Implantable, digestible, interactive, interoperable, and supporting the Internet, the unique needs of these medical devices now and in the future require appropriate IC process technology and packaging. This article will compare the bipolar and CMOS processes used in medical semiconductor devices, and will explain some of the packaging issues that need to be paid attention to.
Developers of medical applications must make trade-offs between power consumption, noise, linearity, reliability, and cost. The process and design architecture need to be carefully selected according to these requirements.
This article will compare bipolar devices with CMOS devices to help users determine the applicability of each device. This article will take high-performance ultrasonic equipment as an example to discuss how to balance noise, power consumption, chip footprint, and integration.
Power consumption is very important in many battery-powered applications. In this type of application, CMOS technology is an excellent choice. However, the balance between leakage and performance is also critical and determines the choice of technology. In addition, in such applications, mixed-signal integration is also an important requirement.
Efficient use of some packaging technologies can meet the need to implement a large number of functions in a single integrated circuit, such as when supporting dense digital functions and requiring low noise at the same time. Such contradictory requirements can sometimes be easily met with multi-chip modules.
This article will also discuss the future development trend of medical equipment, including direct measurement of biological signals and self-powered equipment. These trends will promote the improvement of existing process technology to meet the energy harvesting characteristics and other non-standard sensor functions.
First, take the ultrasonic device as an example to discuss the simulation performance requirements. Through this example, this article will introduce how to make trade-offs between performance, power consumption, size and integration, and test the applicability of bipolar and CMOS process technology. Figure 1 is a system block diagram of a typical ultrasonic machine, showing two parts of transmission and reception. These two parts are responsible for driving the sensor and the digital processing part (not shown) to form a complete ultrasonic device.
Figure 1: Block diagram of ultrasound system
The issues that need to be considered when designing this type of receiving module include input noise, linearity, gain, and power consumption. The number of receiving channels for a given package size determines the degree of integration. The signal received from the sensor can support an amplitude change of more than 100dB. Therefore, the input noise on the low-level signal (about 10uV) and the linearity of the large input signal (about 1V) are very important performance parameters. To adapt to this large dynamic range, the channel gain can be adjusted through a voltage controlled attenuator (VCA) and a programmable gain amplifier (PGA). Figure 3 shows how the overall gain through the device varies with the voltage on the VCA under several PGA settings.
Figure 2: Detailed block diagram of the part that performs the receiving function in Figure 1
Figure 3: The graph of the gain of the receiving module changes with the voltage control
The following will compare the performance of bipolar amplifiers and CMOS amplifiers. Both bipolar devices and CMOS devices can be used to design open-loop amplifier modules that support 4mA bias current to achieve 20dB gain. Here, the BiCMOS process (inside TI) is the target process technology.
Table 1 compares the size of bipolar devices and CMOS devices used in amplifiers. The larger size of CMOS devices and the accompanying input capacitance severely limit the input bandwidth of the amplifier. In this example, a bipolar amplifier can be used to achieve low noise at low bias currents. However, the use of bipolar devices may have base current noise, which is negligible in CMOS devices. The magnitude of the base current noise depends on the impedance of the sensor and the specific implementation of the system.
Table 1: Size comparison of bipolar devices and CMOS devices
Mixed signal and low power applications
It has been observed that in certain medical applications, the analog performance of bipolar devices is better than CMOS devices. However, some applications need to process mixed signals, which require both analog and digital processing capabilities. Such applications generally require extremely low-power operation capabilities.
For example, implantable devices such as cardiac pacemakers have to work for a long time with a limited power supply. This kind of equipment not only needs low-power analog circuits to detect the physiological signals of the body, but also requires low-power digital and memory functions to convert and store these signals. In addition, advanced implantable devices also require low-power wireless communication to transmit information for the basic units outside the body.
Through a more in-depth analysis of signal types and operating modes, it can be seen that these devices generally have a low duty cycle. For example, they are only activated for a very short time during measurement or processing, and most of the rest of the time is in a dormant state. A duty cycle of less than 1% is not uncommon in these applications. Another characteristic is that most of the signals themselves are in a low frequency state. Therefore, the bandwidth and sampling frequency of the data converter can be limited to tens of kilohertz or even lower. In addition, some consumer devices powered by external batteries have similar performance and power consumption requirements.
In addition to sufficient working performance, these devices also need to have low off-state leakage current according to the above requirements. This means that performance and leakage must be weighed in this process technology. Generally speaking, the gate length of these processes is between 130nm and 350nm, and it may reach 90nm in the future. For portable devices, leakage current performance can vary with changes in process, temperature or power supply. This is an important parameter because it will directly affect the service life of the battery. Figure 4 shows how the leakage current (Ioff) and drive current (Idrive) of NMOS process equipment change with temperature. Idrive has little to do with temperature changes, while Ioff has significant temperature dependence. Figure 5 is a temperature dependence diagram of a PMOS device. Since the temperature change is not large, the variation of Ioff with temperature is acceptable. Figure 6 shows the ring oscillator frequency, which is a typical quality factor for the power supply voltage function of a Display device. It can also be used as a criterion for weighing leakage and performance in practical applications.
Figure 4: Leakage current and drive current change with temperature in NMOS devices
Figure 5: Leakage current and drive current change with temperature in PMOS devices
Figure 6: The ring oscillator frequency is regarded as one of the power supply functions
Another important component for designing low-power mixed-signal devices is high-reliability, small, low-power nonvolatile memory. Ferroelectric memory (FRAM) can provide unique performance and is an attractive choice for non-volatile memory in many applications. Its distinctive features include RAM-like fast write speed, low voltage and low power write Work, long service life, and highly flexible architecture. This memory has been integrated into the low-power digital process technology described above.
The working voltage of FRAM is 1.5V. Unlike floating gate devices, it does not require a charge pump. Like all non-volatile memory, its reliability issues mainly involve write/read cycle durability, data retention, and high-temperature service life. Even after multiple operations, FRAM can maintain excellent aperiodic and periodic bit performance.
When it is necessary to achieve different performance indicators in the same IC, packaging technology can be used efficiently. For example, some applications require digital performance with low noise and low power consumption at the same time, which can be achieved by arranging silicon dies of two different processes in the same package. The silicon die can be stacked to save space on the circuit board. With the continuous development of packaging technology, passive components such as inductors and capacitors can also be integrated in the package. Chip on Board technology can completely embed the entire IC in the printed circuit board, saving valuable space for intensive applications.
The medical electronics industry involves a wide range of fields, and innovations in process and packaging in these fields can help produce innovative solutions. For example: the use of sensors to measure physiological signals on the body surface or even subcutaneously is driving the improvement of elastic substrates and special adhesives. The IC taken with the tablet can not only track the applicability of the medicine, but also play a role in measuring or delivering the medicine. Such applications pose challenges for digestible electronics, tablet coating, and human rejection suppression technologies.
The improvement of the high-voltage (about 100V) process can be proportional to the increase in the density of the ultrasonic transmission channel. The innovation of micro-machining can not only realize the miniaturization, mass production and large number of channels of the ultrasonic probe (CMUT, that is, capacitive micro-mechanical ultrasonic sensor), but also conduct comprehensive analysis experiments (lab on chip or LOC).
Energy harvesting is another emerging field that extends the service life of equipment by partially or completely replacing batteries. Several technologies worth considering are thermal energy, vibration energy, and solar energy. These energy harvesting technologies will bring a new round of demand for circuit design and technology.
The medical electronics industry is constantly evolving, and it has unique requirements for performance, power consumption, and integration. This article only introduces some of these needs and future development trends, but there are many things that need to be explored.