comparative evaluation of electrical and calorimetric

IEEJ Journal of Industry ApplicationsVol.8 No.3 pp.386–393 DOI: 10.1541/ieejjia.8.386

Paper

Comparative Evaluation of Electrical and Calorimetric Methodsto Estimate Core Loss in Powder Cores

Yuki Ishikura∗,∗∗a)Member, Jun Imaoka∗∗ Member

Mostafa Noah∗∗ Non-member, Masayoshi Yamamoto∗∗ Member

(Manuscript received June 4, 2018, revised Aug. 30, 2018)

The utilization of powder cores for many power electronic applications has drawn a significant amount of attentionowing to their attractive magnetic properties. However, because of their low relative permeability characteristics, aprecise measurement of the core loss is very difficult to obtain. Imprecise estimation of the core loss may lead to aninaccurate thermal design of power converters. Measuring and calculating the accurate value of the core loss allowsthe circuit designer to design a more efficient system. This paper discusses the revalidation of electrical measurementby comparing the electrical method with the calorimetric method in powder cores. The calorimetric method is one ofthe most promising methods to accurately measure the core loss; however, it is time consuming. A difference less than10% has been reported for the core loss measurement error while comparing the electrical method and the calorimetricmethod in various conditions. These results indicate that the electrical method using the B-H analyzer is effective formeasuring the core losses in some conditions for powder cores. Along with the theoretical discussion, simulation andexperimental tests are also conducted.

Keywords: powder core, core loss, electrical methods, calorimetric methods, relative permeability

1. Introduction

The importance of manufacturing efficient and high-powerdensity converters is growing along with the growth of themarket in many industrial applications. Powder cores, whichhave a high curie temperature and a low relative permeabil-ity with high DC bias current are very attractive to be uti-lized in many power electronics applications, since they pos-sess soft saturation characteristic and high saturation mag-netic flux density. In addition, magnetic powder cores exhibitlower eddy current loss compared to discrete air gap ferrites,since powder cores have tiny air gaps distributed in materialwhich eliminates fringing effects, as shown in Figs. 1(a) and(b). These attractive features have gained much attention indesigning AC and DC inductors (1)–(6).

However, one of the drawback of the powder core is thedifficulty of measuring accurate core losses due to the lowrelative permeability. For instance, the relative permeabilityof ferrite cores (Mn-Zn) is around 5000∼10000, and the rel-ative permeability of powder cores is around 20∼200. Mea-suring the actual value of the core loss in these materials isessential, to ensure an optimum thermal design for the pas-sive component (7)–(10).

In general, core loss measurement methods are categorizedinto two techniques: (1) electrical method (7)–(14), (2) calorimet-ric method (15)–(19).

a) Correspondence to: Yuki Ishikura. E-mail: y [email protected]∗ Murata Manufacturing Co., Ltd.

1-10-1, Higashikotari, Nagaokakyo-shi, Kyoto 617-8555, Japan∗∗ Nagoya University

Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8601, Japan

(a) Discrete air gap (b) Distributed air gap

Fig. 1. Magnetic core with air gap

1.1 Electrical Methods Electrical methods basicallyrely on measuring the voltage and current of the winding ter-minals. By processing these measurements, the core loss tocalculate B-H loop can be obtained. Since these methods canbe performed within a short time, it is highly reproducible.The accuracy of core loss measurement has been discussedin previous studies (7)–(12). Among others, in case of powdercores, huge measurement errors are often encountered, whichoccurs as a result of the phase shift between voltage and cur-rent, due to the low relative permeability (11) (12). Removing thephase-shift error in the core is important while obtaining thecore losses in a magnetic material with low relative perme-ability, and it has been proposed before in (13) (14). However,this method is difficult to implement, since it requires finetuning of the cancellation element which can be very timeconsuming.1.2 Calorimetric Methods Calorimetric methods

have been widely used in many power circuits to measurepower losses of magnetic components (15)–(19). The total powerlosses dissipate as heat, and it leads to a temperature rise ofthe core, measuring the temperature rise using the calorimet-ric methods gives a proper measurement of the core losses.Consequently, calorimetric methods are considered to be one

c© 2019 The Institute of Electrical Engineers of Japan. 386

Comparative Evaluation of Electrical and Calorimetric Methods（Yuki Ishikura et al.）

of the most promising methods to attain accurate power lossmeasurements (15). However, the measurement process is verytime consuming to implement in real industrial applications.

