Thin is In – Making Small, Highly Efficient Power Supplies

There is a huge pressure today to make electronics thin. Televisions must have larger and larger screens, but they must also be thin to easily mount on the wall and fit into any home decor. The same is true about media center computers, devices that coordinate home entertainment, media storage, video-on-demand and the rest of the smart living room. Everything has to get thinner: video games, laptop computers, printers, scanners, copiers, FAX machines, radios and anything else vying for precious space in the office or house, whether living room, kitchen or bedroom. “Slim” is the big selling point and sometimes prominent in a product name. Even portable devices are being squeezed, with thin laptops and super-thin, tiny laptop AC adapters selling for a premium.

To a certain extent, things can get thinner by basic mechanical design. If everything is positioned in one plane rather than stacked up, you can put it in one large, flat enclosure. This style was called the “pizza box” because it resembled the 16-inch square by two-inches high cardboard box for a take-out pizza.

But to get even thinner, serious engineering on all disciplines is required. To shrink a disk drive to fit in an ultra-compact laptop, for example, the number of platters has to be reduced, the platter spacing is reduced, glass platters replace aluminum platters, motors are developed that fit inside the platters rather than under them, and the electronics is miniaturized and built on extremely thin PC boards that are shaped to fit in the corners around the platters rather than under the platters.

Another thick element used in all home and office equipment is the power supply. Making a very thin power supply is equally challenging, because many of the components required are inherently thick: notably the capacitors, inductors, transformers, fans and heat sinks.

There are other significant challenges in making a modern, small, thin power supply. The power supply also must be extremely efficient. To reduce audible noise and save cost, fans and forced air cooling are being eliminated, leaving a sealed plastic box with little ability to dissipate heat. Efficiency standards such as ENERGY STAR® and 80 PLUS® require higher efficiency, both at full load and partial load. In addition, aggressive competition in consumer electronics requires constant price reduction even though larger TV sets and higher performance processors require more power.

In the 1960s and 1970s, the vast majority of power supplies used 50Hz/60Hz power transformers and linear voltage regulators. These power supplies were rarely better than 50 percent efficient. Also, they demanded extremely high peak current from the power grid at the peak of the sine wave. This led to distortion in the power line waveform, overheating of the transformers in the power grid, and most importantly, electromagnetic interference (EMI).

In the 1980s, PCs and workstations were introduced into the office. Because of limited cooling, the earliest PCs used switching power supplies and fans. The average PC consumed 200 watts of power and was conveniently plugged into the wall, but compute-intensive workstations with large disk drives consumed over 1200 watts, more than could be drawn from a 15-amp circuit with a high peak-current switching power supply, let alone a linear supply. These workstations were the earliest adopters of power-factor correction, not because of EMI or power factor specifications, but because of the high power required.

A power factor corrector (PFC) is commonly built as a boost converter followed by a step-down forward or flyback power stage. A boost converter with adjustable duty cycle can produce line current that tracks line voltage if the duty cycle is set correctly. The first PFC ICs accomplished the boost function with a fixed frequency, continuous-conduction mode (CCM) power supply that used feedback to set average input current directly proportional to power line voltage. These converters used a large inductor and kept input ripple current low to allow easier EMI filtering.

More recent PFC ICs regulate input current using a variable frequency (hysteretic) boost stage that operates on the border between continuous and discontinuous, called transition mode (TM). Transition mode power stages have a few advantages over CCM power stages. For the same power level, the inductor can be smaller, which translates to thinner, lighter and less expensive. Also, the boost rectifier is always switched off at zero current (zero-current switching or ZCS), so a less expensive diode can be used with little switching energy loss. Transition mode power stages require more input filtering than CCM power stages because current ripple is higher, but this is more than offset by the reduced cost of the switching diode and reduced size of the input inductor.

Transition mode boost PFCs have been used successfully in volume power supplies up to 300 watts. At higher power levels, the power inductor becomes large and impractical. Also, the input EMI filter becomes much more expensive.

