to report. On an Antminer S9 or T9 the front-page firmware banner is populated by bmminer after it boots, enumerates the three hash-board chains, and starts hashing. If it crashes or exits during init, those fields stay empty and the kernel log fills with bmminer not found, restart bmminer as the watchdog tries to relaunch it. This is a startup/detection failure, and the cause is almost always one of four things: a fan safety trip, an undetected hash board (usually a ribbon-cable seating problem), corrupted firmware, or a failing control board.
What is actually happening
The S9/T9 control board runs a small Linux system on a Xilinx Zynq 7010 SoC (dual Cortex-A9 at 667 MHz). At boot it launches /usr/bin/bmminer — Bitmain’s CGMiner 4.9.2 fork — which reads /config/cgminer.conf, spins the fans, checks each of the three hash boards over their 18-pin ribbon cables, then reports the version strings to the web UI. If bmminer can’t pass its fan check or can’t talk to a chain, it aborts before those strings are ever written. The empty banner is the symptom; the job is to find why the daemon won’t stay up.
Fix it in order — cheapest and most likely first
1. Reboot cold, then read the kernel log
Power the unit off at the PSU, wait 30 seconds, power back on, and give it a full 5 minutes to attempt init. Then open the web UI and go to System → Kernel Log (or Miner Status → Kernel Log). The log tells you which failure you have:
Fatal Error: some Fan lost or Fan speed low→ go to step 2 (fan).no risc detected,Chain X find 0 asic, or fewer than 3 chains detected → go to step 3 (hash board / ribbon).bmminer not found, restart bmminerlooping with no chain detection at all → go to step 4 (firmware) then step 5 (control board).
2. Clear a fan safety trip
The S9/T9 refuses to start hashing if a required fan is missing or reads too low an RPM — a hard safety interlock, since the ASICs will cook without airflow. If the kernel log shows the fan fatal error:
- Confirm both fans are physically plugged into the control board and that the connectors are fully seated.
- Compare the two fans. If one spins weakly or not at all, swap the front and rear fan positions and reboot. If the fault follows the fan, that fan is bad; if it stays on the same header, suspect the control-board fan port.
- Replace any fan that fails to hold rated RPM. Use a correct 6000 RPM-class 4-wire (PWM/tach) 12V fan — a fan without a working tach wire will always trip the check even if it spins.
3. Reseat and verify the hash boards
On the S9/T9 each hash board connects to the control board with its own 18-pin flat ribbon cable (there is no shared bolted bus-bar — that is an S17/S19/S21 arrangement). A cocked or half-seated ribbon, or a bad cable, makes a whole chain vanish and stalls bmminer.
- Power down. Unplug all three signal ribbons and the PCIe power from every board.
- Reseat each ribbon squarely — both ends, control board and hash board — and reconnect power. Inspect ribbons for cracks or scorching; swap in a known-good cable if in doubt.
- Reboot and re-read the kernel log. Each of the three chains should enumerate its full complement of chips. A healthy S9 hash board carries 63 BM1387 chips (arranged in 21 voltage domains of 3 chips each); the whole miner is 189 chips across three boards. A chain reporting
0 asicor a partial count after a clean reseat points to a dead board, not a cable.
To isolate a suspect board, connect just one hash board at a time and reboot. If two of three boards let bmminer start and the banner populates, the odd board out is the fault.
4. Reflash the firmware
A truncated update, a bad flash, or a corrupted config can leave the bmminer binary or cgminer.conf unusable so the daemon dies on launch. Reflash with the official signed firmware image for your exact model (S9, S9i, S9j, T9, T9+) from the manufacturer’s release portal — never a random forum download, and never an image for a different model. Some S9 control boards boot from a microSD card rather than onboard NAND: if yours has an SD slot, a corrupt card is a common cause of this exact symptom — reseat it, and re-image it with the correct firmware if reflashing over the network fails. After flashing, let the miner run at least 20 minutes before judging the result, since it re-tunes frequency per chain on first boot.
5. Isolate the control board
If firmware is clean and the boards are good but the banner is still empty, test the control board alone: disconnect all three hash boards from both the PSU and the control board, then boot with only the control board powered. If bmminer still won’t start and the log shows no chain activity, the control board (Zynq SoC, its NAND/SD, or the signal buffers) is the fault and should be replaced. A control board that won’t get its own OS up will never report a hardware version no matter how healthy the hash boards are.
6. Check pool and network only if the daemon actually runs
Pool or network problems do not cause “No Hardware Version Found” — if bmminer is up enough to reach a pool, it has already written the version banner. Once the banner returns and you’re instead chasing dead pools, confirm the pool URLs on the Miner Configuration tab have no typos or stray spaces and use Network → Diagnostic → Ping to confirm the miner reaches the internet. Treat this as a follow-on, not a cure for the blank banner.
Verify the repair
You’ve fixed it when, after a full boot, the status page shows a populated hardware version and BMMiner version string, all three chains report their 63 chips (189 total), and the kernel log is free of fatal fan or find 0 asic errors. Let it hash for 20 minutes and confirm the total hashrate settles near the model’s rated figure (~13.5 TH/s class for the S9) with no chain sitting at zero.
Common mistakes
- Judging too fast. First boot after a reflash always looks ragged while chains re-tune — wait the full 20 minutes before concluding anything.
- Ignoring the tach wire. A fan that spins but has a broken tachometer wire trips the fatal-fan check exactly like a dead fan.
- Blaming the pool. An empty version banner is a local init failure; pool settings can’t produce it.
- Forcing ribbon cables. The 18-pin connectors are fragile — a bent pin from a forced reseat creates a new “dead chain” fault.
When to escalate
If the control board is confirmed dead, or a hash board reads zero chips after a clean reseat and known-good cable, the unit needs bench-level work — chip-level reflow or a board swap. That’s the point to hand it to a repair shop rather than keep swapping parts blind.
Related: Work the symptom systematically with our ASIC fault finder, source a replacement fan, ribbon cable, or control board from ASIC repair parts, and when a unit needs bench repair, start a repair with us.
