Designing Superior Signal Paths for Portable Medical Devices

Companies developing medical diagnostic equipment face the challenge of providing consumers with high-quality and affordable products. The ability to reduce the size of these medical devices and improve their accuracy is of utmost importance in reducing the cost of medical care and improving patient care services. This need is even more urgent as the population becomes increasingly aging. According to the InMedica report of IMS Research, the total sales of consumer medical devices are expected to exceed $5 billion by 2011.

The development and continuous improvement of medical device monitoring functions will enable remote care providers to provide better diagnostic tools for home patients, emergency room paramedics and even hospitals in several important areas of human health. Monitoring instruments such as blood pressure monitors, blood glucose meters, and defibrillators require clear analog signals to make accurate measurements, otherwise they may endanger lives. Designing excellent analog signal paths can help designers reduce interference from external noise, extend dynamic range and enhance accuracy. In addition, designers must carefully select components to meet the performance requirements of the end product. High performance requirements in small packages Previously, people generally believed that medical equipment in hospitals and clinics was more accurate than portable instruments used at home. However, new technology trends are quickly changing this view. New portable medical devices are used not only by ordinary consumers but also by technologically savvy patients, so customers' needs are no longer limited to taking body temperature, doing electrocardiograms and measuring blood pressure. Customers need a full range of care and measurement functions. In order to meet the urgent demand for home medical diagnostic instruments, equipment suppliers are relying on advanced inventory management and innovative designs to enhance market competitiveness and equip their products with more functions to attract more users. In the field of developing home medical instruments, one factor is very important, that is, the development time required from the initial design to the actual market launch of the product. Shortening the time to market allows manufacturers to seize the market. Whether the development cycle can be shortened depends on whether the system designer's design is flexible enough and cost-effective. Process technology affects system design Although electrical specifications are the main factor for designers to select components, the process used to manufacture integrated circuits is equally important. For example, a typical blood glucose meter usually requires an operational amplifier with very low input bias current, and most designers will choose a JFET amplifier. However, they should consider the temperature issue before making a decision. Because JFETs have a low initial input bias, they are very susceptible to temperature changes, with the input bias approximately doubling for every 10°C rise. To calculate the input bias drift, use the following formula (reference 1). Ib(T)Ib(T0) x 2(T-T0)/10For example, a JFET input op amp such as National Semiconductor's LF411 has an input bias current of 50pA at 25°C, while a better choice is National Semiconductor's LMP7731, a bipolar input op amp with an input bias current of 1.5nA. Using the above formula, we can quickly calculate that at 85°C, the LF411's input bias current becomes 3.2nA, more than twice that of the LMP7731. Evaluating System Trade-offs Speed, noise, and power consumption can be equally important for some designs. A low-noise device consumes more current, while a low-power device offers limited bandwidth. One way to overcome these problems is to use an anti-compensated amplifier in the appropriate application. In addition to being unity-gain stable and fast, the advantages of an anti-compensated amplifier are that it provides wide bandwidth without sacrificing power consumption. Anti-compensated op amps are best used in current-to-voltage conversion (transimpedance) circuits. One of the most common applications in medical instrumentation is to measure the oxygen content in blood cells, called SPO2 or saturated or peripheral oxygen. Figure 1 shows a block diagram of an SPO2 module, where an anti-compensated amplifier (TIA) is used to convert the current from a photodiode into a voltage. Figure 1 Typical block diagram of an SPO2 module Shortcuts to reduce design time The most important parameter in medical instrumentation is noise, which can cause serious interference to the circuit itself and nearby equipment. Calculating noise is a tedious task, especially when you want to calculate the overall contribution of the signal path from the power supply, amplifier, data converter, and external components to the signal-to-noise ratio. Generally, medical instrumentation circuits tend to operate at lower frequencies, so designers of these systems are usually more concerned with noise in the 0.1 to 10 Hz band, also known as peak-to-peak noise. Unfortunately, some data sheets do not provide a value for time-domain noise (peak-to-peak), but instead provide a typical graph of voltage or current noise density. In addition to waiting for the circuit vendor to provide the measured data, there is a quick method to help deduce the peak-to-peak amount of noise. Suppose you want to use National Semiconductor's LMP7731 to estimate the peak-to-peak (0.1 to 10 Hz) voltage noise. First, select a point in the frequency range within the specified frequency band, such as 1 Hz, then the value when comparing the curve is 5.1nV/√Hz (Figure 2). Then use the following formula to calculate the root mean square (RMS) value of the noise: Formula 1: enrms=enf√ln(10/0.1), where enf is the noise at 1Hz Figure 2 LMP7731 input voltage noise vs. frequency Frequency, voltage noise The above formula gives a total root mean square noise of 10.9nV. To calculate the peak-to-peak noise, simply multiply this root mean square value by 6.6, which gives 72.2nV. This is a pretty good estimate, and it is very close to the specification of 78nV listed in the data sheet. If the voltage noise density plot in the datasheet does not indicate the noise value at 1Hz, you can use the following simple equation (Equation 2) to deduce the value at a certain frequency. Equation 2: en = enb * √ (fce / f) where enb is the broadband noise (usually the value at 1kHz), fce is the 1/f corner point, and f is the frequency of interest, in our case 1Hz. For example, the broadband noise of the National Semiconductor LMV851 at 10kHz is 10nV/√Hz. To calculate the RMS noise, first determine the value of the 1/f corner point (fce) from the graph. Using the voltage noise density plot in the datasheet, this will find that fce is approximately 300Hz. Then, using the above equation, we can calculate en = 10 * √ (300 / 1) = 173nV√Hz, which is the voltage noise at 1Hz. Finally, substituting this value into Equation 1 and multiplying the result by 6.6, we get 2.4μV of peak-to-peak noise. Another thing to consider is current noise. Generally speaking, if the impedance of the source is not very large (>100kΩ), you can ignore the current noise and still get a very close estimate, as in the example above. However, if the impedance of the source is very large, then the same technique must be used to estimate the current noise, and the voltage and current noise must be added together in root mean square form. Determining the speed requirement




















































