Embedded resistor design guidelines: test vehicle development is continuing, and thermal guideline development has begun.
Developing design guidelines for embedded resistors is similar to developing guidelines for traces in printed circuits. The general method adapted from printed circuit pioneers is to characterize the temperature rise of the conductor, or in this case the temperature rise of the resistor (a poor conductor), in the dielectric material only, with no traces (except to power the resistors), no planes or components, and the board suspended in still air. This provides a worse-case condition that can be referred to as a baseline.
The process used to develop the guidelines begins by designing and manufacturing test vehicles, testing and collecting data, and then running computer simulations to confirm results and start building models of the parameters that drive the temperature rise. The baseline is then expanded to include some common configurations of internal copper to quantify the copper's heat dissipation effects. Testing, simulation and modeling will further evaluate the specific design features. Since the thermal behavior of the resistor is highly dependent on the total PCB composite, software tools will read in the exact PCB geometry. They will take into account every trace, via, component and connector, as well as mounting configuration and airflow, to completely evaluate the thermal performance of the resistor in the PCB composite.
The process to develop the thermal guidelines has begun. The test vehicle
design concept is complete and as of October 2004, the layout is days from completion. Next comes the manufacture of the test vehicles, probably in early 2005, and then the fun will start. This means running tests, collecting data and running computer simulations to quantify and verify the parameters that drive the resistor's temperature rise.
A primary issue when sizing a resistor for a circuit is its functional characteristic in the application, and this includes current. After numerous discussions with PCB designers from a significant number of organizations, we have concluded that designers are not given current as an input for resistor applications. They are, however, given the resistor characteristics in the parts list in the BOM. Typically the characteristics include package size, tolerance and wattage. The engineer selects these parts from a catalog, lf the engineer is thinking in terms of SMT parts and selects the power rating based on the notion that more is better, then the embedded version of this resistor will be over-designed from the start.
To effectively design a resistor to embed, the designer must be given the current as an input. We could have a similar discussion on resistor tolerance, but we'll save that for later. Embedded resistors should be evaluated with respect to their current carrying capacity, the same way a trace or copper plane is evaluated. Trace and copper plane heating as a function of current flow is well understood. Studies over the past five years have resulted in defining the parameters to be considered when evaluating the temperature rise of a trace as a function of current. The same methods used to evaluate trace heating are being used to evaluate resistor heating and develop guidelines for their thermal management. The power dissipation and power density that's calculated for resistors and compared with trace heating data is our starting point.
The temperature rise of an embedded resistor is similar to the temperature rise of a trace. The resistor dissipates power (watts) and the power is what drives the temperature rise. The current, which the PCB designer does not usually receive with his resistors selection, is required to calculate the power dissipation. Using the relationship P = I2R, the power dissipation can be calculated for a resistor when DC current is applied. For AC applications, the RMS current is calculated and used. Once the power dissipation for the resistor is determined, the temperature rise can be found.
The complexity and variation in PCBs from one board design to another (base materials, stackup or construction, copper mass, thickness and orientation) mean that a general design guideline will be extremely conservative in nature. A design guideline should include the defining parameters used in its creation and provide guidance to apply those parameters to specific applications. For example, consider a PCB 0.062" thick, 5" X 5", with a simple resistor pattern in its center, no more copper than the few terminating traces, suspended in still air. Apply current to the traces and the resistors will dissipate power and heat the board. Three modes of heat transfer occur: conduction in the dielectric material, convection to the surrounding air and radiation to the surrounding environment. This configuration may seem conservative or unrealistic for a PCB, but keep in mind that this defines the starting point. From this point, we can add copper planes, vias, components, mounting bolts, airflow, etc., and characterize the effects of the key elements of the PCB construction.
The example used above is not fictitious; it is the essence of the test vehicle we have created. See FIGURES 1 to 5. From this relatively simple test vehicle we will be able to begin the process of establishing quantitative values for power dissipation and power density of the resistor design as a function of the current and the board construction. Later, we will work our way into more complex constructions. One can only begin to imagine the margin that exists in a 20-layer board consisting of 2 oz. copper throughout the design. It also provokes thought toward designs that have low margin for maximum temperature rise of their semiconductor devices and where the resistors should reside.
[FIGURES 1-5 OMITTED]
FIGURES 6 and 7 illustrate some of the modeling we have done for a 100[OMEGA] resistor, 0.5 mm square, powered with 10 mA. You can see that the temperature rise is very small in the presence of the copper planes. This type of data will enable us to quantify resistor sizing. This modeling work is done with the Harvard Thermal's TASPCB tool. Our next article will review the modeling work that has led us to this particular test design. Following that, after data collection, we will provide confirmation of the models.
[FIGURES 6-7 OMITTED]
RICHARD SNOGREN is a member of the technical staff at Coretec Inc. (www.coretec-denver.com). MIKE JOUPPI is is president of Therman Man Inc. (www.thermalman.com), which has developed a trace heating database and tools to access it. He can be reached at firstname.lastname@example.org.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Getting Embedded|
|Publication:||Printed Circuit Design & Manufacture|
|Date:||Dec 1, 2004|
|Previous Article:||Hotel and travel.|
|Next Article:||A matter of timing: gone are the days when chip delays consumed the bulk of overall timing budgets. Now, board-level timing margins are more...|