On the Performance range, the spectrum will be subdivided in various flavors of digital, from low cost simple and robust state machine with on chip NVM, to sophisticated DSP based ones and analog (Figure 2).

Figure 1: Cost-Value-Performance: Dollars vs. volume

Figure 2: Digital flavors and analog continuum

he real customer is no longer only the design engineer, but all others who would benefit from this enabling technology: purchasing, test engineers, QA, manufacturing, warranty/repair (Figure 3). So the value proposition is expanding from strictly satisfying a technical/parts cost requirement to enabling energy savings, reducing life cycle cost (ownership + disposal) and you name it. "You name it" is the job of non technical folks who up to no long ago had no saying in the creation of a product.

And why not invite the marketing guys at the table and become part of your customer's value proposition. Power management functions at system level can be a significant differentiator. Make known the possibilities [17], like energy saving modes, reporting, remote monitoring, and testing [26].

Figure 3: Value Chain

The proliferation of flavors (part numbers) of analog parts will no longer be supported by high volume demand and mass production, and the digital parts, allowing "in-situ" customization, will be favored. Part number maintenance, availability, reliability issues will keep the cost of analog flat, while the cost of digital will benefit from Moore's law.

Digital power and digital processing power requirements [Figure 4, Table 1] will advance in synch. And "The Paradigm Shift to Digital Control" [1] where "cost is key, in power-supply business" paradigm survives, will be replaced by another paradigm where "value-is-king". Back to quality is a necessity that will turn the tide.

Figure 4: Power Density Trend

Table.1 (Source ITRS) Who are the customers?
Who are the digital power management customers?
Traditionally the design engineers were the "pick-and-chooser" of a project's parts. Their office spaces were dominated by bookshelves cramped with data books and catalogs. Intel, Linear, Maxim, Motorola, National, TI were a perennial sight. Like Methuselah, in the Old Testament's Genesis, the parts used to have a long long life. Markets were to a large extent self-contained and competitive pressures were mainly from the local or neighbor markets.

The ever accelerated speed of technology and the globalization of economical system have shifted the focus from the intrinsic/performance value of a design to the lowest cost (parts, manufacturing, and distribution). In the quest for short term profits, manufacturing and design went offshore. With this reshuffling the customer is no longer the design engineer but a mix of people which may not even include the design engineer. So who are the customers and what are their concerns?

Design engineers of various up bringing, analog, digital, power conversion, etc. are still the main customer, though their ranks are shrinking. Based on the market they are serving, high end (high-end servers, datacomm/networking/storage, ATE, military), mid range ( telecom, midrange servers, industrial, medical, high class consumer) or low-end (servers, low-end industrial, low-end consumer) their main driver shifts from performance, at the high-end, to cost at the low-end.

The cost structure itself is different from the high-end to the low-end. The high-end cost is theoretically the life cycle cost, and includes the cost of ownership and the cost of disposal. At the other extreme, the cost covers basically parts and labor, with no concern for cost of ownership, warranty, service or disposal. So the main concerns order is different depending on the served market, but includes: performance, space, cost, flexibility, risk, design time, availability/second source, agencies compliance, support.

Project leader comes into picture in the high-end and mid-range applications and is the person that monitors the cost at the system level, so what can be attractive to the design engineer may not be so to the project leader. Overall performance, system cost, risk, time-to-market, customer-supplier relationship are the main concerns.

Purchasing group/person, could be the gating factor in having the product offer accepted. Smaller businesses are easier to deal with while bigger ones will have to qualify you to get on their approved or preferred vendor list. So the main concerns may be the company's financial health and business volume, second source availability, price, global distribution and support.

QA issues may be handled by QA personnel or be placed on the design engineer's agenda. Either way, agencies approvals, reliability data and possible site inspection would be the concerns.

Test engineers may come into picture in larger businesses or the issues may be addressed by the design engineers in the smaller ones. The concern here is how much testing is necessary and how easy it can be done (testability)

Though it is very unlikely to meet with manufacturing/CM/OEM representatives it is important to understand their concerns. What are their manufacturing capabilities, how is the part assembled, what could cause a low yield or affect the part reliability.

Customer service is also unlikely to be represented in the acceptance process but knowing and addressing their concerns adds value the offering.

Let's take a look now at how can we add value by addressing our customers' concerns. Options and value added
What are options and the value added for each customer?

Design engineer: (performance, space, cost, flexibility, risk, design time, availability/second source, agencies compliance, support). While most of these are interrelated, some deserve a dedicated treatment.[18] Performance means meeting the specification for all operational and environmental conditions.

Static performance: Here component tolerances, aging effects, temperature compensation are mute issues for the digital solutions being solved internally by A/D calibration upon startup and lookup tables for compensation. Higher performance is achieved at no extra cost or components.

Dynamic performance: The ability of selecting the response based on the magnitude of the disturbance signal places the digital solutions ahead from the start. digital solution benefiting from memory and the ability to communicate can respond intelligently to events in the time continuum (adapt, anticipate, react, and respond to external commands) while the analog solutions can only react.

Proactive vs. reactive: The design benefits both in reduced external complexity (fewer components), and reduced time.

Space (integrate/disintegrate) brings two inter related physics issues: efficiency vs. cooling. It also forces the choice between accepting a single source (integrated solutions) and a higher cost vs. multiple sources, flexibility, and lower cost (discrete solutions).

Reduced space is accomplished by increasing the switching frequency which leads to smaller size L and C. In the 700kHz to 1 MHz range and above the discrete solutions are no longer viable due to the parasitic inductance in the power stage. But while increasing the switching frequency and reducing the space, extra switching losses (heat) are added and less cooling area is available for heat dissipation. A heat sink and adequate cooling air flow are needed. Related to this issue is the mismatch between the allowable operating temperature of the controller (lower) and of the power stage (higher) for equal lifetime. So the extra cost of an integrated solution has this less visible aspect of wasting silicon area by forcing the power stage to operate at the same temperature as the controller, or affect reliability (shorter life) by forcing the controller to operate at a higher temperature (of the power stage).

In certain designs special cooling is not available and there is no choice but to increase the efficiency by lowering the switching frequency, which precludes the use of integrated solutions. Yet, in other designs, high switching frequencies are not tolerated due to the noise sensitivity of the design.

From the above discussion it transpires that the integrated solutions have a smaller market in the higher power range (tens of amps) and a larger market in the low power (a few amps). Regardless of the market, the single source status of most integrated solutions brings a premium cost and the availability issue.

Cost is a catch all term with different meaning/content to different people. From the high-end to the low-end consumer market cost has a different content from "all included" (warranty, maintenance/service, disposal), to bare bones (parts and labor).

Regardless of the application, lower parts content, looser tolerances, more generic parts/higher volume (fewer part numbers), lower time-to-market, flexible /scalable/reusable design, higher reliability, all translate in lower cost.

Flexibility has a few dimensions. It stems from having memory space available, from having communication in the design stage and in operation, from technology (growth Moore's law), and from architecture (modular design/scalability).

Project leader (overall performance, system cost, risk, time-to-market, customer-supplier relationship). The project leader concerns are a subset from the ones listed above, but with strong emphasis on just a few, different from case to case. Purchasing (company's financial health and business volume, second source availability, price, global distribution and support). Apart from the cost, second sourcing, global distribution, purchasing people have an ongoing incentive to reduce the number of parts in the bill of materials (BOM). Here the digital benefit is two fold; the part itself is just one but it can "morph" into different applications according with how it is programmed (ward wired-resistors, or software-nonvolatile memory) then the parts savings due to extra functionality and internal loop compensation (no external R,C).

QA (agencies approvals, reliability data and possible site inspection). Proof of ISO certification, reliability data on silicon and the packaging, memory retention at elevated temperature is information more readily available on standard digital processes than for analog ones.

Test engineers (how much testing is necessary and how easy it can be done - i.e., testability).

The I²C communication allows margining and telemetry feedback. Because communication with the converter is possible and because implementing new features is a lot easier, cheaper and faster that for an analog counterpart, test engineers have a chance to ask for and expect to receive these features. "The IEEE 1149.1 boundary-scan standard [JTAG = Joint Test Action Group] has been adopted industry-wide." [26] Is JTAG possible and at what cost?

Manufacturing/CM/OEM (process variations, parts tolerance). Fewer parts in the system and fewer controller part numbers by themselves reduce manufacturing cost and increase the yield (fewer parts to assemble, fewer parts that could fail). The possibility of working with wider tolerance parts (calibration on start up) and less sensitivity to process variations (fewer external parts) are extra benefits.