Just as the noise of the op amp is critical to the resolution of the ADC, bandwidth is equally important to maintaining the accuracy of the system. To limit the error to 1/2 least significant bit (LSB), a quick check is needed to determine if the bandwidth of the amplifier is sufficient. Instead of using complex and general instructions, you can use the resolution of the analog/digital converter to quickly extrapolate the result. This is done by using 1/2(N/2) and multiplying the result by the amplifier frequency at -3dB (Reference 2).

Using the above shortcut and a 14-bit ADC, this example yields feff = 0.007813*f- 3dB. For the op amp (LMP7711) with a configurable gain of 10 in Figure 3, the frequency at -3dB is 1.7MHz. Thus, the maximum bandwidth (at 1/2 LSB error) is equal to 0.007813*1.7E6=13.3kHz.


Figure 3 Block diagram of a portable electrocardiograph

Monitoring devices and communication devices

Most newer medical diagnostic devices have wireless communication capabilities. Modern electrocardiographs (EKG or ECG) can transmit patient data to a doctor's office or hospital within minutes via a personal electronic notebook (PDA) or other computer peripheral device. Despite the benefits of wireless data transmission, such devices may cause serious interference to medical devices, causing them to have erroneous readings.

To avoid this interference, filters must be used. However, adding filters not only increases the size of the device, but also increases the cost of the design. A more cost-effective and faster method is to use components (including filters) that can suppress radio frequency (RF) noise.

Conclusion

The trend in the medical device field today is to provide consumers with higher value products, that is, home care devices that are inexpensive and can provide diagnostic results quickly. As technology continues to advance, more medical devices will transmit data instantly from the patient's home to the doctor's office via computers. In addition, as users demand more functions, the accuracy requirements of portable medical devices will be further improved to achieve more accurate diagnosis, which will rely on designers' continuous innovation, long-term development and commitment to comprehensive solutions.