As previously mentioned, measurement methods of coreloss are categorized into two techniques; namely, electricalmethods and calorimetric methods. Previous studies reportedin the literature only discussed either one of the two meth-ods, in terms of evaluating the accurate measurements ofthe core losses in powder cores. Furthermore, B-H analyzerwas utilized while evaluating the accuracy of the electricalmethod (11) (12). As a result, the measured value is influencedby the error introduced by the B-H analyzer, therefore, themeasured value of the core losses is inaccurate compared tothe actual value.

This paper discusses the measurement accuracy of thecore loss in powder cores using the electrical measurementmethod. Furthermore, a comparison is conducted betweenthe electrical method and a simple calorimetric measurementmethod. The purpose of this discussion is to reevaluate themeasurement accuracy of the electrical measurement in pow-der cores.

This paper is divided into five sections. Sections 2 dis-cusses the electrical measurement using the B-H analyzer in-cluding the measurement principal, and the measurement ac-curacy of the core losses. Also, the core losses of the twopowder cores which have different characteristics are actu-ally measured including DC bias conditions and temperaturecharacteristics. Furthermore, this paper proposes implemen-tation method for core loss calculation into a circuit sim-ulator. In Section 3, the simple calorimetric measurementmethod is described including the measurement system andthe measurement principal. The inductor losses of two pow-der cores actually measure in various conditions. In addition,the separation method of the core losses and the copper lossesis discussed. In section 4, to validate the accuracy of the elec-trical method in core losses, both electrical measurement andcalorimetric measurement are compared through the experi-mental test in various conditions. Finally, conclusions s arepresented in section 5.

2. Core Losses Calculation using the ElectricalMethod

In this section, the electrical measurement using B-H an-alyzer is discussed. In this discussion, two toroidal powdercores from POCO Magnetic Co., Ltd., which have differentcharacteristics were utilized to evaluate the measurement ac-curacy of core losses. Toroidal core is considered as one ofthe most popular shapes used in inductors. The specificationsof the powder cores are tabulated in Table 1.2.1 Test Set-up of Electrical Method Within various

electrical methods, using the B-H analyzer is one of the mostattractive method, thanks to repeatability and simplicity (11) (12).Figure 2(a) shows the experimental set-up. The experientialset-up consist of a B-H analyzer (SY-8218, IWATSU ELEC-TRIC CO., LTD.), a power amplifier, a DC power supply, anda ripple cut reactor.

The basic concept of measuring the core loss is as follows:three windings are placed around the CUT (core under test),and AC current flow in the excitation winding (primary wind-ings), which is proportional to an AC magnetic field strength

Table 1. Characteristic of powder cores used in the test

(a) Experimental set-up

(b) CUT

Fig. 2. Circuit diagram of the experimental set-up andCUT

HAC. The excitation voltage waveform of the primary wind-ing is sinusoidal. The voltage of sensing winding (secondarywinding) is measured and integrated to calculate the flux den-sity B. The core loss per unit volume (Pcv) is the enclosedarea of the B-H loop, multiplied by the excitation frequency.The DC current in the tertiary winding is proportional to theDC magnetic field strength HDC. Therefore, core losses canbe measured under any DC bias conditions by injecting a DCcurrent into the tertiary winding. Figure 2(b) shows the CUTused in this measurement. The primary windings and the sec-ondary windings used in CUT are litz wire with the diameterof 0.22 mm × 7, to realize low an AC resistances. The tertiarywinding is single wire with the diameter of 0.8 mm, with thepurpose of achieving low a DC resistance.2.2 Accuracy of Electrical Measurement Due to the

low relative permeability of the powder cores, the core lossmeasurement is very sensitive to phase shift error, and it isvery difficult to obtain accurate measurements.

The electrical equivalent circuit of CUT is shown inFig. 3(a) (11) (12). The magnetizing inductance is represented asLm and the resistance RFE represents the loss in the magneticcore (RFE = Vcore

2/Ploss). The winding capacitances of theprimary windings and the secondary windings are denoted byCw. The vector diagram of current and voltage in the CUTis shown in Fig. 3(b). In this test, since the capacitive current

387 IEEJ Journal IA, Vol.8, No.3, 2019


(a) Equivalent circuit (b) Vector diagram

Fig. 3. Equivalent circuit and vector diagram of CUT

Fig. 4. Measurement error properties in electrical mea-surements

icw is much lower than the inductive current iLm, the para-sitic winding capacitance can be neglected (f < 1 MHz) (11) (12).In the powder core, the low relative permeability allows forhigh values of magnetizing inductance current. Therefore,the phase angle between the primary current and the sec-ondary voltage of the CUT approaches 90◦. The electricalmethod is high sensitive to the phase shift error. The phaseshift error is caused by delay of voltage and current sens-ing. The core loss measurement error can be expressed asfollows (7)–(12):

