Power Design
Power Design
kHz to 3 MHz, due to the combination of low core cost and low core losses.
Ferrite is an excellent material for high frequency (20 kHz to 3 MHz) inverter power supplies.
Ferrites may be used in the saturating mode for low power, low frequency operation (<50 watts
and 10 kHz). For high power operation a two transformer design, using a tape wound core as the
saturating core and a ferrite core as the output transformer, offers maximum performance. The two
transformer design offers high efficiency excellent frequency stability, and low switching losses.
Section 4
Power
Design
Ferrite cores may also be used in fly-back transformer designs, which offer low core cost, low circuit cost
and high voltage capability. Powder cores (MPP, High Flux, Kool M) offer soft saturation, higher Bmax
and better temperature stability and may be the best choice in some flyback applications or inductors.
High frequency power supplies, both inverters and converters, offer lower cost, and lower weight
and volume than conventional 60 hertz and 400 hertz power sources.
Many cores in this section are standard types commonly used in the industry. If a suitable size for
your application is not listed, Magnetics will be happy to review your needs, and, if necessary,
quote tooling where quantities warrant.
Cores are available gapped to avoid saturation under dc bias conditions. J and W materials are
available with lapped surfaces.
Bobbins for many cores are available from Magnetics. VDE requirements have been taken into account in
bobbin designs for EC, PQ and metric E Cores. Many bobbins are also available commercially.
4.1
EP CORES
F, P, and R materials, offering the lowest core losses and highest saturation flux density,
are most suitable for high power/high temperature operation. P material core losses
decrease with temperature up to 70C; R material losses decrease up to 100C.
EP Cores are round center-post cubical shapes which enclose the coil completely
except for the printed circuit board terminals. The particular shape
minimizes the effect of air gaps formed at mating surfaces in the magnetic
path and provides a larger volume ratio to total space used. Shielding
is excellent.
J and W materials offer high impedance for broad transformers, and are
also suitable for low-level power transformers.
i (20 gauss)
p (2000 gauss)
Saturation
Flux Density
(Bm Gauss)
Core Loss (mw/cm3)
(Typical)
@100 kHz, 1000 Gauss
FERRITE
POWER MATERIALS SUMMARY
F
P
R
25C
3,000 2,500 2,300
100C 4,600 6,500 6,500
25C
4,900 5,000 5,000
J
5,000
5,500
4,300
W+
10,000
12,000
4,300
100C
3,700
3,900
3,700
2,500
2,500
25C
60C
100C
100
180
225
125
80*
125
140
100
70
*@80C
+@10kHz
CORE GEOMETRIES
POT CORES
Pot Cores, when assembled, nearly surround the wound bobbin. This aids in
shielding the coil from pickup of EMI from outside sources. The pot core
dimensions all follow IEC standards so that there is interchangeability
between manufacturers. Both plain and printed circuit bobbins are
available, as are mounting and assembly hardware. Because of its design,
the pot core is a more expensive core than other shapes of a comparable
size. Pot cores for high power applications are not readily available.
DOUBLE SLAB AND RM CORES
Slab-sided solid center post cores resemble pot cores, but have a section cut
off on either side of the skirt. Large openings allow large size wires to be
accommodated and assist in removing heat from the assembly. RM cores
are also similar to pot cores, but are designed to minimize board space,
providing at least a 40% savings in mounting area. Printed circuit or plain
bobbins are available. Simple one piece clamps allow simple assembly. Low
profile is possible. The solid center post generates less core loss and this
minimizes heat buildup.
4.2
PQ CORES
PQ cores are designed especially for switched mode power supplies. The
design provides an optimized ratio of volume to winding area and surface
area. As a result, both maximum inductance and winding area are possible
with a minimum core size. The cores thus provide maximum power output
with a minimum assembled transformer weight and volume, in addition to
taking up a minimum amount of area on the printed circuit board. Assembly
with printed circuit bobbins and one piece clamps is simplified. This efficient
design provides a more uniform cross-sectional area; thus cores tend to operate
with fewer hot spots than with other designs.
E CORES
E cores are less expensive than pot cores, and have the advantages of simple bobbin
winding plus easy assembly. Gang winding is possible for the bobbins used
with these cores. E cores do not, however, offer self-shielding. Lamination
size E shapes are available to fit commercially available bobbins previously
designed to fit the strip stampings of standard lamination sizes. Metric and
DIN sizes are also available. E cores can be pressed to different thickness,
providing a selection of cross-sectional areas. Bobbins for these different
cross sectional areas are often available commercially.
