Diagnostics · Auradine Teraflux AI3680 (Immersion)

This is shared Teraflux Teraflux reference material covering 3 models. Components, supply voltage/phase, PSU, cooling and service procedures vary by model and revision. Use the exact model nameplate and OEM manual; shared examples do not establish compatibility.

Diagnosis is a discipline: read the evidence, form one hypothesis, test the cheapest thing first, and only escalate to a swap/RMA once you've ruled out cabling, power, cooling, config, and firmware.

Safety first: never reseat cables or touch internals on a live unit. Power off at the PDU before any hands-on step.

Partial board (SMART)A board is down some chips but still hashinglow

What it means: On SMART/graceful-degradation hashboards (Bitdeer, Auradine), a board running below full chip count but still contributing is often DESIGNED behavior - not an automatic failure.

Symptoms

One board slightly below full hashrate
Chip count a bit low but board alive
Total hashrate a few % under nameplate

Field procedure

Confirm the board is still contributing (not at 0) in the web UI/API.
Check temps and reject rate - rule out a thermal/power cause.
Only escalate if the board is fully down, dropping over time, or out of spec after checks.

Resolution: Accept designed graceful degradation; escalate only a fully-down or worsening board.

Escalate when: Board fully at 0, or steadily losing chips after thermal/power checks → RMA.

No hashrateUnit up but not hashinghigh

What it means: Standard modern-miner flow: the unit boots and serves its UI but produces no shares. Diagnose like any Antminer-class unit.

Symptoms

0 hashrate
Pool shows the worker dead
Boards enumerate but no work

Field procedure

Check the web UI/API status and any fault indicator.
Verify network + DNS (pool won't connect without DNS) and the pool URL/worker.
Power-cycle; verify input voltage and temps.
Boards enumerate but no work after that → escalate to vendor support (younger RMA channel).

Resolution: Network/pool/power first; these boot fine but need the basics right.

Escalate when: Healthy boot, correct config, still no hash → vendor RMA/support.

Hydro / thermalCooling faults on air & hydro unitshigh

What it means: Air units throw fan/temp faults like any miner. Hydro units (SEALMINER A2 Hydro, Auradine AH) fault on the WATER side - flow, inlet temp, leaks.

Symptoms

Fan 0 RPM (air)
Flow/inlet-temp fault (hydro)
Over-temp throttle

Field procedure

Air: reseat/swap the fan; clean airflow.
Hydro: confirm pump + flow (e.g. 5-20 L/min), inlet temp in band, no leaks/air in loop.
Use AutoTemp/thermal-protection behavior to avoid hard shutdowns.
See the immersion/hydro Field School guide for the full liquid checklist.

Resolution: Air = fan/airflow; hydro = pump/flow/inlet-temp/leaks.

Escalate when: Hydro flow normal but temps climb → dry cooler / facility loop.

Network / poolUnreachable or pools deadmedium

What it means: Same as any modern miner: no link, wrong subnet, or bad DNS. Auradine's Zero-Touch Provisioning specifically needs DHCP/network correct at deploy.

Symptoms

Not found in a scan
All pools dead
Never onboards to the dashboard

Field procedure

Check the Ethernet link and that the unit is on the miner VLAN/subnet.
Fix DNS (8.8.8.8) - pools resolve by hostname.
For Zero-Touch units, confirm DHCP is handing out a lease at deploy.
Verify the pool URL/worker.

Resolution: Link → subnet → DNS → pool config, in that order.

Escalate when: No IP after cabling/subnet/DHCP checks → controller/board issue, vendor support.

Diagnostics · Auradine Teraflux AI3680 (Immersion)

Diagnosis is a discipline: read the evidence, form one hypothesis, test the cheapest thing first, and only escalate to a swap/RMA once you've ruled out cabling, power, cooling, config, and firmware.

Safety first: never reseat cables or touch internals on a live unit. Power off at the PDU before any hands-on step.

Partial board (SMART)A board is down some chips but still hashinglow

What it means: On SMART/graceful-degradation hashboards (Bitdeer, Auradine), a board running below full chip count but still contributing is often DESIGNED behavior - not an automatic failure.

Symptoms

One board slightly below full hashrate
Chip count a bit low but board alive
Total hashrate a few % under nameplate

Field procedure

Confirm the board is still contributing (not at 0) in the web UI/API.
Check temps and reject rate - rule out a thermal/power cause.
Only escalate if the board is fully down, dropping over time, or out of spec after checks.

Resolution: Accept designed graceful degradation; escalate only a fully-down or worsening board.

Escalate when: Board fully at 0, or steadily losing chips after thermal/power checks → RMA.

No hashrateUnit up but not hashinghigh

What it means: Standard modern-miner flow: the unit boots and serves its UI but produces no shares. Diagnose like any Antminer-class unit.

Symptoms

0 hashrate
Pool shows the worker dead
Boards enumerate but no work

Field procedure

Check the web UI/API status and any fault indicator.
Verify network + DNS (pool won't connect without DNS) and the pool URL/worker.
Power-cycle; verify input voltage and temps.
Boards enumerate but no work after that → escalate to vendor support (younger RMA channel).

Resolution: Network/pool/power first; these boot fine but need the basics right.

Escalate when: Healthy boot, correct config, still no hash → vendor RMA/support.

Hydro / thermalCooling faults on air & hydro unitshigh

What it means: Air units throw fan/temp faults like any miner. Hydro units (SEALMINER A2 Hydro, Auradine AH) fault on the WATER side - flow, inlet temp, leaks.

Symptoms

Fan 0 RPM (air)
Flow/inlet-temp fault (hydro)
Over-temp throttle

Field procedure

Air: reseat/swap the fan; clean airflow.
Hydro: confirm pump + flow (e.g. 5-20 L/min), inlet temp in band, no leaks/air in loop.
Use AutoTemp/thermal-protection behavior to avoid hard shutdowns.
See the immersion/hydro Field School guide for the full liquid checklist.

Resolution: Air = fan/airflow; hydro = pump/flow/inlet-temp/leaks.

Escalate when: Hydro flow normal but temps climb → dry cooler / facility loop.

Network / poolUnreachable or pools deadmedium

What it means: Same as any modern miner: no link, wrong subnet, or bad DNS. Auradine's Zero-Touch Provisioning specifically needs DHCP/network correct at deploy.

Symptoms

Not found in a scan
All pools dead
Never onboards to the dashboard

Field procedure

Check the Ethernet link and that the unit is on the miner VLAN/subnet.
Fix DNS (8.8.8.8) - pools resolve by hostname.
For Zero-Touch units, confirm DHCP is handing out a lease at deploy.
Verify the pool URL/worker.

Resolution: Link → subnet → DNS → pool config, in that order.

Escalate when: No IP after cabling/subnet/DHCP checks → controller/board issue, vendor support.