E =cos(θ + φ) − cos(θ)

cos(θ)· 100% · · · · · · · · · · · · · · · · · · (1)

Where E is the percentage error in the core loss, θ is the ac-tual phase shift between the sensing voltage and the sensingcurrent, and φ is the phase shift error. Figure 4 shows mea-surement error properties. It is noticeable that the when thephase difference get closer to 90◦, it produces more than a100% error. In order to reduce measurement errors, the testis performed using the phase error correction function in B-Hanalyzer, which can achieve a measurement accuracy toler-ance of ±0.15◦.2.3 Core Loss Measurement using B-H AnalyzerThe measurements were obtained under the following con-

ditions: excitation voltage is sinusoidal waveform, frequencyrange is from 20 kHz to 60 kHz, maximum flux density Bm

range is from 0.025 T to 0.15 T and the DC magnetic fieldintensity HDC range is from 0 A/m to 8000 A/m.

Figure 5 shows the core loss per unit volume Pcv - Bm char-acteristic. It is clear that the core loss for each core increaseswith an increase in Bm because the core loss depends on thearea of the dynamic minor loop, which increases as Bm in-creases. In addition, the core losses of core A is almost threetimes of core B. Since core B material consists of Fe-Si-Al,therefore, it exhibits low core loss characteristics (3).

Figure 6 shows DC bias characteristics of core losses. Itis shown that core losses change depending on DC bias mag-netic fields, and the two cores can be characterized as fol-lows: (1) the core losses of the core A gradually increase

Fig. 5. Pcv - Bm characteristic (20 kHz, 25◦C)

Fig. 6. Pcv -HDC characteristic (20 kHz, 25◦C)

Fig. 7. Phase angle θ - frequency characteristic (Bm =0.075 T)

with higher values of DC magnetic field intensity, (2) the corelosses of core B gradually decrease with higher values of DCmagnetic field intensity. These characteristics are highly de-pendent on materials.

Figure 7 shows the phase angles for the cores under test.These results had been averaged, after repeatedly conduct-ing this test for five times. The phase angle values of core Ais between 88.4◦ to 88.8◦, core B is between 89.6◦ to 89.8◦.The tolerance of these results is within ±0.01◦ during fivemeasurements outputs. Figure 7 shows that the phase an-gles decrease from 90◦ as the core loss increase. Also, thephase angles approach 90◦ as the core loss decrease. Theseresults indicate that the core losses measurement of core Bwith low core loss is more difficult than core A. Since thesecores have phase shift tolerance of ±0.01◦, it is possible tomeasure materials up to θ = 89.8 (at 20 kHz) when the ex-pected measurement accuracy is under 5%. Although thereare few variations between measurements in the measure-ment error, the measurement accuracy of the B-H analyzeris 0.15◦. This measurement accuracy in the B-H analyzermay lead measurement errors which is 10∼15% in core Aand 37.5∼75% in core B. In order to verify these possibil-ities, in section 4, measurement results using the electricalmethod are compared with results using a simple calorimet-ric method.

In electrical methods, the measured core losses are



obtained while the excitation waveform is sinusoidal. In or-der to use these data in power electronics applications, thecommon approach for estimating core loss is improved Gen-eralized Steinmetz Equation (iGSE), this formula is used tocalculate core losses under square-wave excitation at the re-spective fundamental frequency (7).

Pcv =1T

∫ T

0ki

∣∣∣∣∣dBdt

∣∣∣∣∣α

(ΔB)β−αdt · · · · · · · · · · · · · · · · · (2)

ki =k

(2π)α−1

∫ 2π

0ki| cos θ|α2β−αdθ

· · · · · · · · · · · · · · · (3)

Where ΔB is peak-to-peak flux density (ΔB = 2Bm) and ki, αβare material parameters. The iGSE is capable of calculatingcore loss of any flux waveform, without requiring extra mate-rial parameters beyond the steinmetz parameters gained fromsinusoidal excitations. In addition, the Steinmetz Premag-netization Graph (SPG) is used for the core loss calculationunder DC bias conditions (9). The SPG is a way of consider-ing the change of Steinmetz parameters of iGSE. By applyingthe SPG to the measured core loss data, it is possible to cal-culate the core loss under any excitation waveform and DCmagnetic field bias conditions.2.4 Temperature Characteristic of Core Loss In or-

der to realize accurate the core losses calculation, it is neces-sary to investigate the temperature dependence of the powdercore measuring the core losses.

