Abstract – New planar construction alternatives for blending
the various power magnetic components of switch mode power
processing circuits and systems are presented. These unique
approaches are based on the use of core designs with multiple
winding areas and core sections which provide common flux
paths for transformer and inductor operations, arranged so as
to minimize magnetic interactions and core material required.
Construction examples for 125-watt & 50-watt DC-to-DC
switch mode converters are detailed, the latter example
consisting of a four-piece three-chamber cylindrical planar core
structure that houses all power magnetic functions, including
input filter inductance. Planar means to control leakage
inductances between inductive windings are also presented for
control and reduction of AC ripple current magnitudes.
I. INTRODUCTION
ONE of the more interesting magnetic design techniques
in practice today by power supply engineers is the
“blending” or “mixing” of transformer and inductive
functions of dynamic power conversion circuits on a single
magnetic core structure. This technique has come to be
known as Integrated Magnetics (IM) design. Although the
origin of the technique can be traced to prior work performed
in the early part of the last century, IM methods have ecome
increasingly popular since 1977, when “coupled-inductor”
assembly concepts and ssociated converter designs [1][2][3]
were more fully disclosed. When implemented correctly, IM
techniques can produce a significant reduction of the
magnetic core content of related power processing systems,
resulting in cost-effective designs of smaller weight and
volume. Recently, IM packaging alternatives have been
extended to include implementations in planar, or “flat”,
forms, using modern printed-circuit approaches for windings
together with low-profile ferrite core constructions.
It was subsequently demonstrated in 1984 [4][5] for circuit
arrangements and in 1987 [6] on a system level that all
power conversion circuit designs of the switch mode variety
have one or more IM forms. Prior to these points in time, it
was commonly believed that only those power-processing
networks wherein transformer and inductor winding
potentials are always dynamically proportional could have
IM versions. Such special converter networks include
transformer-isolated versions of the .uk [2], SEPIC and
ZETA topologies. However, by the use of core structures that
possess more than one major material flux path [4], IM
techniques can be applied to all switch mode power
processing circuits and systems, and can be extended to
include other magnetic elements often excluded. Examples of
such elements are second stage input and output filter
inductances used for reducing conducted AC current levels.
In this paper, two new approaches [8][9] for using IM
techniques in conjunction with planar core and winding
construction methods are described. These approaches
require the use of core structures with more than one winding
window area, and have the distinct advantages of minimal
interactions between the various magnetic elements included
in the IM assemblies. The full benefits of reduced core
system volume and weight by the IM process are also
achieved by these unusual approaches. The resulting core
structures can completely enclose all windings (except for
access slots), providing excellent magnetic shielding
capability. In one of these approaches, “off-the-shelf “planar
core sections can be used to construct the IM system.
II. OVERVIEW OF CONVENTIONAL IM METHODS
IM designs today typically use soft-ferrite E-I or E-E core
structures. Fig. 1 is an collection of some of these
“traditional” IM designs [1][4][6] for a “single-ended”
converter circuit, which differ only in winding locations on
the three “legs” of the core structure. For the core leg where
inductor windings are situated, an air gap is added to obtain
the desired inductance values. Here, the converter topology
shown in Fig. 1(a) is a buck-derived “forward” configuration,
found in many DC power conversion systems today, where
output power needs typically range from 50 to 250 watts.
Effective core leg areas must be chosen in accord with the
maximum flux levels that will occur as a result of converter
operation so as to prevent saturation of any leg under
maximum loading conditions of the system. As can be seen
in the alternative designs in Fig. 1, each leg will see different
flux levels, depending on winding locations and phase
relationships, along with core leg reluctance values. For
example, in the “split-winding” version shown in Fig. 1(c),
the outer leg to the left of the center leg will see the sum of
the transformer produced by the inductor winding. In the right outer leg, the
flux here will be the difference between the transformer AC
flux and the remaining half of the AC and DC components of
the inductive flux generated by the winding(s) mounted on
the center leg of the core structure. In this particular IM
variation, the window area needs on each side of the center
leg are equal. However, in the variation shown in Fig. 1(d),
an optimum core design would require unequal window
areas on each side of the center leg. This disparity is also a
possibility for the IM designs in Figs. 1(b) and (e). In the
variation of Fig. 1(e), the inductive leg is one of the outer
legs of the core structure, and its area would need to be larger
than the others, which is not a standard E-E or E-I core
configuration. Finally, it is apparent that all of the alternative
designs shown in Fig. 1 require winding bobbins that must be
fitted on one or both of the outside core legs. Presently, such
bobbins are non-standard items, requiring custom
manufacturing and added assembly cost.
