Sensorless Starting of Wound Rotor Synchronous

Machines
Maurice Fadel, Stéphane Caux, Giovanni Borghetti, Gianluca Postiglione

To cite this version:

Maurice Fadel, Stéphane Caux, Giovanni Borghetti, Gianluca Postiglione. Sensorless Starting of
Wound Rotor Synchronous Machines. 2019 21st European Conference on Power Electronics and
Applications (EPE ’19 ECCE Europe), Sep 2019, Gênes, Italy. �10.23919/EPE.2019.8915211�. �hal-
02509508�

HAL Id: hal-02509508

https://hal.science/hal-02509508v1
Submitted on 7 Feb 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice

Sensorless Starting of Wound Rotor Synchronous Machines

Maurice Fadel1, Stephane Caux1, Giovani Borghetti2 and Gianluca Postiglione2

1
LAPLACE, Université de Toulouse, CNRS, INPT, UPS, France
[email protected], [email protected]
Tel: +33 (0)5 34 32 24 05, Fax: +33 (0)5 61 63 88 75
URL: http://www.laplace.univ-tlse.fr/

NIDEC ASI S.p.A. Corso Perrone, 11 I – 16152 Genoa, Italy

2

[email protected], [email protected]
URL: www.nidec-industrial.com

Keywords – Sensorless control, Estimation technique, Electrical drive, Synchronous motor.

Abstract – The sensorless operation of wound rotor synchronous machines (WRSM) must meet
several requirements. The most important to be mentioned is how to insure starting at full load
and then operating at low speed in the presence of magnetic saturation. This work presents a
solution for high power machines by combining a method of rotor location by high frequency
injection and then starting it with a Phase Locked Loop (PLL) type solution. It should be noted
that the location of the rotor leaves an indeterminacy of  on the electrical position that needs to
be resolved. Initial experimental tests for rotor location confirm simulations and show how the
scheduling of the two methods can solve the problem.

INTRODUCTION
Wound rotor synchronous machines (WRSM) are synchronous machines (SM), in which the rotor
flux is generated by an electric winding. WRSMs have certain characteristics that meet the needs of
large industrial applications such as compressors or rolling mills very well. When using vector
control, position information is required, which can be obtained using a sensor (resolver, optical
encoder, etc.). The absence of this sensor reduces costs and can increase the reliability of the entire
drive system. Sensorless controls are currently the subject of many studies [3], [4], [5], [6]. From
start-up to nominal speed or higher, it is very often necessary to use several types of observers to
benefit from the best performance depending on the speed. The first problem is the start-up and the
acceleration which is the subject of our contribution. As soon as the machine has started and the
electromotive forces become high, it is easier to rebuild the position. Many methods have already
proven their worth.
First, the current trend is to use a method of injecting a high frequency signal to locate the rotor[x].
Conventional approaches for permanent magnet synchronous machines (PMSM) inject high
frequency signal (HFSI) into the stator dq axes and by measuring currents in the same reference dq,
the position is reconstructed [2], [8], [4]. Unlike PMSMs, WRSMs theoretically allow the rotor
winding to be used as an injector of the HFSI and it is even possible to imagine the following
solutions [7]:
- stator injection and stator measurement
- injection into the rotor and measurement on the stator
- injection into the stator and measurement in the rotor.

The most intuitive approach of the methods mentioned above is to emulate a resolver using the rotor
as the transmitter and the stator as the receiver of the HF signal. However, if a rotating diode rectifier
is used for rotor excitation, injecting a signal into the rotor can become difficult. If the stator is used
as the transmitter and receiver of the HFSI, the rotor winding acts as a damping winding for the
HFSI and the magnitude required for observation may therefore no longer be sufficiently present.
In this article, we discuss the case of stator injection and stator current measurement when the
machine is equipped with a diode rectifier.
In this context it is well known that the position is known with an indeterminate electric  angle.
Several solutions exist to overcome this indeterminacy and here we propose a new method to address
this problem.
The result of this procedure is used to initialize the machine start as well as a PLL procedure to

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.1
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice
ensure the machine start up. We will analyze the benefits using the two methods in a specific
sequencing on position estimation.

MACHINE MODELING
In this case we can establish two types of models, first of all a low-frequency model such as the Park
type reflecting the traditional operation of the machine and a high-frequency model allowing to
analyze the propagation of the injected high-frequency signal.