E cores can be mounted in different directions, and if desired, provide a lowprofile. Printed circuit bobbins are available for low-profile mounting. E
cores are popular shapes due to their lower cost, ease of assembly and
winding, and the ready availability of a variety of hardware.
PLANAR E CORES
Planar E cores are offered in all of the IEC standard sizes, as well as a number of other sizes. Magnetics R material is perfectly suited to planar
designs due to its low AC core losses and minimum losses at 100C. Planar
designs typically have low turns counts and favorable thermal dissipation
compared with conventional ferrite transformers, and as a consequence the
optimum designs for space and efficiency result in higher flux densities. In
those designs, the performance advantage of R material is especially significant.
The leg length and window height (B and D dimensions) are adjustable for
specific applications without new tooling. This permits the designer to
adjust the final core specification to exactly accommodate the planar conductor stack height, with no wasted space. Clips and clip slots are avail-
MAGNETICS
able in many cases, which is especially useful for prototyping. I-cores are
also offered standard, permitting further flexibility in design. E-I planar
combinations are useful to allow practical face bonding in high volume
assembly, and for making gapped inductor cores where fringing losses must
be carefully considered due to the planar construction.
TOROIDS
SUMMARY
These shapes are a cross between E cores and pot cores. Like E cores, they
provide a wide opening on each side. This gives adequate space for the
large size wires required for low output voltage switched mode power
supplies. It also allows for a flow of air which keeps the assembly cooler.
The center post is round, like that of the pot core. One of the advantages
of the round center post is that the winding has a shorter path length
around it (11% shorter) than the wire around a square center post with an
equal area. This reduces the losses of the windings by 11% and enables the
core to handle a higher output power. The round center post also eliminates
the sharp bend in the wire that occurs with winding on a square center post.
Ferrite geometries offer a wide selection in shapes and sizes. When choosing a core
for power applications, parameters shown in Table 1 should be evaluated.
Toroids are economical to manufacture; hence, they are least costly of all comparable
core shapes. Since no bobbin is required, accessory and assembly costs are nil.
Winding is done on toroidal winding machines. Shielding is relatively good.
POT
CORES
6
High
Low
Low
Good
Simple
Good
Poor
Excellent
DOUBLE SLAB,
RM CORES
7-8
High
Low
Low
Good
Simple
Good
Good
Good
EP
CORES
9
Medium
High
Low
Good
Simple
Good
Poor
Excellent
PQ
CORES
10
High
High
Low
Good
Simple
Fair
Good
Fair
E
CORES
11
Low
Low
Low
Excellent
Simple
Good
Excellent
Poor
EC, ETD,
EER, ER CORES
12
Medium
Medium
Low
Excellent
Medium
Fair
Good
Poor
TOROIDS
13
Very Low
None
High
Fair
None
Poor
Good
Good
** Hardware is required for clamping core halves together and mounting assembled core on a circuit board or chassis.
mag-inc.com
4.3
Transf_SIze_Select.eps
General Formulas
TRANSFORMER CORE SIZE SELECTION
FIGURE 1
With larger wire sizes, and/or higher voltages, these K factors may not be obtainable.
To minimize both wire losses and core size, the window area must be full.
NOTE: For Wire Tables and turns/bobbin data, refer to pgs 5.8.
Where:
K=winding factor
I=current (rms)
P i=input power
P o=output power
e=transformer efficiency
4.4
MAGNETICS
P i=El
We obtain the basic relationship between output power and the WaAc product:
8
WaAc = k P 0 x 10 , Where k = C
4eK
Bf
For square wave operation
k = .00633 for toroids, k = .00528 for pot cores, k = .00528 for E-U-I cores
A core selection chart (Table 3) using WaAc can be found on page 4.7. In
addition a A core selection procedure which varies by topology can also be
found on page 4.8. This procedure is based on the book Switching Power
Supply Design by A.I. Pressman. While the formula above allows WaAc to be
adjusted based on selected core geometry, the Pressman approach uses topology as the key consideration and allows the designer to specify current density.
