How to Correctly Apply the Right Power Devices to Meet Industrial Power Supply Requirements

AC/DC power conversion (also called “offline” power conversion) for industrial applications is very different from that used in consumer and mass-market designs. In general, in industrial applications the voltages, currents, and power levels are much higher; there are additional thermal and electrical ruggedness requirements; strict regulatory mandates are in place for active and standby; and on-going operation must be monitored for feedback loops and for fault detection.

The key to effective design under these circumstances is the power converter control circuit or device at the core of the converter, along with switching power devices and their support components. These are used to implement the selected power supply topology and deliver a regulated DC output at the desired voltage and current. The converter may include an integral power device (such as a MOSFET) or act as a controller and driver for an external, discrete power device, including silicon carbide (SiC) power devices. While some converters supply a fixed DC rail to an entire system, others function in less visible but still vital roles as gate drivers requiring special turn-on/off attributes.

This article describes various power converter topologies suitable for industrial power applications and the factors designers must consider before making a choice of topology and its related components. Along the way it introduces example components from Rohm Semiconductor and how to go about applying them effectively.

Power converter topology selection Designers must balance the multiple options and trade-offs versus project priorities when choosing power converters and developing a supply that meets the challenging demands of industrial applications. Although there are many ways to do this, the most common approach is to start with the power (in watts) that the supply needs to deliver, as well as any need for input and output isolation (Figure 1 and Figure 2). These two factors will point toward possible choices for power converter topologies.

Note that these are all switch-mode power supply (SMPS) topologies; none are linear-mode supplies. Linear supplies are used only in those niche applications where their extremely low output noise is critically needed and so their relatively low efficiency of 20% to 40% is tolerable and acceptable. However, such low power-related noise is rarely needed in industrial installations.

The reality, of course, is that determining the “right” approach is often not a simple decision, as there many factors—such as the following—that can affect the decision:

• Basic performance: Includes input and output regulation and transient response
• Ruggedness: Some approaches are more tolerant of, and resistant to, electrical and thermal stresses under some scenarios
• Operating mode: The power supply can be continuous, pulsed, or highly intermittent
• Above-nominal power demands on the supply
• Solution cost
• Need for isolation
• Efficiency: A single, short word with many implications

AC line isolation is inherent in nearly all industrial power converters via the transformer used for voltage step-up/down, user safety, and system performance. Still, even with a primary side transformer, some converters also need internal input/output isolation—sometimes called floating output—for their own operation, for electrical separation between multiple rails, or for high-voltage-rail bootstrapping. This input/output isolation can be implemented using an additional transformer or optocoupler.

Efficiency requirements dominate many designs choices

Any discussion of industrial power converters must consider efficiency as a top priority. Unlike battery-powered devices where efficiency is most closely associated with runtime, efficiency for AC/DC converters is driven by other objectives:

• Operating costs: Many industrial applications require hundreds and thousands of watts; this is especially critical as many such applications run 24/7.
• Dissipation of excess heat: The ambient temperatures in many installations can be high due to limited access to airflow or absence of active cooling. This heat results in component stress, reduced time to failure, and need for replacement downtime and cost. Power converter inefficiency adds to already high ambient heat load.
• Regulatory issues: There are many standards and mandates that set efficiency minimums as a function of application, power level, and region. These standards also define allowable power factor minimums and thus may require addition of power factor correction (PFC) in the power converter and supply.

Basic math shows why even slightly improved efficiency is critical and meaningful. Consider a power converter in which efficiency is increased from 65% to 70%—a seemingly modest improvement of five percentage points. Now look at it from the complementary perspective: inefficiency has dropped from 35% to 30%. While this is also an improvement of five percentage points, it is a decrease in inefficiency of 5/35 or about 14%. So, increasing efficiency from 65% to 70% is also a decrease in inefficiency—and thus cost and heat load, and perhaps need for additional cooling—by 14%. This is a significant improvement and is directly reflected in thermal design requirements and operating costs.

Achieving higher efficiency

There is no single “magic bullet” that leads to a more efficient AC/DC converter design, especially when designers are trying to squeeze out a few more percentage points of improvement. Instead, progress is the result of a combination of several large and smaller scale strategies:

• Choosing an appropriate core topology for the converter and determining the switching frequency that works best for that approach and power level; this frequency is typically between 100 kilohertz (kHz) and 1 megahertz (MHz).
• Optimizing the circuit: there are many small sources of inefficiency in any basic design, and power supply designers have identified ways to partially or largely minimize these; while each may represent a small improvement, they add up.
• Using active and passive components that are inherently conducive to higher efficiency; for power devices (MOSFETs) and some diodes, this now means migrating to components based on SiC process technology.

