A New Series of High Frequency Core Loss Measurement Methods Mingkai Mu, Fred C. Lee Center for Power Electronics Systems Virginia Tech February 8, 2012 February 8 , 2012 Orlando, FL APEC 2012 1 Increasing Frequency for High Power Density Frequency 10MHz Enpirion (9A) TI(1A) Maxim (3.6A) TI (1A) Maxim (10A) 1MHz Linear (8A) Power-one ( 15A) Delta (15A) Year 100kHz 1998 February 8 , 2012 Orlando, FL 2000 2002 2004 APEC 2012 2006 2008 2010 2 Core Loss Map for Magnetic Materials 10 4 Core Loss Density (kW/m3) Bm=20mT 10 10 3 2 1 10 2 10 Iron Powder February 8 , 2012 Orlando, FL 3 10 Frequency (kHz) Ferrite APEC 2012 Flake 10 4 Alloy Thin Film 3 Outlines • Overview • Existing core loss measurement methods • Proposed core loss measurement methods – For sinusoidal excitations – For general excitations • Conclusion February 8 , 2012 Orlando, FL APEC 2012 4 Core Loss Measurement Method I P= c p ⋅ m ⋅ ∆T dt Pros Arbitrary excitation Block Diagram of Closed Type Calorimeter Cons •Setup is complex and time consuming. •Not good for low power measurement •If use winding to excite the core, winding resistance can be hardly separated. *Chucheng Xiao, Gang Chen and .G.Odendaal, “Overview of Power Loss Measurement Techniques in Power Electronics Systems”, Industry Applications Conference, 2002 February 8 , 2012 Orlando, FL APEC 2012 5 Core Loss Measurement Method II ω Li i i ωC Choose C to resonant with L at the measured frequency, so voltage on L and C cancel each other. Vout − pk Vin − pk Rcore ≈ 1 1 jωC = ≈ Rcore + Rcu + Rc ωC ( Rcore + Rcu + Rc ) Rc + Pros. Minimize phase error. Cons. •Hard to exclude winding loss •C value is critical •Sinusoidal waveform only 1 Vin − pk − Rcu − Rc ωC Vout − pk *Yehui Han, Grace Cheung, An Li, Charles R. Sullivan† and David J. Perreault, “Evaluation of Magnetic Materials for Very High Frequency Power Applications”, Power Electronics Specialists Conference, 2008. February 8 , 2012 Orlando, FL APEC 2012 6 Core Loss Measurement Method III Equivalent Circuit model Calculation 1 B= v2 (t )dt ∫ N 2 Ae T iR Lm Pcore = N1 f N 2 Rref ∫ v (t )v 2 R (t )dt T *Thottuvelil, V.J.; Wilson, T.G.; Owen, H.A., Jr.; , "High-frequency measurement techniques for magnetic cores," Power Electronics, IEEE Transactions on , vol.5, no.1, pp.41-53, Jan 1990 February 8 , 2012 Orlando, FL APEC 2012 7 Limitation: Sensitivity to Phase Discrepancy vR iR 1° discrepancy For a film resistor of 2 ohm, if the ESL is 0.5nH, it will cause 0.9° discrepancy at 5MHz. % Power Error v2 vR iR Lm 1° phase discrepancy Phase angle between V2 and IR February 8 , 2012 Orlando, FL ∆ = tan(ϕ v −i ) ⋅ ∆ϕ Phase angle between V2 and IR APEC 2012 8 Sources of Phase Discrepancy Phase Discrepancy Sample rate limitation Mismatch between probes Pros •Exclude winding loss. •Arbitrary waveform. February 8 , 2012 Orlando, FL Cons •Sensitive to phase error. So not suitable for high frequency core loss measurement APEC 2012 9 Outlines • Overview • Existing core loss measurement methods • Proposed core loss measurement methods – For sinusoidal excitations – For general excitations • Conclusion February 8 , 2012 Orlando, FL APEC 2012 10 Proposed New Method I -- capacitive cancellation IR V3 1 V3 = ( Rcore + jω Lm − j )IR ω Cr 1 T Pcore = ∫ V3 I R dt T 0 Cr is used to cancel the observed reactive voltage seen by the probe. * Mingkai Mu, Qiang Li, David Gilham, Fred C. Lee, Khai D.T. Ngo, “New core loss measurement method for high frequency magnetic materials”, Energy Conversion Congress and Exposition, Sept. 2010, pp. 4384 – 4389. February 8 , 2012 Orlando, FL APEC 2012 11 Proposed New Method I (cont’) -- capacitive cancellation Pcore 1 = TRref T ∫ v3vR dt 0 VR V3 % Power Error Tune the capacitor value 1° phase discrepancy Phase