Figure 8 shows the dependence of core losses on the tem-perature, at 25◦C and 75◦C. It can be clearly understood thatthe change of core losses is under 2% between 25◦C and75◦C. These results indicate that core loss of the two pow-der cores under the test is not sensitive to a temperature risefrom 25◦C to 75◦C.

Figure 9 shows the temperature dependence of the relativepermeability in core A. Figure 9 indicates that the lower thepermeability, the less dependence of relative permeability ontemperature. Since the relative permeability of the powdercores used in this experiment is almost not affected by tem-perature variation, the temperature dependence of the corelosses are small. From these reasons, it can be concluded thatthe core losses of the powder cores used in this experimentdoes not need to consider temperature variations up to 75◦C.2.5 Core Loss Calculation using Circuit SimulatorIf the core loss calculation of powder cores can be modeled

into a time-domain circuit simulator, the core loss of powdercores with respect to various circuits can be sufficiently evalu-ated before constructing a prototype, so that number of hard-ware iterations can be reduced. Therefore, this paper pro-poses implementation method for core loss calculation intoa circuit simulator, in this section. Using the core loss dataobtained by the electrical method, core loss calculation is im-plemented in the circuit simulator as follows.

Firstly, non-linearity of the relative permeability is mod-eled. The model based on a permeance-capacitance mag-netic circuit is built up in the system-level simulation plat-form PLECS (Plexim GmbH), by utilizing the magnetic com-ponents listed in the library and c-script (20). The permeance-capacitance based magnetic circuits representing the core canbe directly parameterized using the geometrical and material

Fig. 8. The temperature dependence of the core losscomparing 25◦C and 75◦C

Fig. 9. The temperature dependency of the relative per-meability in cores

information (21) (22), as follows:

P(HDC) = μ0 · μr(HDC) · Al· · · · · · · · · · · · · · · · · · · · · · · (4)

Where the cross sectional area A and magnetic path lengthl are geometry-related constant parameters once the core isselected, and μ0 is the permeability of vacuum. The rela-tive permeability μr(HDC) accounts for material characteris-tics which continuously change depending on the magneticfield strength HDC. While, the characteristic of the relativepermeability μr(HDC) can be expressed by (5) using the fit-ting model proposed in (6).

μr(HDC) =

⎛⎜⎜⎜⎜⎜⎜⎜⎝1 +

p

1 +

( |HDC|q

)r

⎞⎟⎟⎟⎟⎟⎟⎟⎠ · · · · · · · · · · · · · · · · · · · (5)

Where, the parameters p, q and r are material-related con-stant parameters. By solving (4) and (5), the permeance P(HDC) then can be represented as follows:

P(HDC) =

⎛⎜⎜⎜⎜⎜⎜⎜⎝1 +

p

1 +

( |HDC|q · l

)r

⎞⎟⎟⎟⎟⎟⎟⎟⎠ μ0 · Al· · · · · · · · · · · · · (6)

Regarding the relative permeability considering the DCbias characteristics, Fig. 10 shows the comparison result be-tween approximated results obtained from (5) and Table 2,and measured values. It is clear that the approximated resultsusing (5) is consistent with the measurement results. Theseresults shows that the DC bias characteristics of the relativepermeability can be accurately modeled in two cores withdifferent properties. Furthermore, the core loss characteris-tics calculated using iGSE are implemented in the simulationmodel.

In order to evaluate of core loss calculation method us-ing the circuit simulator, we conduct the core loss simulationwith the bidirectional DC-DC converter. Figure 11 shows the



Fig. 10. The comparison between the measured valuesand approximated results using (5)

Table 2. Inductor parameters

Fig. 11. Core loss calculation model in BidirectionalDC-DC converter with a permeance-capacitance mag-netic circuit

circuit configuration of the bidirectional DC-DC converterwith the inductor core loss calculation model. In the bidirec-tional DC-DC converter, the flux density ripple ΔB and theDC bias magnetic field strength HDC of the magnetic corecan be calculated as follow

ΔB =Vi · d

fs · N · A · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · (7)

Where, V i is input voltage, d is duty ratio, f s is switchingfrequency, N is number of turn.