SiC technology is now the most viable candidate for next-generation, low-loss switching and blocking elements due to its low ON resistance and superior characteristics under high temperatures. It offers many advantages compared to silicon alone due to its higher breakdown voltages and other characteristics, including: 

• Much higher critical breakdown electric field voltage, so operation at a given voltage rating is possible with a much thinner layer, greatly reducing on-resistance.
• Higher thermal conductivity, resulting in higher current density in a cross-sectional area.
• Wider bandgap resulting in lower leakage current at high temperatures. For this reason, SiC diodes and FETs are often referred to as wide bandgap (WBG) devices.

As a rough “order of magnitude” comparison, SiC-based MOSFET devices can block voltages up to ten times higher than silicon devices, and can switch about ten times faster with an ON resistance of half or less at 25°C. At the same time, their ability to operate at much higher temperatures of 200°C versus 125°C eases thermal design and management. An example of the power-handling capability of a SiC switching device is Rohm Semiconductor’s SCT3105KRC14, a 1200 volt, 24 ampere (A), N-channel SiC power MOSFET with a typical R_DS(on) of 105 milliohms (mΩ). This device features a favorable thermal resistance characteristic that shows it quickly reaching its maximum value versus applied pulse width (Figure 3). 

Discrete vs. integrated power designs An option at lower power levels is to choose an IC that combines the converter’s regulator with the associated power switching device. The advantage of this is that the interconnection between the regulator and power device has an optimized path, and the unavoidable parasitics are characterized in the specifications provided by the data sheet. Further, the need for external components is minimized as shown by the Rohm BD9G341AEFJ-E2, a buck switching regulator with an integrated 150 mΩ power MOSFET, (Figure 4).

This miniature HTSOP-J8 device measures 4.9 × 6.0 × 1.0 millimeters (mm) and is well suited to industrial distributed power applications. It accepts a 12 volt to 76 volt input and delivers up to 3 A output current. Its current-mode architecture provides fast transient response and a simple phase compensation setup, while supporting a user settable switching frequency from 50 kHz to 750 kHz. As power levels (and thus voltage and current) increase, the packaging of power devices becomes more critical and the use of separate individual devices becomes more challenging. In such situations, a pre-packaged module with two or more power devices is an attractive option. For example, Rohm’s BSM300D12P2E001 is a half-bridge module with two SiC double-diffusion MOSFETS (DMOSFETs) and SiC Schottky barrier diodes that can handle 1200 volts and 300 A (Figure 5).

Including the MOSFETs and their diodes in a single module optimizes performance of the total assembly, which looks like a long, thin brick and measures about 152 mm long × 62 mm wide × 17 mm high (Figure 6). The module also includes an independent temperature sensor (NTC thermistor) to enable monitoring of its thermal situation, while its construction encourages improved thermal management—a major consideration at this voltage and current pairing. It does so by making it easier to physically attach the components to a circuit board or heatsink while providing mechanical integrity and ensuring solid connections of the robust power leads.

The criticality of drivers for effective converter operation

Whether silicon- or SiC-based, MOSFETs must be switched on and off with careful consideration of many details related to the gate drive voltages, currents, slew rates, transients, overshoot, input capacitance, and inductance, as well as many other static and dynamic factors. A gate driver functions as the interface between the low-level, relatively simple signal coming from the output of the controlling processor and the gate input of the switching device. It is a special type of power converter with an output that matches the requirements of the power device as its load.

When a pair of switching devices is used in a common arrangement such as the half- or full-bridge, the driver block must also ensure that the high- and low-side devices are never turned on at the same time even for an instant, as this would cause a short circuit between the power rail and ground. Further, in some power device applications, one or both of the power device paths must be galvanically isolated from system ground while still providing matched performance to each. To meet these requirements, some power device vendors offer a driver IC specifically designed for one or more of their switching devices. For example, the characteristics and features of the Tamura/Rohm2DU180506MR02 half-bridge gate driver IC complement the Rohm half-bridge module cited above. It simplifies the challenge of providing the specified drive to that module while adding various protection modes (Figure 7).

