In 2005, I presented a conference paper in which I argued that the power-supply industry would only switch from analog PWM controllers to fully digital controllers if a killer application were found. Here it is two years later and the killer app has yet to be identified — but why?
The sought-after killer app would be one where a digital power converter can offer needed performance benefits and flexibility, and where no analog solution can be easily and elegantly achieved. This combination of factors, together with the right price, would lure the industry’s executives and procurement managers to accept digital power.
While there have been some impressive strides toward these goals since 2005, there has been no mass adoption of digital power technology in any application that would signal a turning point for the power-supply industry. So what has been holding digital back? Is the killer application idea still valid?
The question of whether digital power needs a killer app to succeed is somewhat controversial. Shortly after I published my conference paper, I remember reading an article that questioned the relevance of a killer application. The author argued that no single application is large enough to cause an industry-wide migration from analog to digital control.
But I maintain that the dc-dc converter used to power the CPU in personal computers (PCs) would be an ideal killer application for digital. There are about 200 million such power converters manufactured every year, and these converters must meet some of the most demanding specifications in the industry.
Most of these power converters now employ analog control. But if the digital approach were to achieve great success in this area, it would almost guarantee that the technology could be easily adapted to all other dc-dc applications. This is true because, to get universal approval in the PC market, a PWM controller must meet all the CPU’s demanding performance specifications at a price that is at least on par with the ubiquitous analog solution.
If that happens, the technology will have matured to a point where the price/performance value proposition is such that digital technology could easily be adopted in other applications. Having conquered PCs, digital power could then be deployed in the telecom industry’s point-of-load (POL) converters and 48-V isolated dc-dc converters, as well as countless ac-dc converters in all shapes and forms.
However, a paradigm shift in the business model of the PWM controller manufacturers must be achieved to pave the way toward such a migration. That is to say, the chipmakers must provide a digital solution for a given application that enables a totally new power system or subsystem. Moreover, the digital solution must greatly expand the domain of the application, providing breakthrough features that are uncommon or even unknown in analog power systems.
These features range from the simple to the complex. Digital technology could perform simple tasks such as power sequencing and communicating power-supply status to a central computer. The technology could also implement more demanding functions like fault diagnosis and even calculating the optimum control-loop parameters on the fly.
The latter capability could be applied in the production environment to a mass-produced power converter to eliminate the unpredictable effects of the converter’s various error-prone components such as inductors, capacitors and error amplifiers. In other words, digital techniques could compensate for the various component tolerances, which produce a statistical distribution of component values when a power converter design goes into high-volume production.
There’s no limit to the number of ways these digital-enabled features may be combined to create power systems that can accommodate the demanding power requirements of the today’s most sophisticated systems as well as those of the future. This clearly brings us back to the concept of the killer application.
Excluding the supremely sophisticated modern military systems, there are still no large applications in the tens of millions of units per year that demand such advanced power systems to achieve success. The reason is partly because power systems have been the exclusive domain of the analog discipline because analog is the most cost-effective way to implement these systems. Design organizations still choose a solution exclusively based on its cost, assuming that it meets the basic design specifications.
Ultimately, digital power solutions will take off if — and only if — they can offer the same or better performance than analog for lower price or if the system is so complicated that the digital approach is the only feasible solution to meet the target price and performance demands of the end users like the giant OEMs.
1. Elbanhawy, Alan, "Going Against the Grain: Digital Power Looking For a Killer Application," 2005 Digital Power Forum, Boston.