Historical Perspective

Early Days

In the early days of Power Electronics, issues in design theory were usually decided by the more experienced members of the industry. A good example of this phenomenon is found on Jerry Foutz’ Website in the “Personal Anecdote” section.

In reading it, you will notice that considerable opposition to the proposed explanation and solution arose from power supply manufacturers’ representatives! This is understandable, as they were salesmen, not engineers. As such, they were principally concerned with protecting the reputation of their existing product lines. But design engineers are concerned not with saving face, but with meeting spec, and thereby serving the general public!

At any rate, in our example, the necessary answers reached the design community, allowing engineers to take this new factor into account in their ongoing work. Likewise, field engineers could retrofit product already in the field before hazardous failures resulted!


A Problem Arises

In 1980, things changed as a deadlocked debate [between Caltech’s Professor Slobodan Cuk and MIT’s E.E. “Doc” Landsman] over optimality threw this weighty question out to the industry itself. [more] Where the academics had failed to decide, the engineer was left to decide for himself! Faith in so-called experts waned, and designers turned to their own experience, and that of other working engineers for guidance.


The Commercial Sector Responds

In the ensuing vacuum, the commercial sector was quick to step in with plenty of free design advice! Chip manufacturers (PWM control, FET driver, power management, and even microprocessor packages) flooded the market with app. notes touting their own products. Software writers (SPICE, PSPICE, MATHCAD, SCAP, and a host of others) all claimed to provide the answers to difficult design problems via computer simulation. None broached the subject of optimality, confining themselves to delineating the various benefits of their own wares.


The Situation to Date

As an industry, Power Electronics has, so far, generally chosen to assume that no circuit can inherently perform any better than any other. Rather, the general belief is that performance depends principally upon the individual engineer’s wit and design skill, and on his case by case choice of circuit topology. I.e. it is how he fills in the values of his circuit topology that matters, and how he modifies it to fit a given application; and further that the correct choice of topology depends strongly on the given application.

The industry’s position is, perhaps, natural, given the nature of circuit design in electrical (and especially electronics) engineering, where circuit topology and component selection are the name of the game!

But the popularity of this opinion does not necessarily make it true!


The Rub

One must remember that the basic task of power conversion remains the same in allapplications! A switching converter is inherently just a four terminal device! Multiple outputs, isolation and inversion are readily available regardless of the basic topology [cf. Rudy Severns’ handout of 100+ topological variations] while AC vs. DC operation depends on the manner of duty cycle variation, not on the topology itself! It is really only the “bells & whistles” that vary from app to app.

Such techniques as resonant and soft switching, and synchronous rectification have turned out to be generally available as well. The journals and conference proceedings are full of such solutions for virtually every topology! The result is that few issues remain, save direct topological comparison.

Even engineers involved in 3-phase power or cycloconversion will be familiar with the necessity of reducing those cases to the simple (1-phase) topology, and of returning the operation to DC via phasor notation, before stability analyses and performance comparisons can be made.


Light in the Darkness

In fact, it is just this truth that allowed the construction of a Canonical Modelin the first place. Its very existence relies on the common four terminal structure of all the basic topologies.

At the “bottom of the pileup” is Dr. Middlebrook’s Canonical Model! Its input and output filters reflect the basic necessity of filtering the switched square waveforms before they impact source and load, the “DC to daylight” transformer models the conversion property of the switching action, while the V and I hat generators contain the residual dynamics of the particular topology!

In other words, the answer was there all along! The existence of a universal canonical model demonstrates that any application is reducable to the simple four terminal case. At which point, one may safely return to the origianl switched topology, and compare it on fair terms to one used in any other application.

In the End
In the end, the intervening quarter-century has done little or nothing to alter the original claim of “optimum topology” laid by Drs. Middlebrook and Cuk! Why not design with the best, then? Just $5.00 gives you the design equations you need to produce a high performance boostbuck (Cuk) converter!



The point is that many topologies can be more or less made to work for a given application! What is weird but true is that the boostbuck (Cuk) converter works better than all the rest!

Though this is surprising, and even counterintuitive, it may actually be a fairly common phenomenon in inventing/engineering. One wonders whether other examples of such “optimum topologies” exist in the engineering world?!