While the path to miniaturization in purely digital applications is well-developed, in analog and power applications the approach remains possible, though far less obvious.
In this technical article, Toshiba will discuss the challenges of miniaturizing mixed-signal devices and examine some of the approaches used.
It has long been expected that each generation of technology will be smaller or more powerful than its predecessor, or both. There are many reasons for this: reducing the weight of electric vehicles to increase their range, miniaturizing industrial equipment such as robots, and making our personal devices more attractive and convenient.
The Challenge of Analog and Power:
Since the 1960s, when Moore's Law was coined, the basic integrated circuit technology underpinning electronic designs has been expected to show significant growth in complexity, accompanied by reductions in size and cost every two years. This has largely held true, particularly in the world of purely digital devices, such as logic integrated circuits and MCUs (microcontrollers). In this case, size reduction primarily refers to reducing the size of features through process evolution and is largely driven by electronic design automation (EDA).
However, in the world of analog and power devices, things are somewhat different. Here, the focus of size reduction is more on the ability to combine design blocks and on improving the design to make this happen. Some components, such as power MOSFETs, require a certain physical size to offer low ohmic (static) losses and handle the higher currents typically required by modern applications. Unless the design can be improved to reduce energy consumption, the laws of physics dictate the size of these types of components.
The key to integrating analog and power designs with digital logic is the ability to merge different manufacturing processes: a common foundation is needed to achieve the highest levels of integration.
Thinking Differently:
To the casual observer, miniaturization simply means squeezing the semiconductor package into a smaller size, and while this can offer some advantages, there is often a better approach. Ultimately, the system designer seeks to reduce the size and weight of the final design and needs semiconductor devices that enable this, depending on all the project's circumstances.
What is often called "significant downsizing" reflects the fact that a change in one semiconductor device can enable size reductions elsewhere that are many times greater than the semiconductor itself. An example of this is the shift from silicon (Si) MOSFETs to widebandgap (WBG) materials such as silicon carbide (SiC) or gallium nitride (GaN). Thanks to the reduction in gate charge (Qg), these devices are able to operate more efficiently at high frequencies. As a consequence of the higher frequencies, the size of magnetic components is reduced, as is the size of thermal management components (such as heat sinks), due to the increased efficiency. Other examples of revised semiconductor designs that provide design advantages are advanced motor controllers that can operate without the need for external components, such as shunt resistors.
Areas to Consider:
In addition to the connectivity between the various components of a mixed-signal semiconductor, many other considerations must be taken into account. This inevitably requires a significant amount of trade-offs and, for this reason, demands highly experienced designers.
The process begins with a thorough understanding of the application(s) in which the device will be used. From there, the interfaces between each of the blocks are studied to ensure that the drive force (the energy supplied from one block to another) is adequate but not over-specified. In some cases, it may be advantageous to transfer certain functions from the analog domain to the digital domain, but in others, this may not be suitable for the application due to aspects of accuracy, reliability, and performance.
One of the main considerations is the path of conductivity losses, both electrical and thermal. Heat generation is one of the factors that determine the size of semiconductors, so minimizing losses is important. For example, designers must consider the trade-offs of connecting the die to the package with a copper wire or clip, not only from an electrical and thermal perspective, but also in terms of manufacturing, performance, and reliability. To minimize thermal conductivity losses, an advanced package that provides cooling from both sides may be necessary, although this can have economic implications.
In harsh applications and environments, such as those found in under-the-hood automotive applications, materials must be carefully considered. Due to extreme (and rapid) temperature fluctuations, the substrate and other components can expand and contract at different rates (differential coefficient of thermal expansion - CTH), which will put stress on the junctions, leading to reliability issues. Materials, such as substrates and adhesives, must be carefully chosen to suit the application environment.
In almost all cases, "manageable" is a better goal than "smaller"; for example, the automotive industry has a particular need for automated optical inspection (AOI) that requires visible tracks, which ultimately define how small an encapsulation can be.
Design Example: Motor Controller.
Motor controllers are relatively complex devices and, therefore, a good example of the integration and miniaturization process. Typically, these devices consist of control logic, gate drivers, and MOSFETs, which can be integrated in various combinations. Aside from what is physically possible, the designer's priority is determining what makes sense to integrate.
Figure 1: Multiple integration approaches are possible for MCDs
It is clear that a fully integrated solution could contain all the main components. However, while this is attractive from a cleanliness standpoint, it limits flexibility. An alternative approach integrates only the control logic and gate drivers, leaving the MOSFETs as discrete external devices. This allows the MOSFETs to be sized for different applications, offering greater appeal (and higher manufacturing volumes, which will impact cost).
Incorporating new design techniques into the control logic can benefit the design and facilitate the integration and miniaturization process. For example, intelligent phase control puts the current and voltage in the same phase, thereby reducing the power factor and energy consumption, which increases efficiency and lowers motor operating costs.
Figure 2: Innovative design techniques favor the integration and miniaturization process
In a conventional motor drive system, a pair of (rather large) external current-sensing resistors are required. However, Toshiba's innovative Advanced Current Sensing System (ACDS) replaces these external resistors with a smart, highly accurate integrated measurement circuit, simplifying the printed circuit board layout and saving space in the final design.
Additional techniques, such as Toshiba's unique Advanced Dynamic Mixed Decay (ADMD), can be incorporated. ADMD provides highly accurate tracking of the input current, enabling very effective motor control at high RPMs. Furthermore, sensorless motor control techniques eliminate the need for external Hall-effect sensors, further reducing size, bill of materials cost, and complexity.
Business Considerations:
While many approaches may be technically feasible, business considerations must be taken into account when integrating and miniaturizing integrated circuits. Supply professionals are increasingly driven by multi-source policies that require footprint-compatible devices to be available from multiple manufacturers. While this is an important issue in itself, it is leading many semiconductor companies to maintain industry-standard footprints and pin assignments for their devices. Even with this requirement, there is still an opportunity to differentiate through good design and innovative techniques, such as those described above.
Abstract:
Miniaturizing mixed-signal integrated circuits containing analog and/or power elements is very different from miniaturizing a digital integrated circuit. Because of the mixed processes involved, a different approach is necessary, and often the size of the integrated circuit itself may not be reduced, but significant size benefits can be achieved elsewhere in the final product.
The complex trade-offs and decisions regarding what to integrate (and how), as well as commercial considerations, demand highly experienced engineers. Toshiba is uniquely positioned in this regard, thanks to its expertise in control logic, MCUs, simple logic, interconnects, bridges, multi-gate driver designs (based on its optocouplers), and power MOSFETs.
Author: Armin Derpmanns, General Manager Semiconductor Marketing, Toshiba Electronics Europe GmbH
