Simulation Helps Solve Difficult
Thermal Challenge in Tower Mounted Amplifier

   < Back to Articles

A tower mounted amplifier (TMA) for a cellular base station presented a challenging thermal management problem. It contains three power amplifier chips each dissipating about 160 watts inside the enclosure. The door of the enclosure itself is finned and serves as a natural convection heat sink. Thermal simulation showed that the heat from the 1 by 1.5 inch amplifier chips did not spread out sufficiently to utilize the full extent of the heat sink, causing the chips to overheat. The problem was addressed by adding a vapor chamber to each chip to reduce the spreading resistance and utilize the full extent of the heatsink. The simulation showed this approach reduced the temperature at the base of the heat sink to acceptable levels.

Thermal challenge in cellular base station

The TMA heatsink is roughly inches 31 inches high, 11 inches wide and 0.300 inches thick. The original design used 2 inch high fins spaced at roughly 3 fins per inch. The fins run from top to bottom along the door so that air currents produced by the heat generated by the amplifier run parallel to the fins in order to maximize natural convection. Each of the power amplifier chips is 1 inch by 1.5 inches, a very high heat flux density. The three power amplifier chips are aligned in the vertical axis.

The company building the TMA came to Thermacore, a company that provides thermal management solutions, because they were concerned about overheating. Matt Connors said that in the past the company would have used a spreadsheet to evaluate the design. “The presence of high heat flux density power sources in this application means that spreading resistance is a critical factor in calculating cooling performance,” Connors said.

“Thermacore has developed a number of theoretical equations to help predict spreading in the base of a heat sink”, Connors continued. “We embedded these formulas into a spreadsheet and have used it as a design tool. But its accuracy is limited by the fact that it cannot take the details of the spatial geometry into account. The challenge is especially great when heat pipes or vapor chambers are used. These devices have multidimensional high conductivity surfaces that are very hard to calculate by hand. Quantifying the interactions of numerous geometries of vapor chambers or heat pipes through varying interfaces and thickness is extremely difficult to do without FloTHERM.”

Simulation Helps Diagnose Thermal Problems

“For the last several years we have used thermal simulation software to evaluate nearly every thermal management challenge provided by our customers,” Connors said. “The advantage of thermal simulation is that it generates a 3D model of the complete geometry and calculates airflow and heat transfer throughout the entire problem domain. This method requires far fewer assumptions and provides much more accurate results than we were ever able to obtain with a spreadsheet. Accuracy is critical in thermal design because our customers want to know what will and won’t work as quickly as possible so they can get their product into manufacturing and start generating revenues.”

“When we first decided to use thermal simulation we polled our customers and asked them which software package they would like to have us use,” Connors said. “They said that they preferred FloTHERM over the other leading thermal simulation packages. They use FloTHERM themselves so they have confidence in its predictions and can easily incorporate our models into their full system models. Since we began using FloTHERM we have been very impressed with its ability to accurately simulate thermal management challenges. When we finally get to the stage of building a prototype we find that the measured temperatures are nearly always within 10% of the values predicted by the simulation. On average, our simulation results predict real-world measurements within 5%. Thermal simulation also takes only about 1/5 the time of manual calculations with a spreadsheet.”

Modeling the TMA

Thermal simulation of standard aluminum heat sink shows natural convection thermal resistance of 0.193 °C/watt

Connors modeled the TMA in FloTHERM focusing on the three power amplifier chips and the heat sink. The simulation results showed three hot spots in the base of the heat sink corresponding to the three chips. The result is that most of the heat from the chips is concentrated on the innermost three or four fins. Very little heat is transferred to the other fins so the effective size of the heatsink is much smaller than its actual size. Air moving upward through the fins is heated by each power amplifier chip so that by the time it reaches the highest chip it is quite hot which further reduces the effectiveness of the heat sink. The net result is that temperatures at the base of the heat sink are higher than allowable for this application with the initial design.

Connors next evaluated the potential for solving the problem by using a vapor chamber which provides much lower spreading resistance than a solid heat sink. Vapor chambers are plate-shaped heat pipes that can be used as the base of the heat sink. They transfer heat more efficiency over a plane, minimizing spreading resistance of the heat source. A vapor chamber consists of a hollow copper shell lined that has an internal wick structure to transport the fluid to the heat source and a vapor area for the transport of the vapor from the heat source.

Heat is absorbed in the evaporator region by vaporizing the working fluid. The vapor transports heat to condenser region where the vapor condenses, releasing heat to the cooling medium such as air. The condensed working fluid is returned to the vapor chamber by gravity or by capillary action if working against gravity. The entire chamber operates at water’s saturation pressure so that the working fluid vaporizes at room temperature. The process allows the vapor chamber to efficiently transport heat from the heat source to the surface of the condenser. This effective spreading allows all the fins on the vapor chamber to work as efficiently as the ones that are over the center of the heat source.

Adding a Vapor Chamber to the Model

The spreadsheet approach assigns a blanket number for thermal conductivity such as 5000 w/mK to the vapor chamber. The problem with this approach is that the vapor space has a very high thermal conductivity while the wick has a much lower thermal conductivity. It is important to break down these conductivities in the thermal model to ensure the accuracy of the predictions. Connors modified the Flotherm model to incorporate three vapor chambers, one for each power amplifier chip. He modeled each of the individual components of the vapor chamber and assigned the correct thermal resistance value to each. The top and bottom wall of the vapor chamber are copper with a conductivity of 380 watts/mK.

The thickness of the wall is dependent on the necessary strength of the overall chamber and the amount of spreading needed within the copper walls prior to reaching the wick. The wick conductivity ranges from 30 to 60 watts/mK and is dependent on the flux of the heat source. The thickness of the wick is dependent on how much power is being transferred and the orientation with respect to gravity. The vapor space conductivity can range from 5000 to 50,000 watts/mK depending mostly on the length of the vapor chamber. Normally there is two or three degrees T in the vapor portion of the vapor chamber.

Results Show Good Thermal Performance

Thermal simulation of vapor chamber heat sink shows natural convection thermal resistance of 120 °C/watt

The simulation results showed that the vapor chamber efficiently spread the heat across the full width of the heatsink so that each of the fins was fully utilized. This reduced the temperatures at the base of the heat sink to acceptable levels. Thermacore built a physical model consisting of a heat source equivalent to a single RF amplifier chip and the vapor chamber and the third of the heat sink that cools this chip. As usual, the physical testing results matched the simulation results within 5%. The improved cooling enables more powerful amplifiers to be placed in compact enclosures. “This application demonstrates how simulation provides a competitive advantage to thermal solutions companies that are experienced in its use,” Connors concluded. “Simulation is much more accurate than conventional formula driven thermal engineering methods and also takes less time. So we can investigate more design alternatives with a very high level of confidence in the accuracy of the results. Thermacore’s strength is not in selling metal by the pound but rather in the design expertise that we deliver to our customers. Our expertise in thermal simulation plays a key role in meeting and exceeding our customers’ expectations.”


Request More Information


   < Back to Articles