Fig. 1b WRSM Structure

Fig. 1a WRSM in high power application Parameters of machine:

P=13.8 kW; Van=6.6 kV, Ian=1200 A,
Fn=60 Hz, rated speed=1800 rpm, two pole pairs

This type of machine (Figure 1a, 1b) actually consists of 2 machines running at the same speed
(Figure 1.b). First of all, an wounded-rotor induction machine (WRIM), acting like a “rotating
transformer”, is supplied by a three-phase VSI or by a thyristor converter; the rotor, seen as the
secondary side of the rotating transformer, feeds a rotating diode bridge which, in turn, allows finally
to energize the main machine (MP). This “brushless exciter” structure is mandatory in Oil & Gas
application, in which the brushes and the related sparks are forbidden due to safety reasons.
Llf

Lsq Llrq

Lsd Llrd Rf ω.Φsd

If I sq I rq
ω.Φsq
I sd I rd
Rs Imq Rrq
Rs Imd Rrd Vf
Φf
Φ sq
Φsd Vsq Lmq
Vsd Lmd
Φrd
Φ rq

Fig. 2a: Electric model on the d axis Fig.2b:Electric model on the q axis

Thus, by assuming an unsaturated functioning, the Park model can be written. This is how Park's
model can be written.

 d
Vsd  Rs .I sd  . sd  . sq
dt

V d
 Rs .I sq  . sq  . sd  sd   Lsd  Lmd 0 Lmd  Lmd 0   I sd  (1)
 sq dt    0 Lsq  Lmq 0 0  Lmq  I 
  sq     sq 
 d   f    Lmd . I f 
 V f  R f .I f  . f with 0 L f  Lmd  Lmd 0
 dt      
 d  rd   Lmd 0 Lmd ( Lrd  Lmd ) 0   I rd 
 0   Rrd .I rd  . rd   rq   0 Lmq 0 0 ( Lrq  Lmq )   I rq 
dt   

 d
0   Rrq .I rq  . rq
 dt

With Tem  p.   sd .I sq   sq .I sd  (2)

For high frequency the equivalent dq circuit of the machine can be summarized as follows [1].
Lls Lls

Rdeff Rqeff
Vdh Lad LlD L lf Laq LlQ
Vqh

Figure 3 a: Equivalent circuit on the d-axis for HF operation Figure 3 b: Equivalent circuit of q-axis for HF operation

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.2
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice

These model show new values such as effective resistances R , Rqeff , effective inductances Ldeff , Lqeff and
deff

leakage inductances LlD , LlQ , L 'lf

which play an important role because of their low value compared to
magnetizing inductances. In addition, saturation plays a major role, in particular by also reducing
the value of magnetizing inductances. The parameters of these models are linked to the parameters
of the low frequency model and also depend on the saturation state of the machine.

SENSORLESS CONTROL OF THE MACHINE

Sensorless operation is compatible with vector control and consists of replacing the position
measurement with a device that reconstructs the position from some measurements on the dq axis
and after digital signal processing, fig. 4.

Figure 4: Control diagram without position sensor

Note that the estimated speed is obtained from the estimated position either by direct filtering or by
a looped device. For machines with smooth poles the current in the d-axis is imposed at zero. On the
other hand, for machines with salient poles an additional torque component can be obtained by
imposing a negative id current in relation to the torque level in order to operate in Maximum Torque
per Amp MTPA [11].

HIGH FREQUENCY INJECTION

Injection-based methods have been known for many years [10]. In our case, the injection is done on
the d-axis in order to limit the impact on the machine's torque. The signal demodulation is performed
using a specific processing illustrated in Figure 6 and uses a PI controller to cancel the DC
component that contains information on the position deviation. This method can be used in
stationary state to locate the rotor or at low speed. In the latter case, it is important to properly design
the filters so that the rotation speed does not disturb the operation.

Figure 6: Position extraction by demodulation

The position estimation converges towards a point of equilibrium with a phenomenon of polarity which
results in an indeterminacy of the position of a value equal to  electric. It should be noted that the
response time varies with the actual position, reflecting non-linear behavior.
It should be noted that this method operates in a closed loop with a bandwidth defined by the regulator.

USING THE ATAN2 FUNCTION

Another method is to measure the current components on the  and  axes and estimate the position
using an ATAN2 function, Figure 7. This method operates in an open loop and is very sensitive to noise.