(2)
(3)
(4)
GENERAL INFORMATION
An ideal transformer is one that offers minimum core loss while requiring
the least amount of space. The core loss of a given core is directly effected by the flux density and the frequency. Frequency is the most important
characteristic concerning a transformer. Faradays Law illustrates that as
frequency increases, the flux density decreases proportionately. Core losses decrease more when the flux density drops than when frequency rises.
For example, if a transformer were run at 250 kHz and 2 kG on R material at 100C, the core losses would be approximately 400 mW/cm 3 . If the
frequency were doubled and all other parameters untouched, by virtue of
Faradays law, the flux density would become 1kG and the resulting core
losses would be approximately 300mW/cm 3 .
Typical ferrite power transformers are core loss limited in the range of 50200mW/cm 3 . Planar designs can be run more aggressively, up to 600
mW/cm 3 , due to better power dissipation and less copper in the windings.
CIRCUIT TYPES
FIGURE 3
Feed forward
Flyback
20
30
40
ADVANTAGES
Medium to high power
Efficient core use
Ripple and noise low
Medium power
Low cost
Ripple and noise low
Lowest cost
Few components
60 80 100
200
300 400
600
Above 20 kHz, core losses increase. To operate the SPS at higher frequencies, it
is necessary to operate the core flux levels lower than 2 kg. Figure 3 shows
the reduction in flux levels for MAGNETICS P ferrite material necessary to
maintain constant 100mW/cm3 core losses at various frequencies, with a
maximum temperature rise of 25C.
FIGURE 3
DISADVANTAGES
More components
Core use inefficient
Ripple and noise high
Regulation poor
Output power limited
(< 100 watts)
1000
Frequency kHz
PUSH-PULL CIRCUIT
A typical push-pull circuit is shown in Figure 2A. The input signal is the output of an IC
FEED FORWARD CIRCUIT
network, or clock, which switches the transistors alternately ON and OFF. High frequency Hysterasis_Loop_Push_Pull.eps
square waves on the transistor output are subsequently rectified, producing dc.
E1
E1 0
In the feed forward circuit shown in Figure 4A, the transformer operates in
the first quadrant of the Hysteresis Loop. (Fig 4B). Unipolar pulses applied
to the semiconductor device cause the transformer core to be driven from
its B R value toward saturation. When the pulses are reduced to zero, the
core returns to its B R value. In order to maintain a high efficiency, the
primary inductance is kept high to reduce magnetizing current and lower
wire losses. This means the core should have a zero or minimal air gap.
mag-inc.com
Feed_Forward_Circuit.eps
4.5
Flyback_
Hysteresis_Loop_Feed_Forward.eps
FIGURE 6A
TYPICAL FLYBACK REGULATOR CIRCUIT
Hysteresis_Loop_Flyb
E1
E1 O
FIGURE 6B
Figure5_Powersec.eps
HYSTERESIS LOOP OF MAGNETIC CORE IN FLYBACK CIRCUIT
For ferrites used in this circuit, B (or B max-B R ) is typically 2400 gauss or
B (as applied
to Equation 4) is 1200 gauss as shown in Figure 4B. In the
B
push-pull circuit, it was recommended that the peak flux density in the core
should not exceed B = 2000
gauss in order to keep core losses small.
B
Because of the constraints of the Hysteresis Loop, the core in the feed
B
forward circuit should not exceed
a peak value of B = 1200 gauss.
H
Core selection for a feed forward circuit is similar to the push-pull circuit
except that B for Equation 4 is now limited to 1200 gauss.
If the transformer operating temperature is above 75, the value of B will be further
reduced. Figure 5 shows the variation of B with temperature. Therefore the
recommended B value of 2400 (B= 1200) gauss has to be reduced, the
amount depending on the final projected temperature rise of the device.
FIGURE 5
In most designs, the air gap is large; therefore, BR is small as noted on the
Hysteresis Loop in Figure 6B and can be considered zero. The maximum flux
density available is approximately 3600. This means B is 3600 or B =
1800 gauss. Core selection for this circuit can be done using Equation 4. The
B value in Equation 4 is 1800 gauss at 20 kHz and is used until a higher
frequency (Figure 3) dictates a lower B required.
GENERAL FORMULA CORE SELECTION FOR DIFFERENT TOPOLOGIES
The following formula has been gained from derivations in Chapter 7 of A.I.
Pressmans book Switching Power Supply Design (see Reference No. 13, pg 14.4.)
A typical schematic is shown in Figure 6A. Unipolar pulses cause dc to flow through
the core winding, moving the flux in the core from BR towards saturation (Fig. 6B).
