Throttle closes itself under load
On a torque-structured ECU the electronic throttle is closed by the torque monitor when modelled torque exceeds a calibrated ceiling — not a mechanical fault.
Typically: Turbo and naturally aspirated petrol with electronic throttle; torque-structured OEM ECUs
What the workshop sees
Power flattens or drops part-way through a pull even though the accelerator is held flat. On the log, throttle plate angle steps closed while pedal position stays at 100%. It usually appears once the build makes more torque than stock, and repeats at the same load or RPM point.
What is actually happening
Modern OEM engine management is torque-structured: the pedal is a torque request, the ECU converts that to an air/throttle target, and a separate torque-monitoring layer continually checks modelled torque against permitted limits. When a higher-output build pushes modelled torque past one of those ceilings — the driver/road limit, a per-gear limit, or a component-protection limit — the monitor intervenes by closing the throttle to bring torque back inside the model.
The key point is that nothing is mechanically wrong: the throttle is doing exactly what the calibration tells it to. The stock torque model simply no longer describes the engine.
How it is addressed
The fix is to make the torque model and its ceilings describe the actual engine, not the stock one — raising the relevant maximum-torque limits (driver/road, per-gear, and any clutch or component limits that apply) to suit the build, and keeping the modelled-torque calculation consistent with real output so the monitor and the air path agree.
This is calibration work, not a switch: the limits exist for a reason, and on a torque-by-gear platform every gear's ceiling has to be addressed or the intervention just moves up the box.
Verify against logged pedal position, requested vs actual (or modelled) torque, throttle plate angle, and the relevant limiter flags. The calibration is right when the throttle tracks the request across the full load and RPM range with no commanded closure, confirmed on more than one pull.
Questions workshops ask
Is a throttle that closes itself a hardware fault?
Usually not. On a torque-structured ECU it is the torque-monitoring system commanding the throttle shut because modelled torque has exceeded a calibrated limit. The throttle body and pedal are doing what the calibration asks.
Why did it only start after the build made more power?
The stock torque ceilings were sized for the stock engine. Once the build exceeds them the monitor begins intervening, so it appears as a new fault even though the calibration is unchanged.
Can you just remove the torque limits?
They are raised to suit the build and kept consistent with the torque model and any component limits — they are engineering boundaries, not switches. On per-gear platforms every gear has to be addressed.
Where this fits
Approved workshops can submit the file and logs and we will develop the calibration with you. Relevant services: ECU tuning files · Workshop calibration guide. See validated results in the tune archive.
Stuck on this on a live build?
Upload the stock read and your logs — we will work it until the data is clean.
