Diagnostics · Avalon A1246

What it means: Avalon reports numerically via the cgminer API on TCP 4028, and estats is the richest dump available. Read it top-down: SYSTEMSTATU → ECMM → ECHU → MM_STATUS → PS → PVT_T/V → GHS. Compare GHSavg (measured) vs GHSmm (theoretical) and TA vs nameplate.

Symptoms

• Red chassis LED (one of 7 fault classes)
• GHSavg << GHSmm
• TA below nameplate

Field procedure

1. Pull estats (or the web UI status) and note SYSTEMSTATU, ECHU, PS, and PVT arrays.
2. ECHU[N 0 0] = fault on ONE board; ECHU[N N N] = controller/PSU/mains-wide.
3. PS ≠ 0 = a power fault (decode the bitmap); PVT extremes = temp/voltage per chip.
4. Use the specific fault class (below) to pick the field action.

What it means: estats shows ECHU[N 0 0] - one board missing/faulted. Because Avalon has per-slot addressing, a slot-swap test cleanly isolates board vs controller.

Symptoms

• ECHU[N 0 0]
• Low MGHS / TA down ~⅓
• Board reads no chips

Field procedure

1. Power off. Reseat that board's ribbon AND its DC power lug (a loose lug mimics a board fault).
2. Power on and recheck estats.
3. Still missing? SLOT-SWAP the board with a good one: if the fault FOLLOWS the board = board fault; if it STAYS with the slot = AUC/MM/controller fault.
4. Board-fault → RMA the board; slot-fault → swap the controller/AUC.

What it means: BOOTBY codes and SYSTEMSTATU tell you why. 0x02 = overheat, 0x03 = pool/DNS, 0x04 = net re-init, 0x11 = no hashrate in 5 min, 0x12 = <70% nameplate.

Symptoms

• GHSavg ~0
• BOOTBY 0x11
• SYSTEMSTATU < 3 ('no asics!!')

Field procedure

1. Check the LED + BOOTBY code. 0x02 → cooling; 0x03 → pool/DNS; 0x04 → reseat AUC USB.
2. Do a 60-second PDU cold-boot.
3. Verify mains voltage, temps, and DNS (1.1.1.1 / 8.8.8.8).
4. 0x11 / SYSTEMSTATU<3 after reseat + slot-swap → bench/RMA.

What it means: Avalon's PS field is a summed bitmap. Bit 1 = input under-voltage; bit 64 = output over-current (usually a hashboard SHORT); per-channel OC bits point to a specific board.

Symptoms

• PS ≠ 0
• Boards unpowered / PG-fail
• Breaker trip

Field procedure

1. Decode the PS value (sum of bits). Bit 1 → raise/steady mains ≥195-200 V.
2. Output over-current can indicate a board or power-path fault; it does not prove a shorted board. Authorized technicians may isolate it using the model-specific service procedure. Never disconnect live: isolate all supplies, control stored energy and verify the safe state before each connection change; use a controlled return to service and re-isolate before any further change.
3. Swap/RMA the shorted board; verify PS returns to 0.
4. Sensor bit (128) / fan bit (2048) → check the PSU sensor/fan.

What it means: The 'Toohot' red-fault class triggers when PVT_T exceeds ~85 °C. One extreme PVT_T vs neighbors points at that sensor/ASIC rather than a whole-board heat event.

Symptoms

• Red LED (Toohot)
• PVT_T > 85 °C
• One sensor wildly off

Field procedure

1. If broad PVT_T is high → clean, improve airflow, check thermal pads and ambient.
2. If ONE PVT_T is extreme while neighbors are normal → sensor/ASIC on that spot.
3. Confirm both front fans spin (No-fan class = RPM 0).
4. Recheck after cooling; persistent single-point extreme → board swap.