What does the future look like for digital power? A call for questions

September 14, 2009

In the past, it’s seemed to me that panels on digital power are mostly a love-fest led by the power IC manufacturers that make, you guessed it, digital power controllers. I’m moderating a Roundtable discussion on digital power for the Digital Power Forum next Monday, September 20 in Orange County, CA, and I think the conference organizers have done a great job of assembling a broad cross-section of panelists.

The Panel’s title is, “Digital & Power: Oil & Water or Dollars & Cents?” The focus will be on how successful digital power has been so far, and what the future holds for the technology.

The key to this panel is the quality and diversity of its panelists: As far as I know, this is the first time digital power will be discussed by such a broad group, with panelists representing systems houses (Cisco, IBM, Juniper Networks) power supply companies (Ericsson, Power-One), and just one lone power IC controller manufacturer, Intersil/Zilker.

The Roundtable members will be:

Brian Zahnstecher, Power Engineer, Cisco

Patrick LeFevre, Director Marketing and Communications, Power Modules, Ericsson

David Owen, Power Systems Engineer, Juniper Networks

Randy Malik, Sr. Technical Staff Member, IBM

Chris Young, Senior Manager, Digital Power Technology, Intersil/Zilker Labs

Dennis Roark, Executive Vice President, Power-One

My personal bias is to focus on using a digital control bus for inter-system power control of POL converters. The Roundtable’s power supply representatives are from the two significant vendors in this technology: Power-One, which holds a patent on dc-dc POL control and communication through a digital bus (the Z-One bus), and Ericsson, which has been forging ahead recently with the largest product offering of digital supplies, incorporating both digital PWMs and digital housekeeping/communication, based on the industry standard PMBus.

However, in some email exchanges I’ve had with Steve Bakota, TI’s representative on the PMBus industry group, he reminded me that the panel also includes an expert in ac-dc power conversion (Randy Malik of IBM) as well as isolated power supply customers. His suggestions for questions and topics were much broader:

Why should power supply engineers – both designers and customers - care about digital?
In what ways would they need to ‘think differently’ during development?
What are the ‘gotcha’s’ to watch for?
When would they recommend going analog? (Great question – I gotta ask that one.)

And then I thought to ask Robert White, who has a long history as digital power expert, (especially in power housekeeping and communications) what his questions would be. As you’d expect, Robert had some suggestions for hard-hitting questions about exactly what the Power-One patents cover. I don’t know if we’ll have time to cover that topic. (On the other hand, it’s open mike, so who knows?) But he also had some excellent questions about the tools as well as the mindset needed for designing with digital power:

The supplier that provides the tools that makes it easiest for the customer to configure, control, and monitor their product will win the biggest share of business - not the supplier with the best technical solution. So, what role do the tools play?
And is there a need for a "universal tool"? Is each supplier having its own tools a hindrance to the use of digital power by OEMs?
What are the OEMs thinking now about digital control loops (that is digital PWM as opposed to digital housekeeping circuits). Is there an advantage? If so, what is it? Flexibility? Cost? Are they using digital PWM? Do they care what is in the power supplies they buy?
What do OEMs think about PMBus? Using it? Requiring it? Avoiding it? Still watching? Dell and Intel are using PMBus, why aren’t the others?
Is there an education gap? Is one of the reasons digital control isn’t going gangbusters (at least on surface appearance) is that everyone is afraid of writing code and Z transforms? Is it a matter of being afraid of the unknown?

Feel free to add you own questions for these panelists in the comments section below. Better yet, come on over to the Digital Power Forum on Monday afternoon and step up to the mike! The Digital Power Forum is September 21-23 in Santa Ana, CA

Posted by Margery Conner on September 14, 2009 | Comments (1)

September 16, 2009

In response to: What does the future look like for digital power? A call for questions
Keng Wu commented:

In my humble opinion, employing digital blocks in peripheral power circuits including sensing, monitoring, protection, status reporting with I2C serial bus, etc in open-loop form shall not, and does not, deserve the status of ?digital power?. The term must be invoked only to the extent to mean a closed-loop control mechanism that totally replaces both the conventional analog error (compensation) amplifier and the analog pulse-width-modulator existing in traditional switch-mode power supplies. In that sense, a ?DIGITAL POWER? converter must have a DIGITAL FILTER and a DIGITAL COUNTER/CLOCK in the control loop. Combining both and with the addition of oscillator, ADCs, and registers for Z-transform coefficients, a standing alone digital controller is born. Talking about DIGITAL FILTER and Z-transform, again my humble opinion, those are the stumbling blocks why ?digital power? is yet to spread. Digital filters are highly mature in image and high-end radar signal processing; since they are all treating low level signals and, in most cases, post-processing, non-real-time. Power processors are however handling very large signal in real-time. Examining conventional SMPS, one quickly realizes that only the error amplifier and the PWM blocks can be replaced by digital circuits, be it Programmable Logic Array with firmware or dedicated microcontroller. The conversion of analog error amplifier to digital filter is not simple at all. Yes, there are MATLAB commands to covert s-domain transfer functions to selected Z-transform with all coefficients provided. But it requires prior knowledge of the analog transfer function. In addition, the designer has to pick his digital sampling clock. One cannot just PICK a clock from thin air. This is problem No. 1. And, of course, problem No. 2 is the analog transfer function to be converted. It seems that a digital power converter has to be first designed along the conventional SMPS approach. Then the designer has to have the ability to IDENTIFY and EXTRACT the analog error amplifier in functional form. The effort required has long been a lost art. Then, what else is standing in the way to digital power? That is problem No. 3. It has been well understood that current-mode control is superior to voltage-mode control for SMPS. However, transporting this advantage to the digital domain is NOT that simple. Current signals, derived from either inductor or power switch, are very noisy and present high slew rate. Digitizing such signals is extremely troublesome and generates peculiar perturbation easily to the control loop. It is better to stop here. Otherwise, this is turning into a Ph.D. dissertation. But, where there are problems, there are also opportunities, and plenty.

About EDN | Site Map | Contact Us | Subscription | RSS

POST A COMMENT
Display Name

Before submitting this form, please type the characters displayed above. Note the letters are case sensitive: