For the PDF version of this article, click here.
Industry trends continue to push power supply technologies to provide higher power density supplies and components with ever-smaller footprints and higher efficiencies. These increasing power and current densities have driven technology innovation in the component and power supply areas, but very little innovation has occurred in the power interconnect field.
This lag in technology within the interconnect can result in designs where the footprint of the power supply is significantly affected by the connector interface, rather than the internal components of the supply. This article describes the fundamental issues faced in the design of power connectors and presents a new class of power interconnects that addresses the limits of existing technology.
The performance of a power connector is dependent on both the bulk resistance (electrical resistance through the metal frame) of the contact and the separable resistance (resistance across the actual separable interface). Most power connector designs available in the marketplace are based on some form of a spring beam. The deflection of the spring beam during the insertion of a mating pin generates a normal force between the contact surface on the spring beam and the mating pin surface. This normal force produces and maintains the electrical path across the separable interface of the connector.
The deflected spring-beam approach is the basic layout in traditional fork and blade contacts, louvered contacts, as well as helical beam and most compression-mount contacts. In the design of spring-beam based connectors, the engineer has to design the spring beam to satisfy both the mechanical (normal force) and the electrical (conductive path) design parameters with a single component. By necessity, this results in a compromised design where neither the mechanical nor the electrical properties are optimized.
For optimal electrical performance, the material in the electrical path should be of the highest electrical conductivity possible with the largest cross-section possible and shortest conductive path. From a mechanical design approach, the material in the spring beam should allow for the largest deflection range possible while generating consistent normal forces.
While designing a beam with small deflections and a high normal force is possible, it is typically not desirable due to tolerance stack-up errors in the finished mated connector. Stiff, low-deflection beams can be designed to produce a desired normal force with little deflection, but if the amount of deflection is too high by a small amount, this can result in excessive normal forces, which can damage plated surfaces or result in excessive insertion forces.
Also, if the deflection of a beam is too high, plastic deformation occurs instead of elastic deformation, and the contact beam takes on a permanent set. This can result in the total loss of contact during operation. On the other hand, if the deflection of a stiff beam is too small, the normal force can be too light to produce a reliable electrical interconnect.
The ideal beam design involves an almost-constant normal force over a wide range of deflection, but the mechanics to obtain these characteristics involve the design of long beam sections with small cross-sections. This approach does not yield small compact connectors with short conductive paths. The design process typically yields a short- to medium-length beam with a small cross-section, and the final mechanical properties become highly dependent on the material selection.
Optimal electrical performance is obtained through the use of a high-conductivity material such as pure copper, but the mechanical properties of the spring beam require a spring alloy. Pure copper cannot be used in spring beams due to the very low yield stress of the material and to the stress-relaxation characteristics of the material over time. Even when loaded below the elastic limit of the material, a spring beam made from pure copper will suffer from stress relaxation (even at room temperature), and the normal force will decrease as a function of time.
To solve these material issues, power contacts are typically formed from a spring alloy such as phosphor bronze or a beryllium copper alloy. These materials meet the mechanical requirements but suffer from low electrical conductivity. Specific values depend on the specific alloy and heat treatment, but conductivity is usually in the range of 15% to 50% of pure copper.
Some connectors do manage to use pure copper in the electrical path. These systems typically use a secondary spring element mounted directly under one of the contact surfaces to force two pieces of high-conductivity copper against each other. While this solves the issue of optimizing the conductive path, the limiting factor in the performance becomes the electrical resistance at the separable interface.
The physics of the separable electrical interface has been studied extensively with numerous publications and papers published on various aspects of the field. Much of the basic understanding of separable interfaces is credited to Holm. The electrical performance across a separable interface is strongly affected by the number of microscopic contact spots (“a-spots”) between the surfaces. Due to surface roughness and irregularities, two flat surfaces when pressed together will only contact in a few spots. It takes a large force to increase the number of actual contact spots between two surfaces, and even if this level of force can be generated in the connector, the forces can be so high that the connection can be next to impossible to make and break, and the plating on the surfaces suffers significant damage in only a few cycles.
Plating is required on most connector surfaces to avoid the build-up of oxides at the separable interface, which can lead to increased resistance and even open contacts. As a result of the contact mechanics between the mating surfaces, most pure copper connectors have only a few contact points for each circuit. Multiple contact points provide the advantage of multiple parallel paths for the current flow, and multiple parallel paths result in lower resistances and higher current-carrying capability.