When the pulses go to zero the flux travels back to BR as in the feed forward design.
However, the difference between the feed forward and the flyback circuit is that the
flyback requires the transformer to act as an energy storage device as well as to
perform the usual transformer functions. Therefore, to be an effective energy storage
unit, the core must not saturate and is usually a gapped structure.
4.6
MAGNETICS
P D
WaAc = o cma
K t B max f
WaAc
Po
Dcma
B max
f
Kt
=
=
=
=
=
=
For individual cores, WaAc is listed in this catalog under Magnetic Data. Choice
of Bmax at various frequencies, Dcma and alternative transformer temperature rise
calculation schemes are also discussed in Chapter 7 of the Pressman book.
RS,DS,HS RM, EP
See Section
0.001
0.002
6
7
40704
40905
8/9
0.004
0.007
0.010
41107
41110(RM)
41010(EP)
0.020
41408
41408
(RS,DS)
0.040
0.070
0.100
0.200
RM SOLID PQ
ETD, EER
EC
TC
11
12
12
13
40601
40603
10
40707 (EP)
11
41309 (EE)
40705
41203
41510(RM) 41510
41313(EP)
41205
41717(EP)
42610 41808
42316(RM) 42316
42016 41810
42614 42510
42020
42620
43214
44308(EI)
46410(EI)
10.00
44022
45021
45528
46016
45530
20.00
48020
40.00
100
49928
43622 43622
(RS,DS,HS)
2.00
44229 44229
44529 (RS,DS,HS)
4.00
7.00
43723(RM)
42625 42520
43220 43515
43723
43230 44317
43535 44721
44040 45724
41003
41106(UU)
40907
41005
41303
42216(EE)
43520
43524
44011
44020
44924
1.00
42819
41106 (UI)
41206
41305
41306
41605
43618(EI)
43208(EI)
0.700
42819(RM)
42120(EP)
41208
41209
41515
41707
41709
42110
42211
42810
43009
42523
42515
43007
43013
0.400
11
40904
40906
41812(RM) 41812
41811 42311
(RS,DS,HS)
42213 42318
(HS)
42616 42318
(RS,DS)
42616
(RS,DS,HS)
43019
(RS,DS,HS)
43019
11
42515
(UI)
41809
42206
43618(EE)
43208(EE)
44308(EE)
45810(EI)
42207
42220(UU) 43517
42512(UU)
42515(UU)
42530(UU) 44119
42507
43434
43521(EER)
44119(UU) 45224
43939
44121(UU) 44216(EER) 43615
44444
45032
44125(UU) 44949
44416
44130(UU)
45810(EE)
46409(EE)
46410(EE)
47035
47228
47054
49938(EE)
42908
43610
43813
44916
44925
46113
47313
47325
48613
49925(UU)
49925(UI)
*Bobbin window and core area product. For bobbins other than those in this catalog, WaAc may need to be recalculated.
mag-inc.com
4.7
@F=
POT-RS-RM DS
250KHZ CORES
CORES
6/7/8
7
7
41408-PC
EP
CORES
9
41313
11
21
41717
12
13
15
18
19
18
20
22
28
30
27
30
32
43
48
53
59
62
84
94
26
28
30
33
40
42
48
60
42
45
49
53
61
70
75
100
58
65
70
80
95
100
110
150
113
127
137
156
185
195
215
293
70
105
110
120
130
140
150
190
200
220
230
260
280
110
160
190
195
205
215
240
300
310
350
350
400
430
170
235
250
270
290
340
380
470
500
530
550
600
650
332
460
480
525
570
663
741
917
975
1,034
1,073
1,170
1,268
300
340
360
410
550
450
550
580
650
800
700
850
870
1,000
1,300
1,365
1,658
1,697
1,950
2,535
650
1,000
1,600
3,120
700
850
900
1,000
1,000
1,400
1,600
2,000
2,800
11,700
1,100
1,300
1,500
1,600
1,700
2,500
2,600
3,000
4,200
19,000
1,800
1,900
2,000
2,500
2,700
3,200
3,700
4,600
6,500
26,500
3,510
3,705
3,900
4,875
5,265
6,240
7,215
8,970
12,675
51,500
4.8
MAGNETICS
41811-PC
42311-RS
42809-RM
42316-RM
42213-PC
42318-RS
42311
PQ
CORES
10
41808
42016
42318
42616
E-CORES
11
41707
41810, 42211
42510
42020
LOW-PROFILE
PLANAR
EC-ETD
CORES
U CORES
11
12
41709
42107
42110
42610-PQ
42216-EC
42106
41809
42810, 42520
42515
42819-RM
42616-PC
42206
42109
42207
42620
43019
43019-RS
43007
42625
43019-PC
43723-RM
43220
43618-EC
44008-EC
43208-EC
43013
42530, 43009
43515 (E375)
43622
41306
41605
42614-PQ
43618-E, I
43208-E, I
44008-E, I
42120
TC
TOROIDS
13
41206
41303
43205
42212, 42507
43517 (EC35)
44308-E, I
44011 (E40)
43622-PC
43230
44119 (EC41)
43521
43524, 43520
44317 (E21)
44308-EC
43806
42915, 43113
43939 (ETD39)
44229
43610
