Design Hangups |
|
Design Hangups |
 |
 |
|
 Integrated Magnetics and the Coupled Inductor
As alluded to above, one of the principal roadblocks to designing the Cuk Converter was the the difficulty engineers had producing an acceptable magnetic piece. This problem was compounded by the general belief that it was necessary to employ advanced magnetics to out-perform some of the other topologies. As has been shown in the first section of this web page, this notion is FALSE. All other topologies suffer from various shortcomings that render them unsuitable for switching power supply design! In fact, only the Boost-Buck Cascade, and the uncoupled Cuk Converter are required to outperform all other topologies. Another hangup was the complexity of the model of the coupled inductor & integrated magnetic piece. Simply stated, the situation is this: When the zero ripple condition is met, the T model of the coupled inductor simply turns into a magnetizing inductor in parallel with an ideal 1:N transformer, with a series non-zero leakage inductor on one side only. 1:N is just the turns ratio of the converter's isolation transformer, if any. This zero inductance path then draws away the ripple current, as mentioned earlier in connection with the Landsman converter. The integrated magnetics piece is also simply modeled as two coupled inductors in parallel, with the isolation transformer added between them, finite leakages blocking ripple at both input and output.   DC Conditions and
Switch Stress Levels
A great deal of confusion results from an attempt to nail down the DC conditions in the L's & C's, and the stress encountered by switch elements. The right way to understand reactance voltages and currents is to work from the input voltage Vg to find a cap voltage at a given D, and to work from spec output current I to find an inductor current at a given D. This avoids complicated expressions involving R or D & D' squared, which are difficult to interpret. Then, all component stresses are easily found from: Ig/I = qD/D' and V/Vg = pD/D' [*]. Here pq is the efficiency Pout/Pin. It is ascribed to voltage drops and current "leakage," as this is usually the case in a switcher. Typical values might be q =.95; p =.85; for an overall efficiency of 81% not counting the housekeping power requirement. From this standpoint, the stresses in the switch elements are easily found. For the Cascade, the Boost Q & P each see Ig x Vg/D', while the Buck Q & P each see I x Vg/D'. For the topologically reduced Cuk Converter, both Q & P see Vg/D' x I/D'**. Note that the sum of the two stresses is equal to twice (I+Ig) x (V+Vg) as it is in all Boostbuck topologies, and, in fact, in all isolated converters. Remember that, as was mentioned earlier, the stress on the elements of the Buck Forward are often believed to be less than that of the Boostbuck topologies. This is not true, and only seems to be true since, for a given R, the Forward puts out less power for a given D! ** This expression may be regarded as the "design oriented" form of the usual DC expressions [*]. It allows the switch stress to be determined for the three variable quantities: line voltage and load current, and the duty cycle calculated to provide output voltage regulation at that point. | Design Hangups | More Hangups | | The Four Topologies | Other Topologies | Philosophy of Design | Do the Math | | Return Home | Old Home | |
||
 |
 |