Applications engineers are in a unique position with respect to dc-dc converter architectures. In the lab, in the field, or on the telephone, they hear a lot. In fact, the centralized control architecture (CCA), which coexists with and adds capability to other power architectures, and the factorized power architecture (FPA) have been getting a lot of press recently.
The CCA controls the power elements through a digital communication bus, providing information and enabling the programming of power components. For example, you could program a power component for output voltage, minimizing the number of parts you need to buy.
Or, you could program for monitoring status to detect and report problems. Or you could program capabilities such as current limit, or built-in functionality that is controlled externally, depending on the application. Digital communication provides greater flexibility for power devices and for power designers.
The FPA, unlike the intermediate bus architecture (IBA), places the isolated element at the point of load. Regulation is performed upstream, and the voltage transformation module is right at the point of load. FPA also provides current multiplication and the option to eliminate an external control loop that typically impedes the ability to have a fast, dynamic response from the converter, leading to much faster response times.
Efficiency is high, and the dynamic performance is extraordinary. Factorized power minimizes distribution losses by enabling the bussing of higher voltages than with an intermediate bus while achieving high performance with low-voltage (less than 5 V) load demands.
With an isolated voltage on the entry point of a printed-circuit board (PCB) and non-isolated voltages at the point of load, IBA made power systems smaller, more efficient, and more reliable, while reducing cost.
Conversations with customers reveal a lot of uncertainty about the architectures and understanding what makes sense for their applications. However, few people fully understand the intermediate bus and factorized power. Of course, the CCA is new, and people are just getting up to speed on it.
In The Field
Many questions or comments focus on new products coming on the market that incorporate a lot of new technology and communication buses. Power-One’s own Z-Bus architecture plays right into CCA. National Semiconductor and Zilker Labs have taken CCA communication to the next level, calling it PMBus. Ericsson has embraced PMBus in its dc-dc converters now, so it also has this I2C communication capability with its converters.
The trend toward server farms, like Lucent’s huge facility in Plano, Texas, with millions of square feet under one roof, demands the capabilities and complexities of the new power architectures and technologies. It’s a lot of data, a lot of power supplies, great amounts of energy being used, and much heat being generated.
Efficiency gains that rely on architectures like IBA and FPA reduce the heat that’s generated. Especially with such large systems with many power components, cooling becomes a major challenge. Any way to reduce the amount of heat that is dissipated is beneficial in the overall design and cost of the system to operate. That’s why such large systems ultimately focus on the power design, because that’s where much of the heat is generated.
Complexity is just one side of the story, though. There are still many simple power applications. They could be high volume, but they just need a couple of voltages, or a single voltage. Maybe they plug into the wall. Maybe they’re a portable battery pack. But they’re not complex.
How do power architectures play into these kinds of applications? If they need a single output voltage, centralized power architecture still makes sense. The distances aren’t that great. Just put it in there and bus it—simple, effective, and low in cost.
Centralized power has been around from the very beginning, and it’s still being used. It’s a very simple concept. The power supply is in one central location with wires that go to where they have to go. Distributed power basically separates the power supply into two or more entities and places them where they need to be. That process was coincident with the advent of dc-dc converter bricks.
Many designers value ease of use, ease of implementation, or low cost of ownership. There’s something to be said for keeping things simple—having a single power supply, already designed, that’s easily dropped in. Maybe it needs a fan. Maybe it mounts to a heatsink and the thermals are already worked out. You just connect the wires. That plays into time-to-market too. When you’re wrestling with a lot of technical issues, the last thing you want to add is power-supply complexity.
So people selling products to address power architectures also need to think about how they’re implemented and ease of use. That’s a very critical consideration for deciding whether a certain architecture makes sense. For a lot of companies, IBA makes sense, because intermediate bus converters and non-isolated point-of-load converters are relatively simple to put together. Factorized power, with its modular elements—BCM, PRM, VTM—is straightforward and simple to use, with few external components that have to be added on by the user.
That again, of course, depends on the demands of the application. If the application requires remote sense, tight regulation, margining, setting the output voltage, or current-source type applications, some type of external circuitry will be required to implement that functionality. But it all comes down to the application. That ultimately makes the decision for you. Yet it’s nice to know that there are a lot of choices, and we can explore the possibilities. There is certainly much more to be said about DC-DC converter architectures, but that will have to wait for another day.