Pin Design of High Current DC/DC Converter

As the supply voltage of digital circuits in electronic systems drops to the 1.0 - 1.5V range, and the power consumption on the load board increases, DC/DC converter modules need to provide very large output currents. Converter manufacturers are increasingly catering to the industry's demand for higher power and smaller packages. For example, five years ago, a half-brick converter could only provide a maximum of 30A of current, but now a half-brick converter can provide up to 100A of current. Also five years ago, a quarter-brick converter only provided 15A, but now it can provide 60A of current. Although this is good news for Chinese communication system design engineers who are seeking greater power density, it also raises the issue of how to solve the problem of how to solve the increase in device temperature due to higher currents. To handle such large currents, the number of pins for output power needs to double.

Figure 1. Voltage across the V+ power plane. The converter has two pins in parallel at each terminal. (Nominal DC voltage not shown)

Figure 2. Voltage across the V+ power plane. The converter has two pins per lead, placed on either side of the converter. (Nominal DC converter not shown)

There is now a consensus among manufacturers and users of DC/DC brick converters that additional power pins are needed to more evenly transfer and distribute heat between the converter module and the load board. But the question remains: where should these additional pins be placed? Ideally, the location chosen should be the most beneficial to the user and provide good performance.

In addition, the placement of the additional pins should be in line with the standards of the converter industry to avoid confusion, avoid exclusive supply, and facilitate user selection. However, contrary to expectations, some DC/DC converter manufacturers have launched some products with incompatible additional pin placements, and there is a lack of consensus in the industry on the standard pin design of high-current converters. Therefore, system designers need to decide for themselves which pin position will become the new standard. If this situation does not change, equipment manufacturers will have to exclusively provide their high-current modules or accept expensive redesigns. It is time to thoroughly investigate this technical issue and develop a new standard pin design solution for high-current DC/DC converters. The purpose of this article is to provide Chinese designers with some technical references for solving this problem.

First, realize that the need to double the number of power pins is not due to pin resistance. An 80mil diameter copper pin has a resistance of about 20 to 100 mils. When this pin carries 100A of current, it dissipates only 0.2W of power. Since this 100A flows out of the V+ pin and back into the return (or ground) pin, a total of 0.4W is lost. By doubling the number of power pins, this is reduced to 0.2W. In other words, a 1.2Vout, 100A converter with an efficiency of 83% will lose about 25W. Therefore, doubling the number of power pins reduces the losses due to pin resistance by less than 1% of the total losses.

Figure 3, high current half-brick pin design, with pins on the same lead-out end placed side by side.

Figure 4: SynQor's high-current half-turn pin design, with the same lead-out pins distributed on both sides of the converter.

The primary reason to double the number of power pins is to reduce the losses that occur on the load board as the output current is transferred away from the pins on the power plane of the load board. SynQor has studied this problem extensively and, through theoretical analysis and carefully controlled laboratory testing, has developed models that help determine the ideal location of additional pins for high current DC/DC converters. To better understand the above analysis: Consider a standard load board that is 12 inches square. Assume that there are four evenly distributed loads around the board, each drawing 25A of current, and also imagine that a 100A half-brick converter is mounted on one side of the board. Further assume that the power plane connecting the converter to these loads is made of 1 ounce copper and has a resistance of 1mΩ per square (taking into account the many vias blocking it).

We first look at drawing 100A from a half-brick converter using only one output pin per pad. We then compare the results to two approaches that use double the output pins per pad.

When there is only one power pin at each end of the converter, we can calculate the voltage on the power plane very easily and accurately. In the simulations performed by SynQor, an 80 mV voltage drop was observed about 6 inches from the pin. This large voltage drop occurs because the current must diverge from a small point (the pin) before the full width of the power plane can be fully utilized. The impedance of this "diverging" area is very large. At 100A, 80 mV dissipates 8W. This number doubles when the losses caused by the current returning to the return pin of the converter are considered. 16W of losses and a voltage drop of 160mV (13.3% of 1.2V) are too much, and once again, we need to double the number of output pins for high current DC/DC converters.