Customer service (Telemetry and remotely identifiable data for maintenance and repairs/warranty). Light can be shed on the so far dark area of how close is the unit from the end-of-life and how has the unit failed. Performance degradation monitoring allows timely replacement and known failure mode can reduces warranty costs (repair, replacement, litigation) and offer precious feedback for product improvement. Discussion of goals
A discussion of goals

"Digital power is mostly analog"
"Power is digital"
"Digital is just a fad"
"Who needs it?"
"It's too expensive"
"It's too slow"
"It's too complex"
"It will put me out of work"

The nay-sayers, with the majority being analog incumbents [10] protect their turf against the digital new-comers [rest], while behind the stage preparing the ground for their digital counter-offensive. The newcomers started the offensive and the incumbents are preparing their digital counteroffensive.

The controversies remind us of either the blind men's description of an elephant, each of them touching a different part of it and therefore "seeing" a different reality, or the description of an airplane by different designers, each hyperbolizing their contribution. Job security concerns (for design engineers) [13] are also mixed in the perceptions. As professionals it is our duty to understand customer's wish list and needs and offer the best suited options. This is the only win-win proposition, where the customer receives the best value and the supplier is constantly monitoring and adjusting its value proposition.

The digital solutions' "sophistication will exact a price in application-development time" [19]. This is not necessarily correct, as development time is practically eliminated by the graphic user interface GUI. Built from scratch, on the spot rather than warmed-up leftovers is what digital approach offers and "let us do the cooking" with the majority being analog incumbents [10] protect theirs.

References and recent articles on the subject

[1] Don Alfano, June 13, 2006, "The Paradigm Shift to Digital Control", EDN,

[2] Margery Conner, May 25, 2006, "Digital Power Lures System Architects, Power Supply Vendors", EDN,

[3] Reno Rossetti, May 16, 2006, "Digital power-conversion schemes fit ultraportable applications", Power Management DesignLine,

[4] Reno Rossetti, Apr 4, 2006, "Digital Power Can Flourish In an Analog World", Power Electronics Technology,

[5] Deepak Savadatti, Feb. 2, 2006 "Driving Digital Power Management into the Mainstream", TechOnline,

[6] Reno Rossetti, Feb. 1, 2006, "Trends in Digital Power", ECN,

[7] Zaki Moussaoui and Greg Miller, Feb. 1, 2006 "Digital Power Control Highlights", ECN,

[8] Patrick Le Fàvre, Feb.1, 2006, "Intelligent Power Control is just Around the Corner for Distributed Power and POL Devices", ECN,

[9] Paul O'Shea, Jan. 23, 2006, "Designers debate merits of digital power management", Power Management DesignLine,

[10] Steve Pietkiewicz, Jan. 1, 2006, "Digital Power Is Mostly Analog", Power Electronics Technology,

[11] Data Points, "Power Controllers Promise Digital Benefits at Analog Prices", Power Electronics Technology,

[12] David Morrison, Jan. 1, 2006, "More Digital Design and Controversy in 2006", Power Electronics Technology,

[13] Jessica Davis, Dec. 27,2005, "Power: Efficiency and Digitization Key Words for 2005,2006", Electronic News

[14] Margery Conner, Dec 22, 2005, "Digital power controller is fast and software-configurable?", EDN

[15] Ashok Bindra, Nov. 1, 2005, "Digital Conversion Invigorates the Power Supply World", Power Electronics Technology,

[16] Margery Conner, Oct. 11, 2005, "Digital power IC combines power conversion and management", EDN,

[17] Joseph G Renauer, Michael G. Amaro, David Figoli, Sep. 12-15, 2005, "Consider the possibilities" Texas Instruments, Digital Power Forum

[18] David Caldwell, Aug. 18, 2005, "Power goes digital", EDN

[19] Joshua Israelsohn, July 21, 2005, "A bit-o'-power: digitally controlled power conversion", EDN,

[20] Vincent Biancomano, May 18, 2005, "Controllers power their way into the digital loop", EE Times

[21] David Morrison, Feb. 1, 2005, "Editor's Viewpoint: Digital Control Treads a Complex Path", Power Electronics Technology

[22] Don Tuite, Oct 18, 2004, "POL Control Options Match System Sophistication", Electronic Design

[23] Geof Potter, July 2004, "Digital Power, A Manufacturer's View", DCDC Technical White Paper, Emerson/ASTEC,

[24] Steve Goldman, May 28, 2004, "Power Is Digital", EE Times,

[25] Geof Potter, Apr., 2004, "An Introduction to Digital Control of Switching Power Converters", DCDC Technical White Paper, Emerson/ASTEC,

[26] JTAG/IEEE 1149.1

About the author:
Sabin Lupan is a senior Field Applications Engineer, at Primarion. You can reach him at sabin.lupan@primarion.com