Dieser Leitfaden ist derzeit auf Englisch verfügbar.
Technology / Technical guide
Yaw Feedback: Setpoint, Measurement & Closed-Loop Error
In a conceptual yaw feedback loop, a requested orientation is compared with an observation. The controller acts on their difference, and the mechanism changes the physical orientation. This chain explains why a motion command, a measured value and the true physical state are different quantities.
The controller reacts to its information, not directly to truth.
A closed loop needs both a defined reference and a meaningful observation; naming feedback does not make either perfect.
01 / Loop boundary
Feedback returns information about an output.
A simple negative-feedback concept compares reference r with measured output y_m, forming error e = r − y_m under a declared sign convention.
A simple negative-feedback concept compares reference r with measured output y_m, forming error e = r − y_m under a declared sign convention. A controller maps that error into an actuator request. The physical system then responds and generates another observation.
MathWorks illustrates this general feedback connection. A motor-drive request and the regulated nacelle heading describe different units and dynamics.
02 / Angle meaning
The compared angles need a common reference.
A wind direction, motor angle and nacelle heading can describe different axes and frames.
A wind direction, motor angle and nacelle heading can describe different axes and frames. Even two heading values need compatible zero and sign conventions. Angular error also needs circular treatment rather than unrestricted subtraction across the north boundary.
The angle-convention guide covers that geometry. A target may be derived from a wind observation, but its derivation is separate from observing whether the nacelle reached the target. One signal should not be assumed to perform both roles.
03 / Biased feedback
Zero indicated error need not mean zero physical error.
Hypothetical scalar example: suppose the requested heading is 25°, the physical heading is 20° and a measurement has a +5° bias.
Hypothetical scalar example: suppose the requested heading is 25°, the physical heading is 20° and a measurement has a +5° bias. The reported 25° equals the request, so the conceptual controller sees zero error even though physical heading differs by 5°.
This invented example is not a yaw diagnosis. A loop can regulate its measured variable while physical interpretation remains biased; repeatability cannot establish accuracy.
04 / Dynamic response
A request and the response occur through different stages.
The mechanism can have inertia, friction, clearance and rate constraints; the observation can have delay and noise.
The mechanism can have inertia, friction, clearance and rate constraints; the observation can have delay and noise. As a result, a request change need not produce an immediate equal change in heading. An estimate based only on requested movement has a different evidence basis from a position observation.
The backlash guide distinguishes mechanical clearance from decision logic. A measured difference can arise from several stages, so it is not a unique diagnosis of one mechanism.
05 / Feedback limits
Negative feedback is a structure, not a stability guarantee.
The loop response depends on gain, dynamics, delay and constraints.
The loop response depends on gain, dynamics, delay and constraints. Feedback may reduce some disturbances while transmitting measurement noise or amplifying variation under other conditions. The word closed-loop therefore does not independently prove stable, fast or accurate behaviour.
A controller may also operate with several distinct references or observation paths. Comparing a requested signal with a displayed value only has meaning when their physical roles are known. The PI/PID guide explains one general controller structure.
06 / Product scope
Keep the conceptual loop separate from installed hardware.
All five current MARSTON models have electric yaw.
All five current MARSTON models have electric yaw. That fact does not establish a specific heading sensor, encoder, loop architecture, feedback accuracy or control algorithm. The concepts here explain why those implementation details cannot be inferred from an actuator photograph.
This page provides no yaw command, sensor substitution, sign change or controller-tuning procedure. Its example is an information-flow model, and achieved behaviour remains a separate equipment-specific question.
Continue the review with PI & PID Control: Proportional, Integral & Derivative Roles.
Continue the review with Wind Measurement Filtering: Bandwidth, Delay & Latency.
Continue the review with Redundant Sensors: Voting, Agreement & Common Causes.
General background: MathWorks: General negative-feedback model connections. These references explain industry concepts; they are separate from the applicable MARSTON product and project documents.
Key concepts and conditions.
Read each quantity with its reference, units and stated conditions. A general relation does not establish a model-specific value.
| Concept / quantity | Meaning / relation | Conditions / limits |
|---|---|---|
| Reference | Requested output r with defined units | Same physical quantity as the compared output. |
| Measured output | Observation y_m and its interpretation | Bias and uncertainty remain relevant. |
| Error | r − y_m under the stated convention | Angles also need common frames and wrapping. |
| Actuator request | Controller output u | Not identical to achieved physical movement. |
| Loop response | Dynamics, delay and constraints | Negative feedback alone does not establish stability. |
Engineering questions.
Is a yaw command an actual heading measurement?
No. A request and an observation have different roles.
Does zero measured error prove correct physical heading?
Not if the feedback is biased or represents a different quantity.
Can motor angle be compared directly with wind direction?
Only through an established mapping and compatible reference conventions.
Does closed-loop mean automatically stable?
No. Dynamics, gain, delay and constraints affect the response.
Are all MARSTON heading sensors specified here?
No installed feedback sensor or architecture is assigned.
Does the 5° example describe a real product defect?
No. It is invented arithmetic illustrating measurement bias.
Technical enquiry
Discuss the technical question.
In a conceptual yaw feedback loop, a requested orientation is compared with an observation. Share the model and applicable operating or design information for an engineering discussion.