Effective heat dissipation is essential to ensure continued and effective power control system operation. The caliber of heat sink used with your power control can determine the reliability of the entire system.

This cross-section of a wing in an air tunnel, taken during a wind tunnel study done by Payne Engineering in the 1950s, demonstrates the principles of laminar airflow.


With microprocessor technology heating up the manufacturing industry, heat dissipation is becoming the new rocket science. Unfortunately, many commercially available heat sinks are poor substitutes for a quality heat exchanger. Microprocessor heat sinks may be adequate in applications where you need only dissipate 30 W. But what if you need to dissipate 250 W? Or 2,400 W? With 100 percent reliability?

All silicon devices experience a forward-conducting voltage drop of roughly 0.5 to 2 V across the semiconductor device. If not for this forward voltage drop, you would not have any heat to exchange. For microprocessors running low amperes, the heat can be discarded relatively easily. But, what about power silicon applications? Thyristors switching and controlling hundreds of amps at 480 V or higher can generate serious heat that must be exchanged reliably. It will certainly take more than an off-the-shelf heat sink to accomplish that task.

Thyristors are employed in solid-state power controllers from 1 kW to 1.5 MW and have been in use since the 1960s. SCR-based devices are rapidly replacing electromechanical contacts and motor starters in nearly every industry. Solid-state power control is a clean, precise, sophisticated technology with a central theme: heat dissipation. The heart of the SCR is the silicon pellet, which is the reason for temperature concerns. The ampere capacity of any SCR is approximately inversely proportional to the temperature of the silicon itself (in applications with case temperatures of more than 212°F [100°C]). The critical task then becomes reliable cooling the SCR.

The typical thermal radiation of the heat sink is simulated.

Heat transfer can occur in three ways:
  • Radiation. Heat is transferred, usually into the atmosphere, by electromagnetic waves leaving the surface of an object.
  • Conduction. Heat is transferred through a solid mass.
  • Convection. Heat is transferred to a moving fluid, usually air, by a process of heated air being replaced by cooler air, either naturally or by force.
Every heat sink requires conduction as the first step in dissipating heat. Clean contact with the device that is generating the heat allows the heat sink to act as a conveyor carrying the heat away from the silicon device. Radiation plays less of a role in many heat sink designs simply because convection is where most of the gains are made in heat dissipation technology.

The typical thermal radiation of an extruded heat sink is simulated.

Heat sinks are rated according to how efficiently they can dissipate heat. This rating is referred to as °C per watt. Simply put, if a 50 W thermal load is applied to a heat sink, which raises the heat sink temperature to 77°F (25°C), then the heat sink rating equals 0.5°C/W. The higher the number, the more efficient the heat sink is at dissipating heat.

After designing and testing heat sinks since the 1960s, it has been found that by applying laminar flow fluid dynamic principles, heat sinks for every power control application can be produced. Extensive testing has yielded interesting results in the area of natural convection. The best heat sink designs flow up to 1 m/sec of air, which is a high speed in natural convection terms. Laminar flow describes the condition of the boundary layer of air as it passes over the surface of the heat sink. As the term suggests, the airflow is smooth, undisturbed and flows in constant contact with the surface of the heat sink material or coating, maximizing heat transfer. This flow pattern is very important to the operation of the unit, as anything that could possibly disturb the flow of air is taken into consideration in the design, layout and construction of every unit manufactured with this form of heat transfer.

SCR disc devices are clamped between the heat sink.

Fans vs. Fanless

Surface coatings often are misunderstood or simply overlooked with common industrial heat sinks. Many companies offer anodized coatings to improve the appearance and, in some cases, anodizing is even touted to improve performance. However, anodizing actually inhibits heat dissipation by placing an aluminum/copper oxide barrier between the heat sink material and the air. Black oil-based lacquers can improve heat transfer by placing a thin layer of thermally conductive paint on the exterior surfaces of the heat sink and improving surface emissivity.1

Forced convection is based on fans with a limited operation of typically 30,000 to 60,000 hr in a favorable environment. Longer life fans can be purchased at much higher cost. About 8,760 hr yields one year of service.

By comparison, some units have been in service for 10, 20 or 40 years with no fans in place. If a solid-state device is designed to operate reliably only in the presence of forced convection cooling, then you have just introduced the only moving part into a completely solid-state device. This is not the best option because fans notoriously fail. And, if the unit is designed to run safely only with a working fan, you could easily overheat a semiconductor when the fan eventually stops working. By contrast, natural convection relies solely on the natural laws of physics to ensure safe and proper cooling of semiconductor devices, with no moving parts.

Power diodes are made in wafer sizes from 3 mm diameter -- typical of the six power diodes in every automobile alternator -- to more than 75 mm diameter.

Unfortunately, many power control companies rely on limited-life, low-cost fans with lifetimes of three to four years to ensure continued operation of an SCR-controlled device. The same companies may also rely on computer models such as computational fluid dynamics (CFD) to determine whether a heat sink is adequate.

One power control systems manufacturer asked software companies specializing in electronic heat transfer CFD software to profile a simple 4" channel design heat sink to determine how accurate the software can be. Most programs were unable to come within 10 percent to 15 percent of actual temperatures of a 15 A heat sink. Some machine run times exceeded 10 minutes.

The results of these informal experiments suggest that to use a computer while designing an aluminum or copper heat sink would be very helpful to ensure you did not have to spend the money for an extruder die change. The study of laminar airflow, however, is more difficult because of the low natural convection velocities up to the transition from laminar to turbulent flow. (This normally occurs at a Reynolds number less than 1 million). Most CFD heat transfer programs assume forced airflow.

In addition, be careful when considering heat sinks sold in catalogs. All devices that become part of your power control solution should be tested to ensure that they meet the advertised ratings.

The well-known laminar flow principles of construction have been developed over a 50-year time span -- NACA laminar flow airfoils were first applied to the P-51 before WWII -- of SCR power control manufacturing. Effective heat sinks used in power control schemes have been developed right alongside the controls for minimum weight and maximum heat transfer. Don’t overlook their essential role in protecting and extending your equipment’s life.

Reference
1. General Electric. SCR Manual, 1979.

Links