What does ERR Subsys=25 ECode=1 mean in an ArduPilot log?

Subsystem 25 is THRUST_LOSS_CHECK and ECode 1 is FAILSAFE_OCCURRED. The autopilot detected loss of thrust for at least one second with throttle above 90% and attitude under control. The matching MSG line names the motor number the firmware suspects.

How does the autopilot decide which motor failed?

It compares the attitude controller's output across motors. The motor that's saturated at maximum while attitude is still drifting is the one not delivering thrust — its diagonally opposite motor is correspondingly throttled down. The firmware reports a one-indexed motor number in the MSG line.

Can I disable the Thrust Loss Check?

Yes via the FLIGHT_OPTIONS bitmask (bit 0 = DISABLE_THRUST_LOSS_CHECK), but you almost certainly shouldn't. Without it, an ESC desync becomes an uncontrolled descent — the thrust-boost compensation depends on this detector firing. Disable it only on rigs where the false-positive rate is too high to fly.

Is RCOU saturation always ESC desync?

No. RCOU at maximum means the autopilot is asking for full thrust from that motor; the reason can also be voltage sag dropping the motor's effective thrust, a bearing failure spinning slower than commanded, or a mechanical bind. Confirm by checking the surrounding power and vibration traces — ESC desync typically has a sharp onset; bearing failure shows a gradual climb.

Does enabling ESC telemetry slow down the autopilot?

Negligibly on modern flight controllers. DSHOT bidirectional telemetry uses the same wire as the motor signal; there's no extra cable. The autopilot writes an ESC log message per motor at telemetry rate, which uses log storage but not enough to matter. The diagnostic upside is large — direct RPM measurement instead of inference.

What's the difference between ESC desync and motor failure?

ESC desync is the speed controller losing its sensorless commutation lock; the motor itself is fine, but the ESC stops driving it correctly. Motor failure is the motor itself becoming unable to spin — bearing seizure, winding short, or magnet detachment. Both produce the same RCOU saturation, but ESC desync often recovers if you cycle throttle, while a failed motor doesn't.

Diagnosing ESC Desync Events in ArduPilot RCOU Logs

TL;DR: ESC desync events on ArduPilot Copter trigger the firmware’s Thrust Loss Check, which writes ERR Subsys=25 ECode=1 plus a MSG line reading "Potential Thrust Loss (N)" with the suspected motor number. The trigger criteria are specific: the vehicle has been descending for at least 1 second, throttle is above 90%, and attitude error is within 15 degrees. When this fires, plot RCOU.C1–C4: the desyncing motor is the one saturated at maximum while the autopilot tries to compensate.

What ArduPilot writes when thrust is lost

The thrust loss check runs continuously while armed and not in ap.land_complete state. It increments an internal counter every loop iteration that meets all three criteria:

Descending despite high throttle. The autopilot has been commanding more than 90% throttle output for the last second but altitude is dropping.
Attitude under control. Roll and pitch error is within 15° (so the descent isn’t just because the airframe is tipped over).
One full second of the above. Brief transients don’t qualify.

When the counter trips, ArduPilot logs ERR Subsys=25 ECode=1 (THRUST_LOSS_CHECK, FAILSAFE_OCCURRED), emits the MSG line naming the suspected motor index, and engages the motor library’s thrust-boost compensation. The compensation uses the remaining motors to maintain altitude with the lost motor saturated.

The RCOU pattern that names the failed motor

Without ESC telemetry, the autopilot can’t directly measure RPM — it infers thrust loss from the attitude controller’s behaviour. Plot RCOU.C1, RCOU.C2, RCOU.C3, RCOU.C4 on one chart:

The desyncing motor will be saturated at SERVOn_MAX (typically 2000 µs).
The motor diagonally opposite it will be at or near SERVOn_MIN (~1000 µs) — the autopilot is throttling it down to compensate for the yaw the saturated motor creates.
The other two motors will sit between hover and saturation.

That diagonal-imbalance pattern is the firmware-confirmed signature of motor failure on a quadcopter. On hex and octocopter frames, the pattern shifts — the affected motor and its symmetric counterpart show the saturation.