So here is the fundamental problem with the design of most power connectors: spring-beam contacts use low-conductivity materials and finite-length conductive paths, but may be able to generate up to 20 to 40 contact points per connection. Pure copper contacts are usually large and limited to only a few contact points at the separable interface. What is needed are designs that optimize both the bulk-resistive path with high conductivity materials, as well as the separable interface path with multiple contact points per circuit.
A New Connector Technology
The goal in the design of the Tribotek power technology was to develop a separable interface that improves and optimizes both the bulk resistance and the separable interface resistance of the connector. Rather than considering various spring-beam designs, the power contact technology was developed around a tensioned-weave approach.
At a fundamental level, think of taking two metal rods of the same size and laying them side by side. Wrap a piece of string once around the outside of both rods. When you pull on the string and apply tension, the rods are pulled into contact with each other and a normal force is generated across a separable interface.
Now try this. Make one of the rods much smaller than the other and again wrap the two rods with a tensioned string. Despite the size difference, the rods are still forced against each other and a normal force is generated at the separable interface. Take the small rod and replicate it many times around the circumference of the larger rod. You now have the basis for a tensioned-weave type connector. Each of the small rods can be formed from drawn wire and, since the normal force between each wire and the larger rod is generated by the tensioned fiber, the wires can be formed from a high-conductivity material, such as pure copper, rather than a spring alloy.
Using multiple wires around a single rod and a single tensioned string (fiber) will produce a single contact point between each wire and the central rod (pin). Each contact point is independent from those to either side, so each contact point generates its own set of microscopic contacts between the wire and the pin surface. For example, if a 1/8-in. diameter pin is used with small rods made from 0.005-in. diameter wire, then upward of 90 individual wires can be placed around the circumference of the pin. With a single tensioned fiber, this will generate 90 contact points. The actual normal force between a wire and the mating pin surface is a function of the system geometry (pin radius, wire spacing, wire size) and the tension in the fiber. Fig. 1 shows the geometry of the mating surface interface.
It is also possible to extend the weave structure along the axis of the pin by adding additional tensioned fibers along the length of the pin. By doing this, a 2-D contacting structure can be generated along and around the pin surface. The pitch in the circumferential direction is set by the number of wires and can be as small as the diameter of the wires. The pitch in the axial direction is set by the spacing between successive bands of tensioned fiber. Fig. 2 shows a magnified view of a weave structure built using 0.005-in. diameter wire with tension bands laid on a 0.020-in. pitch.
The wire between successive bands of tensioned fibers is bent into a loop shape. The loop shape serves two purposes: the loop produces a true weave structure with the tensioned fibers that maintains the weave structure shape, even when a mating pin is not inserted into the contact, and the bend helps to make successive points independent from a mechanical loading point of view, which ensures that each wire contacts the pin surface at every tensioned fiber interface. Fig. 3 shows a view of a weave structure with four bands of tensioned fiber.
To complete the contact, the pin is routed to one side and the wires are routed to the other side and bundled into a common termination. The pin and wire bundles can then be terminated to whatever the specific application requires (wire, busbar, pc board). The fibers need to be tensioned by some external spring element, but now this spring is no longer in the electrical path and the design and material can be selected based solely on the mechanical requirements for the given application.
Fig. 4 shows a round socket designed for a 0.125-in. diameter pin. The socket makes 180 contact points with the mating pin and the total resistance through the mated connector (bulk and separable interface) is below 150 µΩ. The “U” spring that is visible in the picture serves to tension the bands of fiber.
As in all classes of connectors, there are many variations in the design of application-specific connectors, but the basic approach of high-conductivity materials and multiple points of contact remain constant. Pin size and shape, wire size and shape, interface structure, tensioned-fiber structure, fiber-tensioning mechanism and mounting housing can all be modified for specific applications and environments.
The design of the secondary structures for tensioning the fibers and providing the application-specific termination are important aspects of the connector design, but in each case, the fundamental issue of reducing the bulk and separable resistance paths can be addressed using tensioned-woven structures. Other solutions also may be possible as the limitations of existing power connector technologies have been recognized by designers for a number of years. By solving the specific issues around the bulk and separable interface resistances, power connectors can be developed that allow for higher current densities and lower resistances.
