Cuk Power Supplies

The Cuk converter made a splash in the early 1980's, promising to eliminate ripple currents and simplify magnetics. In practice, most engineers couldn't or wouldn't analyze the closed loop response, got a supply that didn't work, and gave up in disgust.

With input and output inductors coupled, you can't effectively close the loop. If you have a stable supply and load, and just want a quiet, reasonably efficient, energy conversion between voltages, then you can do that. But you won't get a dynamic response you can live with.

The inventor is still around, and can be found at Teslaco and excellent tutorial material is available at www.boostbuck.com. You can find some excellent resources at both sites, but in particular, you should buy the book "Advances in Switched-Mode Power Conversion" (Vol I&II, bound in a single volume) if you need to analyze this supply.

If you are the trusting sort, get the design guide from boostbuck.com, follow it, and you'll be okay.

Here are some hints:

1. Do not try to accommodate a wide input (on your first try).
2. Design for the lowest input voltage, highest duty cycle
3. Then, reduce loop gain for stability at minimum duty cycle
4. Do follow the design guidelines at www.boostbuck.com for sizing the components.
5. Do include the damping components from that site. If you think more than 10% of your total power will be dissipated in snubber, you probably didn't account for the effects of duty cycle and wave shape. Dissipation is much less than ppk voltage squared divided by impedance.
6. You can use current mode control, but start with voltage mode control loop. During fast load transients, current mode control will be lost, and you still want a stable supply. Switching to current mode control will give you an additional 90 degrees of phase margin, but only when the supply is in regulation (hint: it won't be during load transients).
7. Following the instructions at www.boostbuck.com, you should concentrate on the "Isolated" topology for your finished design, but
8. Do prototype your design in non-isolated version unless your isolation transformer will be other than 1:1. (Hint: duty cycles as low as 8% are acceptable. Unless you need to do so for safety reasons to reduce component stresses, the 1:1 isolation will probably work well.)
9. Do not attempt a duty cycle greater than 50%, they dynamic response will be very poor.
10. Do use silicon carbide (SiC) rectifiers at the output if you are working with line-level supply voltages.

The best applications for the cuk supply are those in which the primary is at a fixed voltage (like a storage battery or intermediat bus). It might be possible to use a cuk converter as a power factor correction circuit since the loop response must be quite slow anyway.

It's not hard to adapt to step changes in load current, but line voltage changes are another matter. I've done some simulations to see if a wide input range converter is possible, and it may be practical at power levels below 50W or so; but one needs bidirectional switching for it to work well (you need a rectifier AND an active switch in the secondary circuit).

If you are on a budget and like to learn things yourself, the above resources should be sufficient to get you going. If you need to improve your changes of success, probably it would be wise to check out the courses offered at Teslaco. The basic Cuk topology patents have expired, but some of the tricks needed to get the supply working may still require licensed technology.

In my opinion, the inventor's attempts to capitalize on this technology is one of the great failures of the patent system. Teslaco is moderately successful, but has only captured a very tiny fraction of the potential market for this technology. The fact that their book is only available from their own site, and looks like it was printed on a Daisy-Wheel printer, bear silent testimony to the fact that these guys have had to go it alone. And in the hyper-competitive world of power supplies, nobody even gave the Cuk supply a second look until the primary patents expired, delaying industry-wide research into the topology by 20 years.

Responsibility for the slow adoption of this technology, in this author's opinion, rests at least partially with the inventors. Because the technology has remained largely proprietary, it hasn't benefited from the dynamic gain that open competition has produced in other technology areas.

I think this is a great technology, and should be used in power factor correction and distributed bus architectures. It should be the predominant technology used for -48 V IBA (intermediate bus architecture) supplies. But it's not. If you can afford to pay them, and if they have time for you, Teslaco can probably design a supply for your application. It won't likely be royalty-free.

I'm very impressed by the technological genius and solid experimental practice of these guys, don't get me wrong. And I'm not sure I have a right to judge their motives for keeping their technology "Close to the vest." But I can say that this topology hasn't found wide acceptance in the market, and if that is due to a lack of technical information, at least some of the blame falls on Teslaco for trying to capitalize too quickly on their invention. Clearly they succeeded on a personal level, but at what cost to the industry? Adopting a consortium model probably would have made them ten times the money, and would have involved IC manufacturers in the process much earlier. To say nothing of the quality of information that would be available as a result. But I guess I'm just crying over spilled milk.