angle Phase Phase angle between V2 and IR angle between V3 and IR Given 1 ° phase discrepancy, 30° phase angle is enough to reduce phase error below 1% February 8 , 2012 Orlando, FL APEC 2012 12 Measurement Setup Core material: NiZn ferrite (4F1) Number of Turns: 14:14 Core Cross section area: A=5.98*10-6m2 Average flux path length: l=7.19 cm Osc-scope: Tek TDS7054 Hot plate core under test immersed in thermal oil Oscilloscope Power Input February 8 , 2012 Orlando, FL APEC 2012 13 Working Waveform V2 V2 (secondary side) VR VR (current sensing) V3 V3 (cancelled voltage) Lm Cr February 8 , 2012 Orlando, FL APEC 2012 14 Measurement Result @ 10MHz 4F1 (NiZn Ferrite) @100 °C 4 Core Loss Density Pv (kW/m3) 10 3 10 2 10 1 10 1 2 3 4 5 6 7 8 910 20 30 Peak Flux Density Bm (mT) + 。 February 8 , 2012 Orlando, FL Datasheet Result@100°C Without Cap (conventional method) @100°C With Cap (new method) @ 100°C APEC 2012 15 Factors May Influence Accuracy • Phase Error – This error can be minimized when Cr resonates at excitation frequency. • Resonant Cap ESR – An additional resistor is in series with the Rc, Lm, C branch. • Parasitic Capacitor – These parasitic capacitors (inter- and intra-winding capacitor and probe loading effect) will introduce a small current on the secondary side. February 8 , 2012 Orlando, FL APEC 2012 16 Improved Circuit to Reduce the Probe Loading Effect Ll1 i1 Ll 2 Lm Cr C probe Measure the voltage on resonant cap, instead of the voltage of transformer’s secondary. So it merges the parasitic cap into the resonant cap, and has less loading effect by probe input cap. February 8 , 2012 Orlando, FL APEC 2012 17 Test Setup with DC Bias Field Choke Choke + _ AC voltage Add Winding for DC Pre-magnetization Two Choke inductance should be large enough to block AC current 2Lchoke >> Lm February 8 , 2012 Orlando, FL APEC 2012 18 Improved Scheme + _ AC voltage _ AC voltage + The AC voltage seen by the DC current source is minimized. Yongtao Han, Yan-Fei Liu, "A Practical Transformer Core Loss Measurement Scheme for High-Frequency Power Converter", IEEE Transactions on Industrial Electronics, 2008, pp. 941-948. Mingkai Mu, Yipeng Su, Qiang Li, Fred C. Lee, "Magnetic Characterization of Low Temperature Co-fired Ceramic (LTCC) Ferrite Materials for High Frequency Power Converters", Energy Conversion Congress and Exposition, Sept. 2011, pp. 2133 - 2138. February 8 , 2012 19 APEC 2012 Orlando, FL Limited for Sinusoidal Excitation Ll1 Ll 2 * Lm 1 V3 = ( Rcore + jω Lm − j )IR ω Cr The capacitor can only cancel the reactive voltage at a single frequency. February 8 , 2012 Orlando, FL APEC 2012 20 Proposed New Method II -- Inductive Cancellation Ll1 Ll 2 Lm * V3 = (− Rcore − jω Lm + jω L) I R Phase angle V2 VR V3 Pcore N1 f = N 2 Rref ∫ v v dt 3 R T Using inductor, we can still cancel the observed reactive voltage. Mingkai Mu, Fred C. Lee, Qiang Li, David Gilham, Khai D.T. Ngo, "A high frequency core loss measurement method for arbitrary excitations", Applied Power Electronics Conference, 2011, pg.157-162. February 8 , 2012 21 APEC 2012 Orlando, FL Proposed Method II (cont’) -- Compatible with Arbitrary Waveform Ll1 Ll 2 Lm * V3 = (− Rcore − jω Lm + jω L) I R When L = Lm, the reactive voltage is cancelled out, leaving only the voltage on resistor. For this method, a high Q inductor is needed in order to reduce the error introduced by the loss of the inductor. February 8 , 2012 Orlando, FL APEC 2012 22 Using Air Core or Low-loss Core Transformer’s Magnetizing Inductor Ll1 Core under test Ll 2 Air core transformer or Low core loss transformer Using the magnetizing inductance of air core or low loss core to eliminate the error caused by winding resistance. From another perspective, the principle