44721 (E625)
45032
44020 (42/15)
44216
45021 (E50)
44924
44022 (42/20)
45224 (EC52)
43535
43813
43615
44229-PC
44529-PC
45810-EC
44444 (ETD44)
46410-E, I
43825
44949 (ETD49) 44416
44040
45724 (E75)
45528 (55/21)
46016 (E60)
45530 (55/25)
45810-EC
46409-EC
46410-EC
44715
44916
44920
44925
47035 (EC70)
45959 (ETD59) 46113
47228
44932
47313
48020
47054
49938-EC
48613
49925 (U)
Above is for push-pull converter. De-rate by a factor of 3 or 4 for flyback. De-rate by a factor of 2 for feed-forward converter.
NOTE: Assuming Core Loss to be Approximately 100mW/cm3,
B Levels Used in this Chart are:@ 20kHz-2000 gauss @ 50kHz-1300 gauss @ 100kHz-900 gauss @ 250kHz-700 gauss.
SEE PAGE 4.7 Area Product Distribution
TEMPERATURE CONSIDERATIONS
TRANSFORMER EQUATIONS
The power handling ability of a ferrite transformer is limited by either the saturation
of the core material or, more commonly, the temperature rise. Core material
saturation is the limiting factor when the operating frequency is below 20kHz.
Above this frequency temperature rise becomes the limitation.
Once a core is chosen, the calculation of primary and secondary turns and
wire size is readily accomplished.
Temperature rise is important for overall circuit reliability. Staying below a given
temperature insures that wire insulation is valid, that nearby active components do not
go beyond their rated temperature, and overall temperature requirements are met.
Temperature rise is also very important for the core material point of view.
As core temperature rises, core losses can rise and the maximum saturation flux
density decreases. Thermal runaway can occur causing the core to heat up to its
Curie temperature resulting in a loss of all magnetic properties and catastrophic
failure. Newer ferrite power materials, like P and R material, attempt to
mitigate this problem by being tailored to have decreasing losses to temperature of
70C and 100C respectively.
CORE LOSSOne of the two major factors effecting temperature rise is core
loss. In a transformer, core loss is a function of the voltage applied across the
primary winding. In an inductor core, it is a function of the varying current
applied through the inductor. In either case the operating flux density level, or B
level, needs to be determined to estimate the core loss. With the frequency and
B level known, core loss can be estimated from the material core loss curves. A
material loss density of 100mw/cm3 is a common operating point generating
about a 40C temperature rise. Operating at levels of 200 or 300 mw/cm3 can
also be achieved, although forced air or heat sinks may need to be used.
WINDING CONSIDERATIONSCopper loss is the second major contributor
to temperature rise. Wire tables can be used as a guide to estimate an
approximate wire size but final wire size is dependent on how hot the designer
allows the wire to get. Magnet wire is commonly used and high frequency copper loss
needs to be considered. Skin effects causes current to flow primarily on the
surface of the wire. To combat this, multiple strands of magnet wire, which
have a greater surface area compared to a single heavier gauge, are used.
Stranded wire is also easier to wind particularly on toroids. Other wire alternatives,
which increase surface areas, are foil and litz wire. Foil winding allows a very
high current density. Foil should not be used in a core structure with significant
air gap since excessive eddy currents would be present in the foil. Litz wire is
very fine wire bundled together. It is similar to stranded wire except the wire is
TO SMPS between the outside and the inside of the
woven to allow each strand to alternate
bundle over a given length.
