While Power Technologies Met 2012’s Challenges, 2013’s Pitfalls Await

It’s always interesting to look back over the past year and reflect on where we have been, where we are headed, and the changes and adjustments that occur along the way. As an engineer for more than 30 years, for example, I felt a certain sadness following the retirement of the Space Shuttle.

The shuttle program was one of the largest engineering endeavors (sorry, I couldn’t resist the pun) of our generation. It held special meaning, as it was one of the first high-reliability space projects I worked on as a young engineer. It was a truly amazing accomplishment, especially considering that we did not have the PC when the program started, nor did we have highly integrated electronics. The most sophisticated electronics on the Space Shuttle would pale in comparison to a typical cell phone today. Oh, if only the microprocessor were available at that time.

I remember the design of the radar and range indicator that we lovingly referred to as the RRI. It consisted of 13 multilayer transistor-transistor logic (TTL) boards that decoded and displayed the distance and rate of approach for docking. The equivalent complexity could be easily accomplished with an inexpensive PIC processor today. In any event, it is a reminder of how far we have come, how long I have been at this game, and that there is more where those challenges came from.

Power’s Prominence

Much of the focus in the world of power electronics is often viewed as “same stuff, different year,” as in build it smaller and cheaper with more integration and higher efficiency. Always one to push back a bit, I wrote an article this year about the negative impacts of high-efficiency power supplies (see “Low Power And High Efficiency Can Degrade Performance Significantly”).¹ There are many places where efficiency is the number one figure of merit, but as with most things in life, nothing is free.

The pulse-frequency modulation (PFM) and burst mode generated by some of these high-efficiency devices can wreak havoc in instrument systems, especially in analog-to-digital converter (ADC) clocks, as well as low-noise amplifiers (LNAs) and other sensitive electronics. These burst modes also can result in very low-frequency electromagnetic interference (EMI), which is difficult to filter. Nonetheless, technology moves ahead and voltages will continue to get lower, switching frequencies will get higher, and new technologies will emerge to support these changes.

For instance, the Semtech SC221 point-of-load regulator (POL) operates at the fastest switching frequency to date—20 MHz. Such high-frequency switching eliminates output inductors, at least as we know them. The SC221 uses a patented approach of two oppositely wound spiral air core inductors made using traces on the printed-circuit board (PCB), minimizing the EMI from the inductors. This high frequency also allows very small capacitors and very fast response.

Much of the focus on the newer POLs is on lower output voltages, as 1 V has become the norm as we work toward 0.6 V. In measuring some of the higher-speed POLs this year, for the first time, we measured the switch speed at just about 1 ns. This is a milestone, and I published an article this year that focuses on the requirements to measure this speed (see “How To Measure Ultra-Low Impedances”).² One thing that became very clear to me is that average power electronics engineers are severely underequipped and lack the necessary test equipment to accurately measure the devices they are designing and manufacturing.

The world of isolated POLs has also continued the trend toward smaller sizes and better efficiency. The latest introduction from Vicor, an 850-W eighth-brick converter, is a long way from the 100-W full-brick converters we were all working to develop in the 1980s.

This year will likely be remembered as the year that enhancement-mode gallium-nitride (eGaN) FETs started to grow roots. It is reminiscent of the introduction of the hexfet in 1979 or 1980. I recall waiting and finally receiving my first samples, which were all destroyed within a matter of an hour or so. We very quickly learned many lessons from those failures, such as how speed comes at a price, mostly through the interconnecting inductance that bipolar junction transistors (BJTs) could survive and hexfets could not.

In many ways, eGaNs FETs are sparking some déjà vu. The devices offer higher speeds, lower resistance, lower capacitance, and very sensitive gates. We will learn to use them effectively, and this is the year we will recall they started to take hold. Silicon carbide (SiC) seems to have a place in the world, but I clearly see many more articles about eGaN than about SiC.

Digital power has also generated a lot interest this year. There are many approaches to designing digital power, from analog control with digital communications to complete digital control, using high-speed DSPs. As each company attempts to find its way in the digitally enabled world, new problems challenge engineers.

I have written a great deal about the unavoidable collision of the instrumentation, RF, and power supply worlds (see “New Injector Supports Testing Of POLs” and “Assessing Point-Of-Load Regulators Using Non-Invasive Techniques”).^{3, 4} As these changes take place, power engineers are forced to learn how to deal with RF issues due to the higher speeds, as well as how to convert our minds from thinking about resistors and capacitors to create the poles and zeros in our compensation paths to Z domain parameters to code them in software. The upsides of digitally controlled power supplies are often worth the effort since they allow direct communication between the processor and the power supply to optimize performance.

The Right Tools

With all of these changes, we certainly need new equipment to support the higher fidelity measurements required by all of these newer higher-speed technologies. The introduction of new oscilloscopes, such as the MDO from Tektronix and the HRO from LeCroy, offer new highs in the world of high fidelity measurement. New high-performance signal injectors from Picotest allow us to inject and extract high-fidelity signals and generate non-invasive load steps, as well as generate signal integrity (SI) and power integrity (PI) disturbances to ensure the necessary performance levels of the faster and lower-voltage electronics.

We have also seen the world of smart phones and tablet computers explode. A recent study showed that 82% of travelers carry a smart phone and that approximately 38% travel with a tablet or a laptop computer. This is fueled by the higher processing power available for smart phones and tablets along with decreasing cost and longer battery life.

The challenges for the manufacturers of these devices is to continue to improve battery life, while providing faster processors (most are now considerably faster than 1 GHz), high-quality audio, high-resolution screens, and more RF connections than most can imagine in a small handheld device. The stakes are high as the demand for these devices is growing and everyone wants market share.

This war continues to apply further pressure on higher-performance devices. The significant decline of RIM (Research in Motion) this year shows that the smart-phone market isn’t easily placated and it’s easy to lose in this business unless the technology is pushed to its limits.

Despite all of these changes and improvements, our own business experience shows that engineers are not receiving the proper training or equipment needed to develop these newer higher-speed systems. We continuously receive bad or erroneous data, both from IC manufacturers and end unit producers. Many engineers do not have the equipment that’s required to properly measure the performance of their circuits. The continuing trend of shorter development cycles applies even greater pressures in this area.

We also see evidence of a lack of training in the very low attendance at the workshops and seminars that provide the understanding necessary to improve the engineering development process. Companies seem to be more focused on the short term than ever before, and most have either eliminated or greatly reduced the training budgets for their engineering staff.

I always found it interesting that as critical and cutting edge as engineering is, it does not have any formal requirement for continuing education. This is sad, as we all know we will eventually pay a price for designs by undereducated designers. Workshops and seminars are under-attended, and trade magazines aren’t being read or even published. I hope to see this trend reverse in the next few years so we can continue the path of knowledge, passing it along to others in an effort to support continuous improvement.