This gate driver is packaged as a small module measuring 24 mm high that is mounted on a circuit board measuring 65 mm × 100 mm. The board provides connectors for DC power, processor interface and power module drive. The gate driver also provides critical supervisory functions needed by nearly all power devices, especially those for higher power industrial applications. These include overload protection, overheating protection (it interfaces with the thermistor, which is part of the power module), undervoltage lockout, and a gate drive fault indicator. Other half-bridge gate drivers are more general purpose. The Rohm BM60212FV is a 1200 volt high-side and low-side gate driver IC for N-channel MOSFETs and IGBTs (Figure 8). It uses a coreless transformer providing magnetic isolation to implement the required level shifting for the high side. However, it is formally classified as a non-isolated gate driver since the remainder of its internal functions are not isolated.

This diminutive part, housed in an SSOP-20W package measuring 6.5 × 8.1 × 2.0 mm, is compatible with 3 volt and 5 volt drive signals, and includes undervoltage lockout among other features. Note that this IC is AEC-Q100 qualified, meaning it meets stringent automotive reliability standards. Although “automotive qualified” is not “industrial,” some designers prefer to specify AEC-Q100-qualified parts on their BOM as one way to enhance the reliability of their products under environmentally challenging industrial conditions. These conditions include electrical surges and EMI/RFI, thermal stresses of temperature extremes, and mechanical failures that result from thermal cycling and vibration.

Measuring the current

In many power converter designs it’s necessary to know the amount of current flowing from the output to the load, but it is almost always critical in industrial applications. In some cases, this current value is needed to provide feedback to the converter for closed-loop performance; in industrial settings it is also needed to monitor the load and situations such as a motor that is stalled or failing. One of the ways to continuously measure this current in real time is by sensing the voltage across a resistor in series with the load. This is often called a shunt resistor, although that term is a misnomer in this context.

Conceptually, the measurement of current this way is a straightforward application of Ohm’s law. In practice, such as in a high-current industrial converter installation, it brings with it multiple challenges. First, designers must determine an appropriate resistance value. There’s a trade-off here: a larger value resistor provides more IR drop and thus better resolution and noise immunity, but it also dissipates more power, reduces the rail voltage to the load, and can adversely affect the stability of the controller/load loop.

In general, a good starting point is to choose a resistor value such that the voltage drop across it will be around 100 millivolts (mV) at maximum current. Doing the math, this leads to sense resistor values that are in the milliohm range, which is in sharp contrast to the kiloohm and higher values that are common in other circuit functions.

Once the resistor value is determined, designers must select a specific physical component. Due to the magnitude of the current values, this resistor must have a relatively high wattage rating compared to most other resistors. Further, it must be more than just accurate at room temperature; instead, it must use materials and fabrication techniques that together ensure a low temperature coefficient of resistance (TCR). With a low TCR, its value will not change appreciably despite increases in ambient temperature or temperature increase due to self-heating. The Rohm PSR400ITQFF0L50 shunt resistor is a good example showing the complexity of this deceptively simple passive component. This is a 500 microohm (µΩ)—yes, one-half of a milliohm—±1%, 4 watt metal element device (Figure 9).

Although the PSR400ITQFF0L50 may look like a simple piece of bent metal, appearances can be misleading. This 5.2 × 10 mm component is fabricated from a careful blend of copper and metal hydrides to yield a TCR of ±175 parts per million per degree Celsius (ppm/⁰C). Other current sense resistors in the family have both lower and higher TCRs. For comparison, a standard, low-cost, “garden variety” resistor has a TCR of about ±2000 to ±4000 ppm/⁰C, or between ten to twenty times higher than these ultra-low-ohmic metal plate high-power shunt resistors.

When using shunt resistors at high current values, their physical mounting and heat sinking are serious considerations, as are their electrical connections. With a resistor in the milliohm range, the required four-wire Kelvin connection points must have extremely low resistance. They must also be physically located so they produce valid readings not corrupted by those connection resistances.

Conclusion

Designers of industrial power supplies and converters face a unique set of challenges to achieving performance, cost, space, and reliability requirements due to the stresses of the application environment. The high power levels require extra consideration to efficiency, thermal dissipation, and packaging. There are also gate driver and current sense issues to be addressed. With due consideration to the requirements of the application, the core building blocks of an industrial power supply shown here—including discrete devices, integrated, and modular power components—can readily meet the challenges of industrial-class power supplies and converters.