Figure 7a: Position extraction with ATAN2 function Figure 7b: Evolution of the excitation current

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.3
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice

While applying the HFSI voltage, a controlled variation of the excitation current is also given. It is
possible to apply a voltage step to the excitation circuit, or switch off the exciter, or control a descending
ramp. The Figures 8 illustrate the reconstruction of the stationary position with the 2 methods, with
different initial mechanical condition m (0) . Note that the machine has 2 pairs of poles. The curves show
that the procedure based on the ATAN2 function ensures the location of the rotor independently of the
initial conditions while the closed loop method using a PI controller without procedure correction shows
an indeterminacy of the location of the rotor.

Figure 8a: Initial condition on the mechanical position m (0)   / 6 Figure 8b: Initial condition on the mechanical position m (0)   / 3

HYBRID METHOD FOR INITIAL POSITION DETECTION

The experimental results show that the ATAN2 (open loop) calculation is affected by noise, which
results in an uncertainty band of some electrical degrees. On the contrary, the HFSI method provides a
clean measure, although affected by a possible polarity error. This suggests to combine the methods: the
ATAN2 method is used only to understand the polarity and, in case, add π to the position calculated by
the HF method.

Figure 9: Experimental results - excitation current cut-off.

The validation tests (see figure 9) have been carried out by applying the proposed method on the
aforementioned machine, and comparing the results with the measurement of an absolute encoder. A
maximum deviation of ± 2 mechanical degrees has been recorded. In any case, after each test, the
machine has been correctly started without any problem.
The result of the hybrid detection method can thus be used as a good point for starting the machine
without the absolute encoder, and for initializing the PLL in case of full sensorless application.

OBSERVATION WITH PLL

The construction of PLL-based observers is an interesting method that can work as soon as the speed is
different from zero.

Figure 10 : Position observation with PLL

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.4
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice
The idea of this solution is to converge flux on the axes  and  calculated with 2 different models.
The convergence of these two quantities implies that the estimated position converges towards
the real position of the machine recovering the initial error of rotor position. Convergence is
provided by a PI regulator, Fig. 10. This loop is a non-linear system and the convergence dynamics
of the error is a function of the speed, so it is important to adapt the gain of the PI controller
according to speed.

SIMULATION RESULTS
The experimental device we wish to test in simulation includes an incremental encoder to determine
the rotational speed. In the following tests, we propose to compare the performance obtained using
this sensor with the one obtained from the observer.
In addition, we try to evaluate the robustness of our procedure by injecting an initial position error
of 3° mechanical or 6° electrical. The PLL method is evaluated during a start without load and then
for a loaded start (parabolic evolution according Tr=36632 x (ω/1800)2. The reference speeds is first
low (21.6 rpm) and then higher (432 rpm).

NO LOAD – SPEED REFERENCE =21.6 rpm

Speed Reference and Feedback [rpm] Speed Reference and Feedback [rpm]
30 30

25 25

20 20

Speed Ref. Speed Ref.

15 15
Speed Fdb. from Enc. Speed Fdb. from Enc.
Speed Fdb. from PLL Speed Fdb. from PLL

10 10

5 5
Motor starts backwards
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time [s] Time [s]

Figure 11 a: Initial position error 0° Figure 11 b: Initial position error 3°mechanical

Figures 11 show the use of the speed output of the PLL gives lower speed oscillations than for
operation with the incremental sensor, especially when the initial position error is large. It should be
noted that at low reference speed the possibility of a reverse start exists when the position error is
not zero.

NO LOAD – SPEED REFERENCE =432 rpm

Speed Reference and Feedback [rpm] Speed Reference and Feedback [rpm]
500 500

450 450

400 400

350 350

300 300
Speed Ref. Speed Ref.
250 Speed Fdb. from Enc. 250 Speed Fdb. from Enc.
Speed Fdb. from PLL Speed Fdb. from PLL
200 200

150 150

100 100

50 50

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [s] Time [s]

Figure 12 a: Initial position error 0° Figure 12 b : Initial position error 3°mechanical

We notice in Figures 12 that for higher reference speeds the risk of starting in the opposite direction
disappears.

WITH LOAD – SPEED REFERENCE =2160 rpm

The target speed profile consists of a ramp from zero to 2160 rpm in 3s. In steady state the torque is
52750 Nm, corresponding to a power of 11.9 MW.
For a start with quadratic load (Figures 13) we find that the PLL-based observer works correctly

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.5
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)
Sensorless Starting of Wound Rotor Synchronous Machines FADEL Maurice
even in the presence of an initial position error. The incremental encoder does not bring any
additional advantage.

