SPRUHJ1 User guide

SPRUHJ1I January 2013 – October 2021 TMS320F2802-Q1 , TMS320F28026-Q1 , TMS320F28026F , TMS320F28027-Q1 , TMS320F28027F , TMS320F28027F-Q1 , TMS320F28052-Q1 , TMS320F28052F , TMS320F28052F-Q1 , TMS320F28052M , TMS320F28052M-Q1 , TMS320F28054-Q1 , TMS320F28054F , TMS320F28054F-Q1 , TMS320F28054M , TMS320F28054M-Q1 , TMS320F2806-Q1 , TMS320F28062-Q1 , TMS320F28062F , TMS320F28062F-Q1 , TMS320F28068F , TMS320F28068M , TMS320F28069-Q1 , TMS320F28069F , TMS320F28069F-Q1 , TMS320F28069M , TMS320F28069M-Q1

1.1.5.1 FAST Estimator Replaces Mechanical Sensor

Field Oriented Control (FOC) of an electric motor results in superior torque control, lower torque ripple, and in many cases, improved efficiency compared to traditional AC control techniques. For best dynamic response, rotor flux referenced control algorithms are preferred to stator flux referenced techniques. To function correctly, these systems need to know the spacial angle of the rotor flux with respect to a fixed point on the stator frame (typically the magnetic axis of the phase A stator coil). This has traditionally been accomplished by a mechanical sensor (for example, encoder or resolver) mounted to the shaft of the motor. These sensors provide excellent angle feedback, but inflict a heavy toll on the system design.

There are six major system impacts resulting from sensored angle feedback, illustrated in Figure 2-4:

The sensor itself is very expensive (often over $2500 for a good resolver and several dollars for high volume integrated encoders)
The installation of the sensor requires skilled assembly, which increases labor costs
The sensor often requires separate power supplies, which increases system costs and reduces reliability
The sensor is the most delicate component of the system, which impacts system reliability, especially in harsh real-world applications
The sensor feedback signals are brought back to the controller board via connectors, which also increases system costs and can significantly reduce reliability, depending on the type of connector
The cabling required to bring the sensor signals back to the controller creates multiple challenges for the system designer:
- Additional costs for the cable, especially if there is a substantial distance between the motor and controller
- Susceptibility to sources of noise, which requires adding expense to the cable with special shielding or twisted pairs
- The sensor and associated cabling must be earth grounded for safety reasons. This often adds additional cost to isolate these signals, especially if the processor which processes the sensor signals is not earth grounded

Figure 1-4 Sensored FOC System

In some applications where the motor is enclosed (for example, compressors), a sensored solution is impractical due to the cost of getting the feedback wires through the casing. For these reasons, designers of FOC systems are highly motivated to eliminate the sensor altogether, and obtain the rotor flux angle information by processing signals which are already available on the controller circuit board. For synchronous machines, most techniques involve executing software models of the motor being controlled to estimate the back-EMF waveforms (rotor flux), and then processing these sensed waveforms to extract an estimation of the rotor shaft angle, and a derivation of its speed. For asynchronous machines the process is a bit more complicated, as this software model (observer) must also account for the slip which exists between the rotor and rotor flux.

However, in both cases, performance suffers at lower speeds due to the amplitude of the back-EMF waveforms being directly proportional to the speed of the motor (assuming no flux weakening). As the back-EMF amplitude sinks into the noise floor, or if the ADC resolution cannot faithfully reproduce the small back-EMF signal, the angle estimation falls apart, and the motor drive performance suffers.

To solve the low-speed challenge, techniques have been created that rely on high frequency injection to measure the magnetic irregularities as a function of angle (that is, magnetic saliency) to allow accurate angle reconstruction down to zero speed. However, this introduces another set of control problems. First, the saliency signal is non-existent for asynchronous motors and very small for most synchronous machines (especially those with surface mount rotor magnets). For the motors that do exhibit a strong saliency signal (for example, IPM motors), the signal often shifts with respect to the rotor angle as a function of loading, which must be compensated. Finally, this angle measurement technique only works at lower speeds where the fundamental motor frequency does not interfere with the interrogation frequency. The control system has to create a mixed-control strategy, using high-frequency injection tracking at low speed, then move into Back-EMF based observers at nominal and high speeds.

With any technique, the process of producing a stable software sensor is also extremely challenging, as this motor model (observer) is essentially its own control system that needs to be tuned per motor across the range of use. This tuning must be done with a stable forward control loop. Needed is a stable torque (and usually speed) loop to tune the observer, but how do you pre-tune your forward control without a functioning observer? One option is to use a mechanical sensor for feedback to create stable current and speed loops, and then tune your software sensor in parallel to the mechanical sensor. However, the use of a mechanical sensor is often not practical. This problem has delayed market use of software sensors for sensorless FOC control.

In summary, these existing solutions all suffer from various maladies including:

Poor low speed performance (back-EMF and SMO)
Poor high speed performance (saliency observers)
Poor dynamic response
Calculation intensive (multi-modal observers)
Parameter sensitivity
Requirement for observer tuning

The most recent innovation in the evolution of sensorless control is InstaSPIN-FOC. Available as a C-callable library embedded in on-chip ROM on several TI processors, InstaSPIN-FOC was created to solve all of these challenges, and more. It reduces system cost and development time, while improving performance of three-phase variable speed motor systems. This is achieved primarily through the replacement of mechanical sensors with the proprietary FAST estimator. FAST is an estimator that:

Works efficiently with all three phase motors, taking into account the differences between synchronous/asynchronous, salient/non-salient, and permanent/non-permanent/induced magnets
Dramatically improves performance and stability across the entire operating frequency and load range for a variety of applications
Removes the manual tuning challenge of traditional FOC systems:
- observers and estimators, completely removes required tuning
- current loop regulators, dramatically reduces required tuning
Eliminates or reduces motor parameter variation effects
Automatically designs a stable and functional control system for most motors in under two minutes.