HDC =N · IL

l· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · (8)

Where, IL is inductor average current.Figure 12 shows the simulation algorithm flowchart to cal-

culate core losses including DC bias conditions in the c-script. For example, the simulation is conducted consideringthe following two conditions

1. V i = 20 V, D = 0.5, Fs = 20 kHz, IL = 0 Arms.2. V i = 20 V, D = 0.5, Fs = 20 kHz, IL = 10 Arms.Figures 13(a) and (b) show the simulation results in IL,

Pcv, μr. As shown Fig. 13, when inductor average current

Fig. 12. The flowchart of the simulation algorithm ofthe core loss calculation including DC bias conditions

(a) IL = 0 Arms (b) IL = 10 Arms

Fig. 13. Waveforms of simulation

IL change from 0 A to 10 A, inductor ripple current increasesfrom 2.60 A to 3.52 A, as a result of a drop in the relative per-meability. Since these results almost agree with the DC mag-netic field bias characteristic of the relative magnetic perme-ability shown in Fig. 10, it can be said that the DC magneticfield bias characteristic of the relative permeability can be ac-curately modeled. Furthermore, when inductor average cur-rent IL change from 0 A to 10 A, core loss increases 0.43 Wto 0.57 W, as a result of a drop in the relative permeability.The simulation results of the core losses are discussed in sec-tion 4.

3. Core Losses Calculation using the Calorimet-ric Measurement

3.1 Experimental Setup of Calorimetric MethodFigures 14(a) and (b) show the test system and a picture

of the calorimeter. The thermal pot (JBI-273, THERMOSK.K.) has a simple double jacket structure with the vacuumbetween inner and outer insulating enclosures. This struc-ture allows to separate the inner hot air from the ambient air,easily improve the measurement accuracy of the calorimetricmethod. The measurement sample of the inductor is insulatedfrom the thermal pot with the insulating tape. During the ex-periment, the calorimeter is covered with another thermal in-sulation sheet to minimize the heat leakage. There is certainamount of coolant in the thermal pot. In this test, Fluorinert R©(FC-43, 3M Company) is used as the coolant, which has high



(a) Overview of the test system

(b) The picture of test system and inner of the calorimeter

(c) Bidirectional DC-DC converter

Fig. 14. Overview of the test system and picture

Table 3. Specification of bidirectional DC-DC con-verter

electrical insulation. Also, Fluorinert R© is high heat conduc-tion capabilities owing to high density and low kinematic vis-cosity. Three thermocouples are used for temperature mea-surements. All of these are attached to the bottom of the ther-mal pot to measure temperature of the coolant as shown inFig. 14(b). All wires are passing through the hole in the ther-moelectric pot cap. The hole is covered with insulation tape.Figure 14(c) shows the simplified schematic of bidirectionalDC-DC converter, and specifications are listed in Table 3.

Before measuring inductor losses, calibration test are con-ducted which the relationship between the loss and tempera-ture rise is measured with a reference heat generator. In thistest, the winding resistance of the CUT as the heating ele-ment is selected. By using the winding resistance of CUT,it is possible to measure the relationship between losses andtemperature rise under the same condition as the actual coreloss measurement test. The winding resistor is heated by DCcurrents. Stirring is necessary for the uniform temperature ofthe coolant. After a fixed period of time, around five minutes,the injecting current shall be stopped, and the value of tem-perature rise in the pot shall be taken. The energy dissipatedby the CUT can be calculated from measured values of thetemperature rise ΔT [◦C] of the coolant as follows:

Fig. 15. Calibration test result

Q = m · c · ΔT · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · (9)

Where Q [J] is heat energy, m is mass of the coolant, andc [J/(g K)] is the specific heat capacity of the coolant. Sincethe heat energy Q [J] is proportional to temperature riseΔT [◦C], the power loss [W] can be calculated form the re-lation of the time and the temperature rise. After conduct-ing the experiments, the CUT shall be cooled down to reachthe ambient temperature. By repeating the measurement pro-cess under different power loss conditions, the relationshipbetween the power loss and temperature rise can be plotted.Experimental results are shown in Fig. 15. The least squaremethod of the linear regression calculations is utilized to lin-earize the temperature rise of the calorimeter, and a slope(◦C/W: heat resistance) in (9) for each loss was calculated.For example, in Fig. 15, when the CUT is excited at the powerlevel from 0.2 W to 1.9 W, the relationship between powerlevel and temperature rise can be expressed as

y = 2.71x + 0.04 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · (10)