is to compare the core loss of DUT with the reference core. February 8 , 2012 Orlando, FL APEC 2012 23 Reference Core Choice 1: Air core Advantage: no core loss Disadvantage: larger parasitic capacitance and inductance Choice 2: Low loss core Advantage: smaller parasitics Disadvantage: hard to find if measure low loss material not convenient to adjust inductance value. Choice 3: Low loss core with adjustable air gap Advantage: smaller parasitics easy to adjust inductance value Disadvantage: hard to find if measure low loss material February 8 , 2012 Orlando, FL APEC 2012 24 Excitation The circuit is excited by the function generator signal amplified by power amplifier. The power amplifier has 50Ω internal impedance, so an impedance transformer is needed to reduce the voltage drop on the internal impedance. Programmable function generator is used to generate desired waveforms. February 8 , 2012 Orlando, FL APEC 2012 25 Test Example Core-under-test and Reference Core Core Loss @ 1.5MHz Sinusoidal Excitation* 4 10 3 10 2 10 LTCC 40011 (ESL®) 4F1 (Ferroxcube®) Core sample Reference Toroid LTCC 40011 n1:n2=10:10 Ae=3.82mm2 le=64.5mm ve=246.4mm3 Toroid 4F1 N1:N2=9:9 Ae=11.88mm2 le=24.5mm ve=291mm3 1 10 0 10 BLTCC B4 F 1 1 10 2 10 1 vLTCC dt ∫ nA = 1 eLTCC = 2.8 1 v4 F 1dt ∫ N1 Ae 4 F 1 Core loss in 4F1 core is much smaller than LTCC 40011. February 8 , 2012 Orlando, FL *The core loss data is measured using the capacitive cancellation. 26 APEC 2012 Measurement Setup 4F1 Test circuit LTCC 40011 in oil bath Hot plate Impedance transformer Test LTCC 40011 using 4F1 February 8 , 2012 Orlando, FL APEC 2012 27 Measurement Waveforms Sinusoidal Voltage V2 (secondary side) VR (current sensing) V3 (cancelled voltage) February 8 , 2012 Orlando, FL APEC 2012 28 Core Loss of Sinusoidal Excitation Measured core loss with capacitive and inductive cancellation 4 10 3 10 2 10 Method with Capacitor Method with Transformer 1 10 3 4 5 6 7 8 9 10 20 30 Peak Flux Density B (mT) *1.5MHz, 100°C February 8 , 2012 Orlando, FL APEC 2012 29 Measurement Waveforms Rectangular Voltage (50% Duty Cycle) V2 (secondary side) VR (current sensing) V3 (cancelled voltage) February 8 , 2012 Orlando, FL APEC 2012 30 Measurement Waveforms Rectangular Voltage (25% Duty Cycle) V2 (secondary side) VR (current sensing) V3 (cancelled voltage) February 8 , 2012 Orlando, FL APEC 2012 31 Core Loss of LTCC 40011 Rectangular Voltage Waveforms of Different Duty Cycles 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0 February 8 , 2012 Orlando, FL 0.2 APEC 2012 0.4 0.6 0.8 1 32 Conclusion • A new series of high frequency core loss measurement methods are proposed, which excludes winding loss and reduce phase discrepancy sensitivity. • The principle is using capacitor or inductor to cancel the reactive voltage seen by the probe to reduce the sensitivity to phase error. The capacitor version is good for sinusoidal excitation; the inductor version is broad-band, so it can measure the core loss of general waveforms. February 8 , 2012 Orlando, FL APEC 2012 33 Conclusion (cont’) • For capacitor version, high Q capacitor is needed to reduce the ESR induced error. For inductor version, the reference core is important for this measurement. Air core, low loss core or air gapped core should be selected according to the core under test. Adjustable capacitor or transformer will make the measurement more easier. • With this new series of methods, high frequency core loss can be accurately measured at different excitations. The core loss measurement problem can be addressed for power electronics applications and material science research. February 8 , 2012 Orlando, FL APEC 2012 34 Q&A Thank you! mmk@vt.edu February 8 , 2012 Orlando, FL APEC 2012 35