Speed Reference and Feedback [rpm] Speed Reference and Feedback [rpm]

2000 2000

1500 1500

Speed Ref. Speed Ref.

Speed Fdb. from Enc. Speed Fdb. from Enc.
1000 Speed Fdb. from PLL 1000 Speed Fdb. from PLL

500 500

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [s] Time [s]
Figure 13a: Initial position error 0°
Figure 13b: Initial position error 3°mechanical
CONCLUSION

In this work we have shown that it is necessary to know the position of the rotor well to ensure a correct
start of the WRSM machine, especially when the load is high. The solution we have studied is
satisfactory and based on 3 specific steps in succession. First of all, a localization of the rotor by
injecting a high frequency signal into the stator, which makes it possible to know the position with an
indeterminacy of . Then by acting on the excitation current we lift this indeterminacy and the position
of the rotor is then known but the mechanical accuracy is not high. Combining the HF and arctangent-
type methods, experimental results have been proven that is possible to get a good accuracy and speed.
The result obtained can be used as an initial condition for a closed-loop observation procedure based
on PLL. This procedure gives satisfactory results and the incremental speed encoder is no longer
necessary.

REFERENCES
[1] Alexander Rambetius, and Bernhard Piepenbreier., "Comparison of carrier signal based approaches for
sensorless wound rotor synchronous machines," International Symposium on Power Electronics, Electrical
Drives, Automation and Motion, 2014
[2] Beciu A., E. Godoy, P. Rodriguez-Ayerbe, I. Bahri, A. Maalouf, " High Frequency Impedance Analysis for
Sensorless Starting of Wound Rotor Synchronous Machines,". IFAC World Congress, 2016.
[3] Tian-Hua Liu, Shao-Kai Tseng, Ting-Wei Lin, and Jui-Ling Chen, “Sensorless IPMSM Position Control
System Using a High Frequency Injection Method” 2016 IEEE 2nd Annual Southern Power Electronics
Conference (SPEC)
[4] Oliver Benjak, Dieter Gerling “Review of Position Estimation Methods for PMSM Drives without a Position
Sensor, Part III: Methods based on Saliency and Signal Injection”, 2010 International Conference on Electrical
Machines and Systems, Oct. 10-13, 2010, Incheon, Korea
[5] Matti Eskola, Heikki Tuusa, “Sensorless Control of Salient Pole PMSM Using a Low-Frequency Signal
Injection” EPE 2005 – Dresden
[6] Alfio Consoli, Giuseppe Scarcella, and Antonio Testa, Member, “Industry Application of Zero-Speed
Sensorless Control Techniques for PM Synchronous Motors”, IEEE transactions on industry applications, vol.
37, no. 2, march/april 2001
[7] Alexander Rambetius, and Bernhard Piepenbreier, “Comparison of carrier signal based approaches for
sensorless wound rotor synchronous machines”, International Symposium on Power Electronics, Electrical
Drives, Automation and Motion, 2014
[8] Yu-seok Jeong, Robert D. Lorenz; Thomas M. Jahns, and Seung-Ki Sul, “Initial Rotor Position Estimation of
an Interior Permanent-Magnet Synchronous Machine Using Carrier-Frequency Injection Methods”, IEEE
transactions on industry applications, vol. 41, no. 1, january/february 2005;
[9] Michael C. Harke, Dejan Raca, Robert. D. Lorenz, “Implementation Issues for Fast Initial Position and Magnet
Polarity Identification of PM Synchronous Machines with Near Zero Saliency, EPE 2005 – Dresden
[10] Joachim Holtz, Fellow, “Initial Rotor Polarity Detection and Sensorless Control of PM Synchronous
Machines” Conference Record of the 2006 IEEE Forty-First IAS Annual Meeting
[11] Sung-Yoon Jung, Jinseok Hong, Kwanghee Nam, “Current Minimizing Torque Control of the IPMSM Using
Ferrari’s Method” IEEE Transactions on Power Electronics Year: 2013 Volume: 28 , Issue: 12 Pages: 5603 –
5617.

EPE'19 ECCE Europe ISBN: 978-9-0758-1530-6 - IEEE catalog number: CFP19850-USB P.6
Assigned jointly to the European Power Electronics and Drives Association & the Institute of Electrical and Electronics Engineers (IEEE)