8
Np = V p x 10
4BAf
Ns = Vs Np
Vp
l p = P in = P out
P in eE in
l s = P out
E out
Switch Mode Power Supplies (SMPS) normally generate excessive high frequency
noise which can affect electronic equipment like computers, instruments and
motor controls connected to these same power lines. An EMI Noise Filter inserted
between the power line and the SMPS eliminates this type of interference
(Figure 8). A differential noise filter and a common mode noise can be in series,
or in many cases, the common mode filter is used alone.
FIGURE 8
CORE GEOMETRYThe
core shape also affects temperature and those
DIFFERENTIAL
that dissipateNOISE
heatFILTER
well are desirable. E core shapes dissipate heat well.
Toroids, along with power shapes like the PQ, are satisfactory. Older
telecommunication shapes, such as pot cores or RM cores, do a poor job of
dissipating heat but do offer shielding advantages. Newer shapes, such as
planar cores, offer a large flat surface ideal for attachment of a heat sink.
mag-inc.com
MODE
LTER
Considerations
Com_Diff_Noise_Filter.ep
4.9
Z(ohms)
10,000
1,000
5
FIGURE 10
In a CMN filter, each winding of the inductor is connected in series with one of
the input power lines. The connections and phasing of the inductor windings are
such that flux created by one winding cancels the flux of the second winding. The
insertion impedance of the inductor to the input power line is thus zero, except
for small losses in the leakage reactance and the dc resistance of the windings.
Because
tan /i of the opposing fluxes, the input current needed to power the SMPS
therefore will pass through the filter without any appreciable power loss.
Common
100KHz
1MHz mode noise
10MHz is defined as unwanted high frequency current that
appears in one or both input power lines and returns to the noise source
FREQUENCY
through the ground of the inductor. This current sees the full impedance of either
one or both windings of the CMN inductor because it is not canceled by a return
current. Common mode noise voltages are thus attenuated in the windings of the
inductor, keeping the input power lines free from the unwanted noise.
i
100
10KHz
Inductor Design
Imped_vs_Freq.eps
Figure 11 shows total impedance vs. frequency for two different materials.
J material has a high total impedance over the range of 1 to 20MHz. It is
most widely used for common mode filter chokes. Under 1MHz, W material
has 20-50% more impedance than J. It is often used in place of J when low
frequency noise if the major problem. For filter requirements specified at
frequencies above and below 2MHz, either J or W is preferred.
FIGURE 11
FIGURE 9
100MHz
CORE SHAPE
The wound unit reaches its highest impedance between 1 and 10MHz.
The series inductive reactance X s and series resistance R s (functions of
the permeability and loss factor of the material) together generate the total
impedance Z t .
Figure 10 shows permeability and loss factor of the ferrite material in
Figure 9 as a function of frequency. The falling off of permeability above
750kHz causes the inductive reactance to fall. Loss factor, increasing with
frequency, cause the resistance to dominate the source of impedance at high
frequencies.
Additional detailed brochures and inductors design software for this application are available from Magnetics.
4.10
MAGNETICS
Toroids are most popular for a CMN filter as they are inexpensive and have
low leakage flux. A toroid must be wound by hand (or individually on a
toroid winding machine). Normally a non-metallic divider is placed between
the two windings, and the wound unit is epoxied to a printed circuit header
for attaching to a pc board.
An E core with its accessories is more expensive than a toroid, but assembly
into a finished unit is less costly. Winding E core bobbins is relatively inexpensive.
Bobbins with dividers for separating the two windings are available for pc
board mounting.
E cores have more leakage inductance, useful for differential filtering in a
common mode filter. E cores can be gapped to increase the leakage inductance,
providing a unit that will absorb both the common mode and differential
unwanted noise.
cmf_li_vs_ap_400amp.eps
CORE 2 SELECTION
I vs AP at 400 amps/cm
F
J
The following is a design procedure
for a toroidal, single-layer common mode inductor,
see Figure 12. To minimize winding capacitance and prevent core saturation due to
W layer design is often used. This procedure assumes a miniasymmetrical windings, a single
mum of thirty degrees of free spacing between the two opposing windings.
The basic parameters needed for common mode inductor design are current (I), impedance
(Zs), and frequency (f). The current determines the wire size. A conservative current density of
400 amps/cm2 does not significantly heat up the wire. A more aggressive 800 amps/cm2
may cause theHwire to run hot. Selection graphs for both levels are presented.