For “push-pull” versions of converter circuits where twoquadrant
B-H loop operations occur from transformer
actions, the centermost leg is used for the inductive part of
the converter, with dual primary and secondary windings
placed on the outer core legs. In these situations, the window
area requirement is balanced, like the core-and-winding
arrangement shown in Fig. 1(c).
A variety of methods for core reset [1] due to the “singleended”
transformer action of the converter in Fig. 1(a) is
available. Some of these methods utilize resonant reset
techniques involving the parasitic capacitances of T1, Q1, D1
and D2 and the self-inductances of the transformer windings.
III. PLANAR IM CONCEPTS
The use of printed wiring methods for the windings of an
IM can lower the height profile of the overall package of the
magnetic component, and since the windings are mounted on
a laminate base, the cost of special bobbins mentioned earlier
is eliminated. One example of a Planar Integrated Magnetic
(PIM) design for a “split-winding” version of the forward
converter topology is shown below in Fig. 2.
In the PIM design of Fig. 2, the primary and secondary
windings are interleaved using selected layers of the PCB on
the outside portions, while the inductor winding is split into
parallel sections of the inside layers. Feed-thru vias in the
PCB are used to interconnect the various parts of all three
windings. Terminations of the winding ends can be
accomplished in a number of ways, such as the use of solid
pins running vertically from the PCB end patterns down to
the PCB of the converter assembly. The use of “gull-wing” leads is another alternative, and this termination method is
particularly useful when placing the PIM on a “motherboard”
assembly with surface-mounted components.
“Conventional” IM constructions, whether they use PCBstyle
or wire windings, do have some undesirable limitations.
Because windings are required on the outer legs of the core
system, it is not possible to completely surround all windings
with core material to restrict magnetic leakage levels. This
situation is easily seen from the PIM construction example
shown in Fig. 2. Also, because conventional constructions
using E-I or E-E cores restrict window locations for windings
to two locales and material flux paths to a maximum of three,
other power magnetic components in a conversion system
(like input and added second-stage output filter inductances)
cannot be easily accommodated in an IM arrangement
without significant topology changes. This, in turn, often
leads to undesirable compromises in power processing
performance.
Rather than placing the various windings of a planar IM
(PIM) in a conventional “open, side-by-side” manner, they
can also be arranged in a “closed, top-to-bottom” core
configuration [7][8], as shown in the cross-sectional diagram
in Fig. 3(b). In this example, the upper chamber of the core
serves as the location for the output inductive windings of
the converter, while the lower chamber is reserved for the
windings of the transformer. The core piece separating the
two chambers provides a material path for both inductive and
transformer flux elements. With the windings phased as
shown in Fig. 3(b), the effective flux level in this common
path will be the difference between these flux values.
Therefore, the core material needed for this common path
can be reduced accordingly. Although the center post areas
are shown in Fig. 3(b) to be identical, this condition is not an
absolute design requirement. In fact, optimum size and unit
height studies for the PIM form shown in Fig. 3(b) for a
particular converter application may indicate otherwise.
A practical construction of the PIM concept of Fig. 3(b) is
shown above in Fig. 4. Here, a “pot” format is used,
consisting of three separate core pieces, stacked to form the
two window areas, or winding “chambers”, needed for the
transformer and inductor parts of the IM. Within the lower
chamber are then placed the PCB assembly for the primary
and secondary windings, with the upper chamber reserved
for the PCB assembly for the inductor winding. Note that it
is possible to use double-sided PCBs in place of a single
multi-layer PCB in each locale, and interconnect them by
external wiring methods using termination “tabs” on the
PCBs. These tab areas are accessed by means of “slots” cut
into the core sections, as can be seen in the example
assembly in Fig. 4.
Fig. 4. A “closed, top-to-bottom” PIM assembly, using three separate pieces
of magnetic material. Only the center post (CP) of the core top is gapped to
provide the desired filter inductance value.
In this construction example, a small hole through the
center posts of the cores has been provided to permit a single
non-conductive screw-and-nut combination to be used to
hold the assembly together.
Adding one or more inductive windings to a “stacked”
PIM construction to accommodate additional filter
inductances of a converter system is simply a matter of
adding a third chamber area [7][8] to the core arrangement.
This chamber can be located below the chamber in Fig. 3(b)
where the transformer windings are housed. Fig. 5 is an
illustration of this construction, wherein the lower chamber
contains the winding(s) for the inductance of an input filter
network for the forward converter topology shown in Fig.
1(a). Now the core piece below the transformer chamber serves as a differential flux path for the transformer AC flux
and the DC+AC flux generated by the lower winding
associated with the input filter inductance.
|