Confirming it in Mission Planner

Open the log, plot RCOU.C1–C4, and look for the saturation pattern at the timestamp of the ERR Subsys=25 line. Cross-reference against ATT.Roll, ATT.Pitch, RATE.R, RATE.P to confirm the attitude controller was working against the unbalanced thrust.

Confirming it in MAVExplorer

MAV> graph RCOU.C1 RCOU.C2 RCOU.C3 RCOU.C4
MAV> graph RATE.RDes RATE.R RATE.PDes RATE.P
MAV> messages ERR MSG

The MSG stream will contain the "Potential Thrust Loss" line naming the motor. The frame layout (see FRAME_CLASS and FRAME_TYPE) tells you which physical motor that number refers to on the airframe.

With ESC telemetry, the picture is clearer

If your ESCs report telemetry over DSHOT bidirectional or a separate UART, ArduPilot logs an ESC message family with RPM and ESC temperature per motor. The desyncing motor shows RPM crashing to zero or oscillating while RCOU.Cn is still commanding high — conclusive evidence that the ESC is no longer driving the motor at the commanded duty cycle.

Enabling ESC telemetry is worth the half-hour of setup for any operator running a fleet — the post-incident analysis goes from inference to direct measurement.

Why ESCs desync — ranked by what we see most often

High-throttle commutation failure. The ESC’s sensorless commutation algorithm loses track of the rotor position at high RPM, especially after a sudden throttle change. BLHeli_S ESCs with older firmware are particularly prone. Flash to current BLHeli_32 or AM32 firmware and the rate drops sharply.
Voltage sag below the ESC’s startup threshold. A battery dipping under load below the ESC’s minimum operating voltage can cause one ESC to drop and not recover. Coincident with BAT.Volt dips. See our power module post.
Mechanical bind. A bearing failure or debris in the motor briefly stalls the rotor; the ESC sees current spike and loses sync. Often correlates with a sudden VIBE spike on one axis.
Prop strike. Impact damage to a propeller unbalances the rotor; the ESC’s sync timing tolerates much less imbalance than the airframe’s vibration tolerance.
ESC firmware bug. Rare but real. Update to the latest firmware for your ESC family before chasing harder explanations.

Preventing it

Update ESC firmware — current BLHeli_32, AM32, or the manufacturer’s latest. Many desync events are firmware-fixable bugs.
Calibrate ESC range using Mission Planner’s ESC calibration sequence. An ESC seeing a 1500-1700 µs range will desync differently than one calibrated for 1000-2000 µs.
Set MOT_PWM_MIN and MOT_PWM_MAX carefully — matching the ESC’s expected range.
Avoid prop strikes — trivial sounding, but a chipped prop is the most common precursor.
Enable ESC telemetry for ongoing visibility — you catch a developing bearing problem before it becomes a flight failure.
Soft-mount ESCs away from the main harmonic of the airframe — reduces sustained vibration to the ESC PCB.

When this isn’t actually ESC desync

Motor unplugged or wire broken. RCOU saturates the same way, but ESC telemetry (if enabled) shows zero RPM with zero current. The cause is electrical, not the ESC’s sync algorithm.
Bearing failure. The motor still spins but with high friction; RPM stays low while ESC current climbs. Distinguishable from desync only with telemetry.
Wrong FRAME_CLASS. Mixing assignments are off; the autopilot is commanding the wrong motors. RCOU patterns look chaotic from the start, not after a clean takeoff. See our takeoff-flip post.
Voltage failsafe kicked in and reduced power to all ESCs. RCOU.C1–C4 all drop together rather than one saturating.

When LogHat helps — and when it doesn’t

LogHat flags every ERR Subsys=25 event and highlights the corresponding RCOU saturation pattern in the 3D replay, with the motor number annotated. When ESC telemetry is present, we also surface the RPM-versus-RCOU trace for each motor so you can see the sync failure directly. What we can’t tell you is whether the ESC’s firmware version is the latest — that’s information the autopilot doesn’t capture. Check it manually before assuming the problem is hardware.