Overview of Sensorless Vector Control for Permanent Magnet Synchronous Motors: PMSM_FOC_sensorless Method – EMF + PLL
Preface:
Sensorless control strategies are one of the key research areas in electric motor control. This article is the first in a series focusing on sensorless techniques for Permanent Magnet Synchronous Motors (PMSM). Each article will analyze the proposed methods and strategies from both theoretical and verification perspectives.
I. Research Background of Sensorless Control for Permanent Magnet Motors
Research into sensorless control mainly aims to address limitations due to external constraints and cost requirements. The main methods are divided into high-frequency signal injection and model-based approaches. Unlike induction motors, Permanent Magnet Synchronous Motors (PMSM) exhibit characteristics that vary across different speed ranges, requiring different estimation methods. In these applications, a reliable bldc motor driver or brushless dc motor driver—or even a 3 phase brushless motor driver—is often essential.
Figure 1.1 Classification of Sensorless Control Techniques for Permanent Magnet Synchronous Motors
Besides the methods mentioned above, there are also single-observer sensorless control strategies covering the full-speed domain, such as the nonlinear flux observer and dq-axis current estimation.
II. Overview of Sensorless Control Strategies for Permanent Magnet Synchronous Motors
2.2 EMF‑Based Observer Method
2.2.1 Understanding the EMF Observer
Figure 2.2.1 Control Block Diagram of the Estimation Method
The estimation of speed and rotor position using back electromotive force (EMF) and a Phase-Locked Loop (PLL) consists of two steps:
- Acquire back EMF
- Use PLL to compute motor speed and rotor position
The algorithm extracts physical quantities related to speed—such as voltage, current, flux linkage, and back EMF—from which it derives the speed signal. Estimation accuracy depends on the signal-to-noise ratio of the back EMF; higher speed leads to larger EMF amplitude and better estimation. However, at zero or low speed, the signal-to-noise ratio is very low, and external interference prevents obtaining useful data. Thus, model-based methods are unsuitable for zero or low-speed regions.
The open-loop algorithm mainly refers to the EMF integration method, based on the motor’s electromagnetic relationship. It estimates speed from real-time stator current and voltage measurements. Since it doesn’t use iterative calculations, each computation is independent, ensuring rapid dynamic response and improving the control system’s bandwidth. Such implementation is relevant for an industrial dc motor controller or sensorless bldc controller in simpler applications. However, its major limitation is susceptibility to external disturbances and parameter variations. Therefore, it demands high accuracy and is often coupled with error correction and parameter identification to mitigate these issues.
In summary, open-loop algorithms are hard to apply in high-precision servo drives—they are more suitable for applications with minimal disturbance and lax precision requirements, such as basic bldc foc controller applications.
2.2.2 EMF Mathematical Models for Interior and Surface-Mounted PMSMs
2.2.2.1 EMF Mathematical Model
There are various ways to obtain EMF, typically via observers (such as sliding-mode or Luenberger types). To facilitate analyzing back EMF, we present the motor’s mathematical model in the stationary reference frame. These models align with designs for brushless dc motor controller or integrated 3 phase bldc motor controller systems.
2.2.2.2 LPF‑Enhanced EMF Observer
Traditional EMF observers produce waveforms with distortion and weak sinusoidal qualities. Since accurate EMF estimation directly affects speed and position estimation, improvements—such as those found in a sensorless bldc controller enhanced with low-pass filters—are necessary.
2.2.3 Simulation Verification for Interior and Surface-Mounted PMSMs
Figure 2.2.3 Back EMF waveform using traditional differential-equation calculation
As shown in Figure 2.2.3, the back EMF waveform exhibits distortion and poor sinusoidal quality. This directly impacts the precision of speed and position estimation, necessitating method improvement.
Figure 2.2.4 Back EMF waveform estimated with low-pass filter–based observer
As seen in Figure 2.2.4, the waveform estimated via the filter-based extended EMF observer is highly sinusoidal.
- Traditional extended EMF observer method
(b) Filter-enhanced extended EMF observer method
Figure 2.2.5 Rotational speed estimation tracking
From Figure 2.2.5, it’s evident that the filter‑enhanced observer yields low speed estimation error and minimal steady‑state error, which is what high-performance industrial dc motor controller designs aim to achieve.
III. Conclusion and Discussion
A standalone EMF observer cannot independently estimate rotor speed and position across the full speed range of a PMSM. Because EMF amplitude is proportional to signal-to-noise ratio, accurate EMF cannot be obtained at zero or low speed, preventing further estimation of rotor information.
To achieve full-speed domain rotor speed estimation, the commonly used strategy is IF + SMO, which will be introduced and validated in future articles.