High Frequency Magnetics; Black Magic, Art or Science?
Magnetics Core Loss
Dr. Ray Ridley
APEC Industry Session on Magnetics
Wednesday March 23 2016
www.ridleyengineering.com
1
Introduction
Core loss calculations and measurements
New core material needs
Need for extensive data
Approximations to data
Different excitation waveforms
Temperature variations
Standardized data base proposals
© 2016 Ridley Engineering Inc
2
Immediate Goals for Core Materials
0.6 T
MPP
SATURATION
LOSS
0.3 T
FERRITE
Improvements
SATURATION LIMITED CORE
LOSS LIMITED
10K
100K
1 M
Frequency (Hz)
10 M
100 M
Would be nice to at least extend saturation range further
© 2016 Ridley Engineering Inc
3
B-H Loop and Core Losses
B
𝛥𝐵
H
Core loss calculations and measurements
© 2016 Ridley Engineering Inc
4
Core Loss MPP 200u
5000 mW/cm3
300 kHz 0.1 T 5 W/cm3
© 2016 Ridley Engineering Inc
5
Core Loss PC95
350 mW/cm3
300 kHz 0.1 T 0.350 W/cm3 > 10 times better than MPP
© 2016 Ridley Engineering Inc
6
Core Loss Steinmetz Equation
Pc  kf B
x
y
Steinmetz equation can
be used to approximate
the actual loss measurements
Three coefficients define the
loss for a given material
© 2016 Ridley Engineering Inc
7
Steinmetz Equation Limitations
Pc  kf x B y
Equation assumes curves are
1) Equally spaced with frequency
2) Equal slopes at different frequencies
© 2016 Ridley Engineering Inc
8
Modifying the Steinmetz Equation
Pc  kf x B y
Discrete step changes in coefficients from Mag Inc.
<100 kHz
100-500 kHz
>500 kHz
© 2016 Ridley Engineering Inc
a
c
d
0.074
0.036
0.014
1.43
1.64
1.84
2.85
2.62
2.28
9
Steinmetz Equation Changing Coefficients
Pc  kf x B y
<100 kHz
100-500 kHz
>500 kHz
© 2016 Ridley Engineering Inc
k
x
y
0.074
0.036
0.014
1.43
1.64
1.84
2.85
2.62
2.28
10
Ridley-Nace Variable Steinmetz Equation
Continuously-variable coefficients with Ridley-Nace formula
Pc  (a ln f  b) f B
c
( df  e )
Five coefficients needed to describe materials
Example for Magnetics R material:
Pc  (3.626 ln f  28.32) f
© 2016 Ridley Engineering Inc
1.729
B
( 0.00076 f  2.8332)
11
Continuously Variable k Term
Magnetics R Material Measured Core Loss
20
10000
Loss
mW / cm3
k  3.626 ln f  28.32
16
1000
12
100
8
10
Measurements
4
0
1
0
200
400
600
800
1000
Frequency
Pc  (a ln f  b) f B
c
0.1
0.01
0.005
( df  e )
a  3.626 b  28.32
0.01
Flux Density T
0.1
0.2
12
Continuously Variable y Term
Magnetics R Material Measured Core Loss
10000
3
Loss
mW / cm3
1000
y  0.00076 f  2.8332
2
100
1
10
0
1
0
200
400
600
800
1000
Frequency
0.1
Pc  (a ln f  b) f B
c
0.01
0.005
0.01
© 2016 Ridley Engineering Inc
Flux DensityT
0.1
0.2
d  0.00076
( df  e )
e  2.8332
13
Oliver Variable Steinmetz Equation
(Powdered Iron Core Material)
Pcore
f

 df 2 B 2
a
b
c
 2.3  1.65
3
B B
B
Continuously-variable coefficients with Oliver formula
Four coefficients needed to describe materials, plus 2.3 and
1.65 exponents subject to change with different materials
www.micrometals.com/appnotes/appnotedownloads/coreloss update.pdf
© 2016 Ridley Engineering Inc
14
Core Loss Temperature Variation
Loss Multiplier
5
g (T )  a 5T 5  a 4T 4  a 3T 3  a 2T 2  a1T  a 0
4
Manufacturer’s Data
3
2
1
-60
0
60
120
Temperature
Six coefficients needed to describe temperature variation
Every material has a different curve
Does the curve also change with frequency and flux level?
© 2016 Ridley Engineering Inc
15
Different Core Excitations
B
B
t
t
B
B
t
t
Need to modify equations to suit each of these waveforms
© 2016 Ridley Engineering Inc
16
General Triangular Flux Waveform
B
2B
D2Ts
D1Ts
Pc ( D)  D1 Pc
© 2016 Ridley Engineering Inc
f
2 D1
 D2 Pc
f
2 D2
Ts
t
Pc  (a ln f  b) f c B ( df  e )
17
Duty Cycle Core Loss Multiplier
B = 0.15 T
6
5
DCM : D '  D
4
3
2
1
0
CCM : D '  1  D
0
0.1
0.2
Pc ( D)  D1 Pc
© 2016 Ridley Engineering Inc
0.3
f
2 D1
0.4
 D2 Pc
0.5
0.6
f
2 D2
18
Putting it All Together
© 2016 Ridley Engineering Inc
19
What We Need from the Manufacturers
1) Raw core loss sinewave test data over wide range of
frequency, B, temperature
2) Who is going to fund the modeling into a standard
format database?
20
Magnetics Forecast
High-ripple current inductor designs will dominate the practical high-performance market
Practical “high performance” means converters up to 5 MHz
Core material improvements will be just incremental (unless some breakthrough material emerges)
Insufficient funding is going into fundamental magnetic material research to give a
good chance of any breakthroughs.
Data Standardization is badly needed from the core manufacturers to ease confusion
Core loss raw data is needed, not just curves.
Creative core geometries need to be applied to POL inductors and other high-ripple parts
Magnetic core will not go away (however much it is wished for)
Isolation transformers will creep back into PoL applications to boost efficiency – More magnetics!
Multi-converter processing will continue to proliferate.
© 2016 Ridley Engineering Inc
21
Some References
Cliff Jamerson:
Targeting Switcher Magnetics Core Loss Calculations,
Power Electronics Technology, Jan 2001.
Ed Herbert:
http://www.psma.com/technical-forums/magnetics/core-loss-studies
Christopher Oliver: www.micrometals.com/appnotes/appnotedownloads/coreloss update.pdf
Power Supply Design Center Articles http://www.ridleyengineering.com/design-center.html
[89] Core Loss Modeling
[90] Core Loss Modeling with Non-Sinusoidal Waveforms - Part II
[91] Core Loss Modelling - Sinewave Versus Triangle Wave - Part III
Ray Ridley and Art Nace
[A03] Modeling Ferrite Core Losses
Christopher Oliver
[A06] Core Loss Modeling & Measurement
Power Supply Design Software http://www.ridleyengineering.com/software.html
© 2016 Ridley Engineering Inc
22