Recent efforts to produce thinner, more efficient, and less expensive power supplies resulted in development of a novel integrated circuit that implements two interleaved channels of TM PFC in one device, the UCC28060. The UCC28060 takes advantage of all of the benefits of TM and delivers the advantages of interleaving to make up for the disadvantages of TM. For example, as shown in Figures 1A and 1B, interleaving reduces input ripple current, so a much smaller EMI filter can be used. Interleaving also reduces output ripple, allowing for a smaller output capacitor and longer capacitor lifetime. Interleaving also splits one huge inductor into two smaller inductors, which increases power level and enables thinner packaging.

Fig 1

If interleaved power converters are operating at 50 percent duty cycle, the input ripple from one phase completely cancels the input ripple from the second phase, producing input current that has no switching component and produces no EMI. As duty cycle deviates from 50 percent, cancellation reduces and residual ripple increases (see Figure 2). However, much of the benefit of ripple cancellation is still realized.

Fig 2

Figures 3A and 3B show two prototype power supplies. Figure 3A is a single-phase CCM PFC with a forward converter for output regulation and an EMI filter. In this power supply the inductor, bulk filter capacitor and heat sink are all huge and centrally located, resulting in a thick, bulky power supply. Figure 3B is a two-phase interleaved TM PFC. The interleaved power supply is considerably thinner because the inductor size is reduced, output filter capacitor size, and layout considerations for heat dissipation. If laid out with equivalent rules, the interleaved TM PFC also would be noticeably smaller.

Another way to implement a thinner and more efficient power supply is to use a direct flyback interleaved PFC-based on the UCC28060. Although not practical at very high power levels, the direct flyback interleaved PFC uses only one power conversion operation, so the parts count is lower and the efficiency is higher. If thickness is truly the most critical factor, planar magnetics could be used also.

Fig 3a

Fig 3b


In a perfect world, one 600 watt power stage would be smaller and less expensive than two 300 watt power stages. In practice, two 300 watt interleaved power stages are actually smaller and less expensive than the single 600 watt power stage. Even more importantly, lower-power stages are thinner and more practical to implement. A typical power factor corrector can be easily produced that is 94 – 96 percent efficient. For a 600 watt system, this means that the power stage can dissipate ~30 watts of heat, a significant challenge in a box without forced air and with minimal convection. The vast majority of that heat is dissipated in the input bridge, the inductor core, and the power FET. By implementing the 600 watt system as two 300 watt interleaved stages, heat can be spread among two inductors, two power FETs and two boost diodes, and these can be placed in different parts of the power supply to further spread the heat. This greatly simplifies the job of designing heat sinks and air flow, leading to a thinner, less expensive power supply.

Fig 4


Fig 5


Figure 4 shows the schematic diagram of an actual interleaved PFC preregulator. Although the parts count is higher than for a single-stage PFC, the components are smaller and less expensive. To improve light load efficiency, the UCC28060 also includes programmable phase management controls that reduce operation to one phase at light load and switch to burst operation at very low load.

Conclusion

In summary, the TM Interleaved topology is empowering a new generation of power supply engineers to cross thickness boundaries their forefathers imagined and customers dreamed about hanging on their family room walls. The fact that Interleaving with phase management delivers efficiency gains, especially at light load, makes the solution even more compelling for today’s engineer challenged by a world growing increasingly conscious of energy consumption. The UCC28060 from Texas Instruments takes that dream one step further by making the solution cost effective, easy to implement and loaded with features which ensures overall system safety.

References

For more information on creating a thin interleaved power factor corrector using the UCC28060, visit: www.ti.com/sc/device/ucc28060.

About the Author

Bob Neidorff is an IC Design Engineer and TI Fellow with High Performance Analog at Texas Instruments. For over 30 years, he has designed and managed the design of a wide variety of power and analog circuits.

Texas Instruments