Where y is the temperature rise of the calorimeter, x is thepower, and the second term of (10) is the offset temperatureof the calorimeter. Microsoft Excel is used to perform the lin-ear regression calculation. Once the reference line equationis obtained, losses of the inductor in the calorimeter box canbe calculated.3.2 Separation of Core Loss and Copper Loss In

calorimetric methods, since the total loss of the CUT includesthe copper loss, it is necessary to separate the core loss andthe copper loss in order to calculate the core loss. The copperloss consists of the DC resistance loss and the AC resistanceloss. It is extremely difficult to accurately calculate the ACresistance loss caused by the skin effect and the proximityeffect. In this study, the copper loss is calculated assumingthat the resistance of the copper wire is the same as the DCresistance. Since the thickness of the winding of the induc-tor is 0.8 mm and the skin depth at the switching frequencyof 20 kHz is 0.46 mm, AC resistance is not taken into con-sideration. Also, the proximity effect is not considered forsimplifying the calculation (23) (24). Using the measured DC re-sistance shown in Table 4, the copper loss can be calculatedas follows:

Pcu = RdcIdc2 + RacIac

2 � RdcIL rms2 · · · · · · · · · · · · · (11)

Where, Pcu is the total copper loss, IDC is the AC current,IAC is the ac current and IL rms is the inductor effective cur-rent. The core loss can be calculated by subtracting the cop-per loss from the total loss which is the result of the simple



Table 4. Inductor parameters

calorimetric method.

4. Comparison between Electric and Calorimet-ric Calculation Method

In this section, in order to reevaluate the core losses cal-culation accuracy of the electrical method, results using theelectrical method are compared to the calculation results us-ing the calorimetric method.

Figures 16(a) and (b) show the dependence of the corelosses on ΔB, under different DC bias conditions. The simplecalorimetric method is compared to electrical method usingthe circuit simulator and experimental results. Shown thatcore losses change depending on DC bias magnetic fields,and, the measurement tolerance of both methods is below0.03 W. Since core losses measurement range of the core Ais from 0.7 W to 1.5 W, the core loss measurement accuracyrange is from 2% to 5%. On the other hand, the core lossesmeasurement range of the core B is from 0.3 W to 0.8 W,and the core loss measurement accuracy range is from 4%to 10%. In addition, Fig. 16(c) shows the characteristics ofthe duty cycle in the core loss. The measurement tolerance ofboth methods is below 0.04 W. Since the core loss measure-ment range of core A is 0.4 W to 1.1 W, the accuracy of coreloss measurement is 2 to 10%. As these results, the actual er-rors are sufficiently less than the assumed error mentioned insection 2. These results indicate that the accuracy of the elec-trical method is sufficiently high within these measurementrange. From these results, it is conclude that the accuracyof the electric method is sufficiently high using the powdercores used this test.

When measuring a smaller core loss range using the calori-metric method, the measurement accuracy decreases becausethe loss becomes lower. Therefore, another verification suchas adjusting the amount of coolant is necessary. In the electri-cal method, the frequency increases, or when measuring thecore loss of the powder core having less core loss characteris-tics, the measurement accuracy decreases. Also, if a powdercore having a lower relative permeability is used, the mea-surement accuracy of the core loss decreases. These prob-lems shall be addressed in the future.

5. Conclusion

In this paper, in order to reevaluate the measurement accu-racy of the electrical measurement in powder cores, a com-parative study of two core loss calculation methods for pow-der cores has been performed considering DC bias conditionsand duty cycles. The first method is the electrical method us-ing by the B-H analyzer and IGSE. The electrical measure-ment method can easily measure the core loss under manyconditions including the DC bias characteristics. The secondone is the calorimetric method using the simple calorimeter.A simple calorimetric method is characterized by high mea-surement accuracy, although it is a time consuming method

(a) Core loss - ΔB characteristic in core A

(b) Core loss - ΔB characteristic in core B

(c) Core loss - Duty cycle characteristic in core A

Fig. 16. Results of comparison between the electricalmethod and the calorimetric method

compared to the electrical measurement method. In addition,core loss calculation method of powder cores using the time-domain circuit simulator is proposed. The core loss calcu-lation method using the time-domain circuit simulator caneasily calculate the core loss of the inductor according tothe circuit topology under various conditions. As a resultof the evaluation, in two different core materials, the elec-trical method and calorimetric method were compared un-der various conditions, and core loss calculation difference ofless than 10% was confirmed. These results indicate that theelectrical method is very practical in terms of accuracy andreproducibility since the measurement accuracy is almost thesame as the measurement method of simple calorimetry insome conditions.