The impedance of the inductor is normally specified as a minimum at a given frequency.
This frequency is usually low enough to allow the assumption that the inductive reactance,
Xs, provides the impedance, see Figure 9. Subsequently, the inductance, Ls can be
calculated from:
Ls = Xs
2f
cmf_li_vs_ap_800amp.eps
(1)
With the inductance and current known, Figures 13 and 14 can be used to select a core
size based on the LI product, where L is the inductance in mH and I is the current in
amps. The wire size (AWG) is then calculated using the following equation based on
the current density (Cd) of 400 or 800 amps/cm2:
10
100
( )
1.889I
Cd
(millihenry amps)
AWG = -4.31x In
The2 number
AP at 800 amps/cm
(2)
FIG. 14: CORE SELECTION AT 800 amps/cm2
DESIGN EXAMPLE
Inductor Design
10
100
lihenry amps)
mag-inc.com
4.11
Inductor Design
HALL EFFECT DEVICES
Edwin H. Hall observed the Hall Effect phenomenon at John Hopkins University
in 1897. He monitored the current flowing from top to bottom in a thin
rectangular strip of gold foil by measuring the voltages at the geometric center
of the left edge and the right edge of the strip. When no magnetic field was
present, the voltages were identical. When a magnetic field was present
perpendicular to the strip, there was a small voltage difference of a predictable
polarity and magnitude. The creation of the transverse electric field, which is
perpendicular to both the magnetic field and the current flow, is called the Hall
Effect or Hall Voltage.
In metals the effect is small, but in semiconductors, considerable Hall voltages can be
developed. Designers should consider using Hall sensors in many applications where
mechanical or optical sensors have traditionally been used. To monitor ac or dc current
flow in a wire, the wire is wrapped around a slotted ferromagnetic core, creating an
electromagnet. The strength of the resulting magnetic field is used by the Hall sensor,
inserted in the air gap, to measure the magnitude and direction of current flowing in
the wire.
CORE SELECTION
In all cases, the effective permeability of a gapped core will be a function of the size
of the air gap and the initial permeability of the core material. Once the gap becomes
greater than a few thousandths of an inch, the effective permeability is determined
essentially by the air gap.
ANALYTICAL METHOD
1. Determine the flux operating extremes based on either the V/B of the circuit
(volts/gauss), or the maximum flux sensitivity (gauss) of the sensor (as provided
by the sensor data sheet).
Gapped
2. Choose
a Toroid
coreSelector
basedCharton the maximum or minimum dimension requirements to
allow windings, and based on the core cross-section dimensions. The cross-section
dimensions should be at least twice the gap length to ensure a relatively
homogeneous flux distribution bridging the gap.
0.6
0.5
(1)
flux density (gauss)
path length (cm)
turns
current (amps peak)
44
91
0.3
61
0.4
(2)
41
60
42
20
6
42 425
90 07
8
0.2
41
50
0.1
4.12
0.0
0.00
0.05
MAGNETICS
0.10
0.15
Gap Lentgh (inches)
0.20
0.25
1. Calculate NI/B (amp turns per gauss), knowing the flux operating extremes of
V/B or the maximum B sensitivity of the sensor.
2. Using Figure 15, follow the NI/B value from the vertical axis to the diagonal line
to choose a ferrite core size. Drop down from the diagonal line to the horizontal
axis to determine the gap length. The core sizes indicated on the selector chart take
into account gap length versus cross-section dimensions in order to maintain an
even flux distribution across the gap under maximum current.
TOROID GAPPING
The value calculated for (i) is somewhat arbitrary and can be adjusted to
obtain a practical value for the inductance.The minimum capacitance is given by
C = i /8f min e o
(6)
(7)
INDUCTOR DESIGN
INDUCTOR CORE SIZE SELECTION (USING CORE
SELECTOR CHARTS) DESCRIPTION
(1)
COMPONENT SELECTION
(2)
(4)
(5)
Ferrite E cores and pot cores offer the advantages of decreased cost and low core
losses at high frequencies. For switching regulators, F or P materials are recommended
because of their temperature and dc bias characteristics. By adding air gaps to these
ferrite shapes, the cores can be used efficiently while avoiding saturation.
These core selection procedures simplify the design of inductors for switching
regulator applications. One can determine the smallest core size, assuming a winding
factor of 50% and wire current carrying capacity of 500 circular mils per ampere.
