Advertisement

Raising the bar in isolated dc/dc brick module technology

Raising the bar in isolated dc/dc brick module technology

Understanding the changes in the power supply design fronts is crucial

BY BILL ANDREYCAK
Texas Instruments
Dallas
http://ti.com

One look at any of the new products introduced each week in the modular, isolated dc/dc “brick” power supply market reveals that incredible technology advancements are taking place. In the past year alone, power densities have doubled, so what once required a quarter-brick solution is now supported by a solution that's half the original volume.

Opening photo

And generally, those supplies are available from numerous suppliers at mere fractions of the cost of their predecessors. In this competitive, cost-driven business, capitalizing on the latest design, component and packing technologies is essential for survival.

Notable changes occur on three key design fronts: distributed-power system architecture, switch-mode topology selection and robust-control pulse-width modulators (PWMs). In each example, the catalyst has been reducing size, increasing power density, and maximizing efficiency.

The changing power system architecture
Due to the number of low-voltage supplies required by most systems, the use of several single-output, isolated converters becomes unattractive from both a cost and size standpoint. Instead, the most recent system power distribution trend incorporates a single isolation and step-down converter, often referred to as an isolated bus converter (IBC). These power supplies deliver an isolated low-voltage dc bus in the 5 to 12-V range to feed localized point-of-load (POL) nonisolated dc/dc modules.

What is unique is the delivered voltage is generally semi-regulated and sets within a specified range. The more popular design approach has been with a feedback-less power stage, so the output voltage will depend on the IBC's input-voltage variations, which can be minimal.

The “bus converter” output is then piped into high-density nonisolated dc/dc converters, one for each output voltage needed. A variety of industry- standard modules and discrete solutions have been optimized to provide tight regulation at very high efficiencies right at the point of load. A huge assortment of output voltages and currents are readily available from multiple vendors.

Topology focus
Due to its overall simplicity and excellent transformer core use, the push-pull topology has been supporting module applications for many years. There is a slow but increasing migration away from the push-pull topology in favor of other solutions.

Interest in interleaving forward converters is significant, both for the conventional, single switch and active clamp/reset versions. Much like their nonisolated dc/dc counterparts, “brick” designers seek advantages of using multiple phase power paths to generate high-current outputs.

In comparison to the push-pull, each interleaved forward converter runs at one-half the switching frequency (see Fig. 1 ). This translates into lower magnetic core losses, although, a key benefit is the ability to half the individual converter currents.

Fig. 1. The interleaved forward converter offers lower magnetic core losses and half the individual converter currents.

Additionally, interleaving spreads out each transformer and inductor power dissipation over two magnetic components, often having a combined larger surface area thus lowering component temperature. In height restrictive applications it is also generally easier to select and fit two smaller magnetic elements than one processing twice the power in a given area.

Interleaving also has one unique technical attribute that warrants attention. Compared with a single forward converter, interleaving reduces the input current ripple by a factor of the square root of two. On the secondary side, when operating below 50% duty cycle, one converter's inductor current is charging while the other converter's inductor is discharging. At precisely 50% duty cycle, the output inductor ripple currents completely cancel one another and the ripple current is zero.

This would be very beneficial in applications with a limited input voltage span, hence narrow duty cycle range. Interleaving also leaves the opportunity to achieve higher-frequency conversion using smaller magnetic devices provided that overall conversion efficiency doesn't suffer, as in the case of the zero-voltage-switched active-clamp/reset technique.

Managing material costs

In many brick power supply designs, integrated magnetic elements use pc board layers as the transformer and inductor windings. One practical advantage of the conventional forward or active reset/clamp technique is the simplicity of the transformer construction.

It uses only one primary side and one secondary side winding requiring only a two-sided pc board. The push-pull requires a total of four windings, each of which generally, but not always, requires one pc board layer per winding.

Assuming that all other electrical inter-connections can be made on the pc board, the Forward would result in the lowest cost pc board solution needing only a double-sided pc board. Often, the choice of how many layers to use may depend on how much copper is actually needed altogether to keep losses at an acceptable level.

Control circuits
Higher levels of integration within the PWM controllers reduce the external component count. One challenging, but necessary, function in most power supply designs is generating a startup bias supply. While most PWM controllers feature a low startup current and undervoltage lockout thresholds with volts of hysteresis, often there are other system issues that complicate a simple bias facilitation.

Once the control IC is functional, its supply and gate drive current increases and an auxiliary power source is needed. The start-up bias network is commonly switched out of operation to increase efficiency once soft start is complete and the bootstrap supply is operational.

There's not much circuitry required to perform this task, but it's not free of cost and consumes valuable pc board area. Due to advanced high voltage IC process technology, many new controllers are integrating this high voltage start-up circuit on-chip. Ratings of 110 Vdc at 10 mA or more simplify the startup task to a single pin connection to the supply input voltage.

While the sophistication of control, diagnostic and interface logic continues to increase, so too will the overall parts count. Even diminutive surface-mount-packaged amplifiers, comparators, and timers, in addition to their associated resistors and capacitors, all consume valuable board real estate. However, these straightforward low-cost solutions are counterproductive to achieving high power density because each requires room.

In many situations, enlisting the use of a microcontroller is a logical choice and may well be the most cost-effective solution too. Microcontrollers commonly house numerous I/O pins along with several A/D interfaces and can be programmed to perform the variety of tasks required.

Functions such as tracking or sequencing of the output voltage, power-good, and auto-retry following a fault are easily configurable. But most of the tasks are limited to speeds well below the converter's switching frequency due to the limited performance of inexpensive microcontrollers. Any high-speed activity, for example, current limit protection in the order of nanoseconds is far beyond the capabilities of all but possibly the fastest microcontrollers. High-speed DSPs might be applicable, although a portion of their overall capability could be consumed on this single task.

So while the trend in control logic could be toward digital, analog circuitry will remain irreplaceable for some functions, at least for now. Superior performance has an associated cost, and in the design decision process containing conflicting interests, lowest cost often wins over enhanced technology. Partitioning the control logic into what is best done by digital or analog circuitry will certainly influence new designs and potentially choreograph new IC solutions.

Undeniably, the expectations to achieve higher power densities at lower overall cost are continually challenging power supply designs. In response, the cumulative effects of advancements in distributed-power system architecture, topology selection, and control IC technology address these industry needs.

Advertisement

Leave a Reply