References

( 1 ) T. Saito, S. Takemoto, and T. Iriyama: “Resistivity and core size dependen-cies of eddy current loss for Fe-Si compressed cores”, IEEE Trans. on Mag-netics, Vol.41, No.10, pp.3301–3303 (2005)

( 2 ) C. Appino, O. Bottauscio, O. de la Barri‘ere, F. Fiorillo, A. Manzin, and C.Ragusa: “Computation of eddy current losses in soft magnetic composites”,IEEE Trans. on Magnetics, Vol.48, No.11, pp.3470–3473 (2012)

( 3 ) T. Ishimine, A. Watanabe, T. Ueno, T. Maeda, and T. Tokuoka: “Develop-ment of low-iron-loss powder magnetic core material for high-frequency ap-plications”, SEI Tech. Rev, No.72, pp.117–123 (2011)



( 4 ) B.G. You, J.S. Kim, B.K. Lee, G.B. Choi, and D.W. Yoo: “Optimization ofpowder core inductors of buck-boost converters for hybrid electric vehicles”,J. Electr. Eng. Technol., Vol.6, No.4, pp.527–534 (2011)

( 5 ) Y. Wang, G. Calderon-Lopez, and A.J. Forsyth: “High-Frequency GapLosses in Nanocrystalline Cores”, IEEE Trans. on Power Electronics, Vol.32,No.6, pp.4683–4690 (2017)

( 6 ) J. Imaoka, K. Okamoto, M. Syoyama, Y. Ishikura, M. Noah, and M.Yamamoto: “Modeling, Magnetic Design, and Simulation Methods Consid-ering DC Superimposition Characteristic of Powder Cores Used in PowerConverters”, in Proc. of the 2018 International power Electronics conference-ECCE ASIA- (IPEC-Niigata 2018), pp.1095–1102 (2018)

( 7 ) K. Venkatachalam, C.R. Sullivan, T. Abdallah, and H. Tacca: “Accurate pre-diction of ferrite core loss with nonsinusoidal waveforms using only Stein-metz parameters”, in Proc. of IEEE Workshop Comput. Power Electron.,IEEE Workshop on, pp.36–41 (2002)

( 8 ) T. Shimizu and S. Iyasu: “A practical iron loss calculation for ac filter induc-tors used in PWM inverters”, IEEE Transactions on Industrial Electronics,Vol.56, No.7, pp.2600–2609 (2009)

( 9 ) J. Muhlethaler, J. Biela, J. Kolar, and A. Ecklebe: “Core Losses Under theDC Bias Condition Based on Steinmetz Parameters”, IEEE Trans. on PowerElectronics, Vol.27, No.2, pp.953–963 (2012)

(10) R. Beres, X. Wang, F. Blaabjerg, C. Leth Bak, H. Matsumori, and T. Shimizu:“Evaluation of core loss in magnetic materials employed in utility grid ACfilters”, in Proc. of Applied Power Electronics Conference and Exposition(APEC), pp.3051–3057 (2016)

(11) F.F. Javidi and M. Nymand: “Error analysis of high frequency core loss mea-surement for low-permeability low-loss magnetic cores”, in Proc. of IEEE2nd Annual Southern Power Electronics Conference (SPEC), pp.1–6 (2016)

(12) N.F. Javidi, M. Nymand, and A.J. Forsyth: “New method for error com-pensation in high frequency loss measurement of powder cores”, in Proc. ofApplied Power Electronics Conference and Exposition (APEC), pp.876–881(2017)

(13) D. Hou, M. Mu, F. Lee, and Q. Li: “New core loss measurement methodfor high frequency magnetic materials”, IEEE Tran. on Power Electronics,Vol.29, No.8, pp.4374–4381 (2014)

(14) D. Hou, M. Mu, F. Lee, and Q. Li: “New high frequency core loss mea-surement method with partial cancellation concept”, in IEEE Transactionson Power Electronics, Vol.32, No.4, pp.2987–2994 (2016)

(15) R. Linkous, A.W. Kelley, and K.C. Armstrong: “An improved calorimeterfor measuring the core loss of magnetic materials”, in Proc. of 15th AnnualIEEE Applied Power Electronics Conference & Exposition (APEC), Vol.2,pp.633–639 (2000)

(16) C. Xiao, G. Chen, and W.G.H. Odendaal: “Overview of Power Loss Mea-surement Techniques in Power Electronics Systems”, IEEE Trans. on Indus-try Applications, Vol.43, No.3, pp.657–664 (2007)