Consider the voltage profile of the converter in Figure 1: In this scheme, a second power pin is added to the V+ terminal. In the test pattern, the nominal DC voltage (e.g., 1.2V) at this pin is removed from the scale, allowing us to focus entirely on the voltage drop from the pin to the load. The scale of Figure 1 also prevents us from separating the second pin from the first. Note that the simulation includes the heat dissipation around the pins, although they are not well represented in this diagram. For this converter, the additional output pins are placed within 0.2 inches of the original pins (aligned). This pin arrangement (see Figure 3) has been adopted by some major DC/DC converter manufacturers. The adjacent pin location may be convenient for converter layout, but it does not help much in solving user problems. The voltage drop 6 inches from the converter is about 70mV, which is 10mV less than when only one pin is used. The overall power savings is 2W, and the power loss

It is 16W.

The reason for this improvement can be found by measuring how the current spreads away from the pin. As shown in the simulation, the current must travel several inches to fully spread out to take advantage of the width of the power plane. But because the two pins are placed only 0.2 inches apart, their currents quickly add up, just like there was only one pin. Therefore, the effect of having two pins is limited to the vicinity of the pin, which is only a small fraction of the total spreading resistance.

However, let’s look at what happens when the additional pin is placed on the other side of the half-brick converter. Figure 4 illustrates a SynQor high current half-brick pinout where the additional pin is placed 0.2 inches outside the opposite polarity pins and aligned. With the polarity reversed, each pair of power pins is 1.6 inches apart. As the voltage profile in Figure 2 shows, the voltage drop at 6 inches from the converter is approximately 40 mV lower than with only one pin. In this case, the overall power savings is 8W and the power loss is 16W. The difference between Figures 1 and 2 is obvious because in the first pinout design, the two output pins are superficially indistinguishable, while in the second design, they are clearly separated.

The most significant improvement in the second design is that the distance between the two pins is 1.6 inches instead of 0.2 inches. This allows the current from each pin to spread out before it overlaps with the other current. Since there are two parallel channels, the effective spreading resistance is almost split.

Obviously, from the user's perspective, the location of the additional pins in Figure 4 is better than that in Figure 3.

In addition to the advantages described above, SynQor's design significantly reduces the resistance spread on the customer's load board, reducing the parasitic inductance of the connection by 90%. This improvement enables the module to have better transient response, reduce output ripple, and improve current sharing capabilities.

Further, the placement of the additional pins in Figure 4 not only reduces the power loss in the load board. Some of the heat from the converter can also flow down the pins and into the load board. How much heat can flow in this path depends on how hot the load board is due to other heat sources. In our half-brick example, we saw that the power loss in the power plane was reduced to 8W by choosing a better pin location. This reduction allows the load board to be cooler in the converter area, allowing more heat to flow from the converter to the pins. This reduces the converter temperature and increases the converter reliability.

Since different manufacturers have different pinouts, one might think that designing for interoperability between different pinouts is impossible. Interestingly, different pinouts can be substituted for each other with only a few minor modifications, just a few more holes in the load board. However, the issue is not just about non-exclusive supply, but performance. Designs with opposite polarity pinouts result in lower voltage drop, inductance, and power losses, greatly improving performance. Other converters would be able to achieve similar performance gains if they used the opposite polarity pinout set, even though they would place the additional pins where they currently are. This is why the choice of industry standards must be very wise, considering not only the location of the pins, but also the pin polarity and the resulting performance changes.

The same problem applies to where to place additional pins on high-current quarter-brick converters. For example, SynQor places additional pins 0.15" outside the original pins on its 60A quarter-brick. More importantly, like the half-brick layout, this design places opposite-polarity terminals side by side. Although for this smaller converter, the distance between two pins on a given terminal is now only 0.75" instead of 1.6", this design is superior to designs where the distance between two pins on a given terminal is only 0.15".

This analysis shows that the pinout of high current converters should not be ignored but taken seriously. The performance results are real, and design engineers are not sure what the industry standard will be. Now is the time to determine a sensible standard for the pinout of high current converter bricks. This requirement will become more urgent as more quarter brick converters exceed 60A and also require double the pinout design. Doubling the opposite polarity settings and maximizing the distance between the pins can provide the same technical and commercial advantages to the quarter brick as the half brick. However, some DC/DC converters still maintain a dual positive and dual negative configuration. Since there is no standardization body for DC/DC converters, any decision on the pinout standardization of high current DC/DC converters will be determined by the market. In other words, everyone may act in the way of the market leader. If so, let us hope that OEMs and ODMs can choose wisely.