A fan reading is useful when it is attached to the right machine, firmware and operating state. Before treating a missing fan or unusual RPM value as a diagnosis, capture the complete status view and compare it with the log from the same interval. The useful result is a clear decision: continue a documented observation, stop under the site's operating procedure, or hand a specific fault record to qualified support.
This guide explains how to read and record Antminer fan status. It does not prescribe replacement parts, a universal RPM target or a repair sequence. The worksheet is LeedMiner's suggested operating record; the model-specific examples below come from BITMAIN documentation. The mining-site photographs illustrate equipment and operating environments, not a test of the machine you own.
1. Separate fan speed, control settings and machine state
BITMAIN's status-page reference defines the fan field as rotational speed in RPM and distinguishes chip temperature from PCB temperature. Its 19-series interface guide separately documents fan-status displays, fan-percentage settings and work/sleep modes. A control percentage and a measured speed therefore belong in different columns of your record.

Start with the full model name, installed firmware version and a private asset identifier. Then record the capture time, time zone and whether the equipment was starting, working, sleeping or stopping. If the interface does not expose one of those states, write “not shown” rather than inferring it from the fan sound. Include the time since the last restart when available.
Record every displayed fan channel, retaining its label and unit. A screenshot cropped to a single red number may remove the machine identity or another relevant reading. Keep the original capture and make a separate sharing copy if account or network details are visible. A repeated screenshot should have a new timestamp, so two observations cannot be mistaken for simultaneous readings.
| Field | What to record | What it does not establish by itself |
|---|---|---|
| Fan speed | Channel label, displayed RPM and capture time | Airflow through every part of the machine |
| Fan setting | Automatic/manual state or percentage, if shown | The actual measured RPM of every fan |
| Temperature | The full sensor label, value and unit | That an inlet, board and chip reading are interchangeable |
| Operating state | The state actually reported by the interface | Normal mining output merely because the page is reachable |
| Hashrate | The named measurement window and local source | The pool's accepted work over a different interval |
For comparisons, use a known working record from the same unit where possible. If comparing another machine, document the differences in model, firmware, operating mode and capture conditions. A convenient neighbor is not automatically a valid reference. Do not turn the highest RPM in a group into a target setting for the rest of the fleet.
If the question is whether mining output recovered, compare the appropriate intervals using our miner and pool hashrate guide. Fan readings provide equipment context; they do not replace accepted-work evidence.
2. Read an alert with its channel and surrounding log
BITMAIN's fan-problem reference describes a four-fan configuration whose FAN1–FAN4 interface labels correspond to controller connections. Use that mapping only for the documented configuration. It is not a statement that every air-cooled ASIC, hydro unit or modified machine has four monitored fans.

Save the exact alert wording and enough preceding and following log entries to preserve the sequence. Record whether it appeared immediately after startup, during an otherwise steady period, or following an operator action. Avoid translating an unfamiliar code into a fault name before checking the documentation for that model and firmware.
For example, BITMAIN's S19 XP log guide associates F:1 with fan abnormalities, P:1 with high-temperature protection, and J:6 with temperature-sensing trouble. Those examples help separate a fan event from other conditions in that documented family. They are not a universal code dictionary for newer firmware or another manufacturer's equipment.
| Observed symptom | Useful evidence to add | Decision to avoid |
|---|---|---|
| One channel shows zero or disappears | Full channel list, operating state, timestamp and matching log interval | Ordering a replacement from the screenshot alone |
| One reading differs sharply from the others | The same channel's earlier record and any associated alert | Applying a universal percentage difference as a failure rule |
| All readings change together | Mode, settings and site events recorded at the same time | Assuming every fan independently failed |
| Fan alert and temperature alert appear together | Exact sensor labels and the order of the events | Deciding which event caused the other without further evidence |
These are questions for an incident record, not automatic diagnostic rules. Mark each field as observed, reported by another person or still unknown. That distinction lets support request the missing evidence without treating an assumption as a measurement.
Do not bypass fan monitoring, disable protection or keep restarting an unsafe machine to obtain a cleaner screenshot. Physical inspection and any component work belong to the site's isolation procedure and the appropriate qualified service personnel. Preserve what can be captured safely; a complete log is not more important than stopping an unsafe operating condition.
3. Hand over a record that supports the next action
A useful handover connects the original symptom, the equipment identity and the decision still required. Separate “fan alert observed” from “component fault confirmed” and “repair verified.” Each is a different stage, and a case should not advance simply because someone refreshed the dashboard.

Use a short handover note with the asset identifier, exact model and firmware, first-observed time, affected channels, log interval and actions already taken. List the original and current settings where a change was authorized. If nothing was changed, say so. Keep a named owner and a specific next question, such as whether support needs another log interval or an isolated physical inspection.
Share records through the agreed private support channel. Our log export and redaction checklist explains how to preserve diagnostic context while removing unnecessary account and network identifiers. Do not include passwords or wallet secrets in the case. A public discussion copy and a manufacturer support package may need different levels of detail.
For equipment identification, the Antminer S21 XP catalog page is an example of an exact product reference. Its listing does not identify your individual unit or make the S19 XP fault-code examples applicable to it. Confirm the model on the equipment and use the documentation that matches that model.
Antminer S21 XP 270 TH/s

Air-cooled Bitcoin miner. Confirm your unit’s model before using a fault-code reference or service documentation.
After an approved service action, retain the original record and add a new one. Capture the same fields, identify the action and state the actual observation period. Check whether the original symptom recurred and whether relevant alerts remain. If the machine is intended to mine, record fresh worker activity separately. Do not invent a universal number of minutes that proves every repair is complete.
Close the case only against the agreed acceptance conditions. If the alert remains unexplained, label the case unresolved and carry forward the missing evidence. If it disappears, state what was actually verified and over what interval. This makes the record useful when a later operator sees the same symptom again.
Need help preparing an equipment or service inquiry? Contact LeedMiner with the model, firmware and a concise symptom summary. For hosted equipment, establish who can capture the local status and who owns the next service action before requesting additional screenshots.