(17) N S.D.J. Weier, M.A. Shafi, and R.A. McMahon: “Precision Calorimetry forthe Accurate Measurement of Losses in Power Electronic Devices”, in Proc.of IEEE Industry Applications Society Annual Meeting, pp.1–7 (2008)

(18) P.D. Malliband, N.P. van der Duijn Schouten, and R.A. McMahon: “Preci-sion calorimetry for the accurate measurement of inverter losses”, in Proc.The 5th International Conference on Power Electronics and Drive Systems(PEDS), Vol.1, pp.321–326 (2003)

(19) F. Blaabjerg, J.K. Pedersen, and E. Ritchie: “Calorimetric measuring sys-tems for characterizing high frequency power electronic components and sys-tems”, in Proc. Industry Applications Conference., 37th IAS Annual Meet-ing, Vol.2, pp.1368–1376 (2002)

(20) [Online]Available: https://www.plexim.com/plecs(21) J. Allmeling, W. Hammer, and J. Schonberger: “Transient Simulation of

Magnetic Circuits Using the Permeance-Capacitance Analogy”, in Proc. ofIEEE 13th Workshop on Control and Modeling for Power Electronics (COM-PEL), pp.1–6 (2012)

(22) M. Luo and D. Dujic: “Permeance based modelling of the core corners con-sidering magnetic material nonlinearity”, in Proc. of Annual Conference ofthe IEEE Industrial Electronics Society (IECON), pp.950–955 (2015)

(23) H. Matsumori, T. Shimizu, K. Takano, and H. Ishii: “Three-phase AC Fil-ter Inductor Design for Three-phase PWM Inverter for Conversion EfficiencyImprovement at Low Load”, IEEJ Trans. on IA, Vol.137, No.5, pp.395–405(2017) (in Japanese)

(24) K. Emori, T. Shimizu, and Y. Bizen: “Discussion on design optimization ofinductor loss focused on copper loss and iron loss”, in Proc. of 1st Interna-tional Future Energy Electronics Conference (IFEEC), pp.241–245 (2013)

Yuki Ishikura (Member) received his M.S. degree electrical and elec-tronic system engineering from Shimane Universityin 2012. From 2012 to 2014, he was a circuit de-sign engineer at Hitachi, Ltd., Japan. Since 2014,he has been with Murata Manufacturing Co., Ltd.,Japan. He is currently pursuing the Ph. D. degreeat Nagoya University, Japan. His research interestsinclude power electronics for DC-DC converter andinverter in renewable energy system.

Jun Imaoka (Member) received his M.S. and Ph.D. degrees in elec-tronic function and system engineering from ShimaneUniversity in 2013 and 2015 respectively. Since Oct.2015 to Mar. 2018, he was with Department of Elec-trical Engineering, Kyushu University as an Assis-tant Professor, Fukuoka, Japan. From Apr. 2018,He is currently with Department of Electrical Engi-neering, Nagoya University as an Assistant Profes-sor, Nagoya, Japan. His research interests includedesign of integrated magnetic components, modeling

for high power density power converters, thermal management for powerconverters, magnetic material application, EMI of switching power supply.

Mostafa Noah (Non-member) was born in Cairo, Egypt. He receivedhis B.Sc. and M.Sc. degrees in Electrical Engineeringfrom Cairo University, Egypt in 2009 and 2014. From2009 to 2015, he was an Electrical Design Engineerwith multinational consultant firm Dar Al-Handasahand SCG. From 2015 to 2017, he was a research as-sistant at the Power Electronics Lab of Shimane Uni-versity, Japan. Currently, he is with the Power Elec-tronics Lab of Nagoya University, Japan, where he ispursuing Ph.D. His research interests include the de-

sign of the integrated magnetics in multi-phase DC-DC converters, and LLCresonant converters.

Masayoshi Yamamoto (Member) received his M.S. and Ph.D. degreein science and engineering from Yamaguchi Univer-sity, Yamaguchi, Japan in 2000 and 2004 respectively.From 2004 to 2005, he was with Sanken Electric Co.,Ltd., Saitama, Japan. From 2006 to 2017, he waswith the Interdisciplinary Faculty of Science and En-gineering in Shimane University, Japan, as an Asso-ciate Professor. He is currently a Professor at Instituteof Materials and Systems for Sustainability (IMaSS),Nagoya University, Japan. His research interests in-

clude power supply for HEV (boost converter, buck converter, 3-phase in-verter, digital control), charging system for EV, LED illumination systemfor a tunnel, EMI of switching power supply, and wireless power transfer.


comparative evaluation of electrical and calorimetric

Documents