Only two parameters of the two design applications must be known:
(a) Inductance required with dc bias
(b) dc current
1. Compute the product of LI 2 where:
L= inductance required with dc bias (millihenries)
I= maximum dc output current - I o max + i
2. Locate the LI 2 value on the Ferrite Core Selector charts on pgs
4.154.18. Follow this coordinate in the intersection with the first core
size curve. Read the maximum nominal inductance, A L , on the Y-axis.
This represents the smallest core size and maximum A L at which
saturation will be avoided.
3. Any core size line that intersects the LI 2 coordinate represents a workable
core for the inductor of the cores A L value is less than the maximum
value obtained on the chart.
Inductor Design
L
AL
where L is in millihenries
5. Choose the wire size from the wire table on pg 5.8 using 500 circular
mils per amp.
mag-inc.com
4.13
Inductor Design
EXAMPLE
4.14
MAGNETICS
7. Given the A L, the number of turns needed for the required inductance is:
AL
Turns
250
21
315
19
400
17
8. Use #14 wire
Note: MAGNETICS Molypermalloy and Kool Mu powder cores have a distributed
air gap structure, making them ideal for switching regulator applications. Their dc
bias characteristics allow them to be used at high drive levels without saturating.
Information is available in Magnetics Powder Core Catalog and Brochure SR-IA,
Inductor Design in Switching Regulators.
FOR REFERENCES, SEE PAGE 14.4
dc_bias_pc_pot.eps
PC (POT) CORES
dc_bias_rs_ds.eps
A 40704
B 40905
C 41107
D 41408
E 41811
F 42213
G 42616
H 43019
J 43622
K 44229
L 44529
dc_bias_rm_ep.eps
Core Selection
A 41408 (RS)
B 42311 (DS, RS)
42318 (DS, RS)
C 42616 (DS)
D 43019 (DS, RS)
RM AND EP CORES
A 40707 (EP7)
41010 (EP10)
41110 (RM4)
B 41313 (EP13)
C 41510 (RM5)
D 41717 (EP17)
E 41812 (RM6)
F 42316 (RM8)
G 42120 (EP20)
H 42819 (RM10)
J N43723 (RM12)
mag-inc.com
4.15
Core Selection
dc_bias_pq.eps
Selector Charts
PQ CORES
dc_bias_lam_size_e.eps
A 42016
42020
B 42614
C 42610
42620
42625
43214
D 43220
43230
E 43535
44040
dc_bias_e.eps
A 41203 (EE)
B 41707 (EE)
C 41808 (EE)
D 42510 (EE)
E 43009 (EE)
43515 (EE)
F 44317 (EE)
G 44721 (EE)
H 45724 (EE)
E CORES
A 41205 (EE)
B 42515 (EE)
C 41810 (EE)
43007 (EE)
D 42530 (EE)
43520 (EE)
E 42520 (EE)
4.16
MAGNETICS
dc_bias_e_ei.eps
E, EI CORES
dc_bias_e_ei2.eps
E, EI CORES
dc_bias_ec.eps
Core Selection
Selector Charts
A 44016 (EE)
B 44011 (EE)
C 44020 (EE)
D 44308 (EE, EI)
E 44022 (EE)
46016 (EE)
F 45528 (EE)
45530 (EE)
47228 (EE)
48020 (EE)
G 46410 (EE)
H 49938 (EE, EI)
EC CORES
A 43517
B 44119
C 45224
D 47035
mag-inc.com
4.17
Core Selection
dc_bias_etd_eer.eps
Selector Charts
ETD AND EER CORES
A 43434 (ETD34)
B 43521 (EER35L)
C 43939 (ETD39)
D 44216 (EER42)
44444 (ETD44)
E 44949 (ETD49)
F 45959 (ETD59)
dc_bias_eem_efd_er.eps
4.18
MAGNETICS
Gapped Applications
DC Bias Data
Gap_dc_bias.eps
Effective Permeability
NI = 0.80 x H x le
Where
NI = maximum allowable ampere-turns
H = DC Bias level
le = core path length (cm)
AL le
4 A e
1 1 lg
= +
e i le
A e = effective cross sectional area (cm 2 )
A L = inductance/1,000 turns (mH)
i = initial permeability
l g = gap length (cm)
e =
4.19
mag-inc.com