There are three main situations where better electrical-performance power contacts can result in more flexibility in system design. For designs where the current/power requirements don't increase but where everything needs to be packaged in a smaller space, an optimized contact can yield a significantly smaller profile. This is useful if you have been limited in the allowed footprint for the power connector or if you need more space for airflow/cooling or data pins.
Fig. 5 shows a mixed data/power connector that uses a tensioned-weave contact for power distribution. The height of the connector above the board has been kept below 7.5 mm with four 30-A power contacts and 24 data pins. Traditionally, a 30-A connector results in a connector height around 15 mm. The reduced height with a tensioned-weave contact still allows for significant airflow, even with 1U height restrictions.
Another case where improved power contact technology can enable a design is in applications where the power levels are increasing and total available space remains the same. With higher power, the thermal dissipation through the connector becomes more important since the losses are proportional to I2R. Doubling the current without reducing the resistance in the connector results in a four-fold increase in losses across the connector, which become part of the thermal load that has to be managed in the system. Being able to increase the current through the same size connector footprint can also enable system upgrades without the need for a complete redesign around the power distribution network.
Fig. 6 shows a busbar-to-busbar connector with four #1/0 contacts capable of carrying 325 A per pin with 60 µΩ of resistance per pin. This represents a 20% to 30% increase in current density over existing connecter technology.
The final case where improved power connector performance becomes important is for low-voltage applications where there is a limited voltage drop budget between the source and load. As the current in a system increases, it is harder and harder to limit the losses through the system, and the connector can contribute significantly to this problem. For a 5-V system, maintaining a 0.5-V tolerance is easier if the resistance across the connector can be measured in the micro-ohm range rather than the milli-ohm range. This can be true at any current level. The table shows the current capabilities and resistances for a full range of contacts designed using the tensioned-weave approach.
The selection of a power connector is based on many issues. This can include current capacity, size, environmental ratings, cost and availability. The manufacturing process for tensioned-weave type connectors is specialized and completely different from most other connector types. This results in a small cost premium compared to high-volume commodity products, but this premium is comparable to that of other high-current-density power connectors. The cost for the tensioned-weave fabrication machinery is merely substituted for the cost of progressive die and high-speed stamping processes required for conventional contacts.
The termination of the contacts for each specific application is the same as for other types of interconnects, so the packaging, assembly and application are exactly the same from the customer's point of view. In some cases, the current capacity and performance of a system is limited by components in the system other than the interconnect. In this case, the end user will receive very little electrical performance benefit from selecting a high-performance power interconnect over a commodity product.
Mechanical issues other than size also are factors in the selection of a power connector. These include the ability to handle misalignment and meet blind-mate requirements. Since the tensioned-weave contact interface is effectively the end of a bundle of wires with a spring attached to this bundle, the separable interface of the interconnect can “float” with respect to the termination based. This produces a level of built-in gatherability in the interface that can make blind mating between two connector halves easier with fewer restrictions on alignment. Also, with large numbers of contact points, the interface becomes more resistant to degradation in the presence of severe environmental contaminants (i.e. dirt).
|Pin Diameter (in.)||Mil-Std Size||Metric Size||Resistance (µΩ)||Current (A)|
Similarly, with the round pin design, no mode of vibration can generate complete separation of all contact points at a single time. The challenge for severe shock and vibration conditions is in the design of the housing for proper support and anchoring to the mating interfaces.
The tensioned-weave interface is very versatile and can be adapted to provide better performance in almost all applications. For example, it can be adapted for high-temperature operation by designing the tensioning spring from materials such as stainless steel. Or for situations with multiple small pins with very tight array spacing, a single tensioning structure can be shared over multiple contacts to reduce spacing.
The tensioned-weave structure does not provide significant performance benefits in situations where other components constrain the design. But where the connector itself limits the design, the tensioned-weave structure can enhance performance dramatically.
Davis, J.R. (ed.). ASM Specialty Handbook: Copper and Copper Alloys. ASM International, Materials Park, Ohio, 2001.
Holm, R. Electric Contacts, Theory and Applications. Springer-Verlag, New York, N.Y., 1967.
Moran, J., Sweetland M., and Suh, N.P. Low Friction and Wear on Non-Lubricated Connector Contact Surfaces. Proc. 50th IEEE Holm Conference on Electrical Contacts CMPT-IEEE, Seattle, WA, Sept. 20-23, 2004.