Originally Published MEM Fall 2008
COMPONENTS
) |
(click to enlarge)
An EMI-hardened op amp (top) and a low-noise precision op amp (bottom) from National Semiconductor. |
Companies developing medical diagnostic equipment are faced with the challenge of providing customers with the best possible functionality at the most attractive price. Reducing the size and improving the accuracy of these medical tools is important for lowering healthcare costs and improving patient care, especially for the growing senior population. According to InMedica, the research division of IMS Research, sales of consumer medical devices will reach more than $5 billion by 2011.
Through new and increased monitoring functions of medical equipment, remote caregivers can watch key areas of a patient's health, providing better diagnoses for review by at-home patients, emergency field personnel, and clinicians. Monitoring equipment, from blood pressure monitors to glucose meters and defibrillators, requires clear analog signals for precise measurement of critical vital signs. Designing an optimal analog signal path helps engineers overcome challenges such as reducing outside noise interference, maximizing the dynamic range, and improving accuracy. Designers must carefully choose the components that make up the final product to meet these demanding specifications.
Great Performance in Small Packages
Until recently, medical equipment in hospitals and practitioners' offices was generally considered more accurate and more precise than portable counterparts used at home. The current trend, however, is changing rapidly to accommodate not only average consumers, but also tech-savvy patients whose interest goes beyond taking one's temperature or determining systolic and diastolic pressure.
With an increasing demand for home-use medical diagnostic equipment, manufacturers of such devices rely on state-of-the-art inventory and design talent to remain competitive and strategize their products by offering more functionality. One of the most important factors in the development of medical apparatuses for consumers is development time, from the initial design to the actual marketing of the product. Reducing time to market enables manufacturers to have their products adopted among users sooner. And, in order to minimize cycle time, system designers rely on cost-effective and clever designs.
Process Technology Affects Designs
Although electrical specifications are a major factor in choosing the components for a design, the process on which integrated circuits (ICs) are built can be equally important. As an example, glucose monitors typically require an operational amplifier (op amp) with a very low input bias current. Many designers might choose a junction field effect transistor (JFET) amplifier, but it is important to consider the temperature effect before making a decision.
While JFETs have a very low initial input bias, it is greatly affected by temperature variations, so much so that it roughly doubles for every 10°C increase. To determine the input bias drift, use the relationship shown in the following equation:1
Ib(T) ≈ Ib(T0) × 2(T – T0)/10. (1)
For example, a JFET input op amp offers an input bias current of 50 pA at 25°C. A better choice in this case is a bipolar input op amp with a bias current of 1.5 nA. Using the equation above, at 85°C, the bipolar input op amp input bias current becomes 3.2 nA, more than double that of the JFET input op amp.
Evaluate System Trade-Offs
Speed, noise, and power consumption trade-offs can all be important for some designs. A low-noise device consumes more current, and a low-power device has a limited bandwidth. One way to overcome such trade-offs is to use decompensated amplifiers in appropriate applications. Decompensated amplifiers have an advantage over their unity-gain stable, high-speed counterparts. The advantage that decompensated amplifiers have is wide bandwidth without a power consumption penalty, in addition to a lower overall price.
) |
(click to enlarge)Figure 1. A typical block diagram foran SP02 module.
|
Decompensated op amps are well suited for I-V conversion (transimpedance) circuits. One particularly popular application in medical instrumentation is the measurement of oxygen saturation within blood cells. This is known as saturation or peripheral oxygen (SPO2). Figure 1 shows the block diagram of an SPO2 module in which a decompensated amplifier is used to convert the current from the photodiode into a voltage.
Clever Shortcuts Minimize Design Time
One of the most important parameters in medical instrumentation is noise, which can cause serious interference problems in the circuit itself and in neighboring equipment.
Noise calculations can be rather tedious, especially when trying to determine the total contribution of the signal path from the source, amplifiers, data convertors, and external components to the signal-to-noise ratio (SNR).
Generally, medical instrumentation circuits tend to operate at very low frequency, so designers of these systems are often concerned with noise contained in 0.1–10 Hz, also known as peak-to-peak noise. Unfortunately, some data sheets do not specify noise in the time domain (peak to peak), but will have a typical graph of noise density, be it voltage or current noise. Rather than wait for the IC manufacturer to provide the actual measurement, it is possible to quickly estimate the peak-to-peak noise.
) |
(click to enlarge)Figure 2. The input voltage noise versus frequency for the LMP7731 op amp.
|
Suppose you want to approximate the peak-to-peak (0.1–10 Hz) voltage noise using a bipolar input op amp such as National Semiconductor's LMP7731. First choose a point in the frequency range within the specified band, for example at 1 Hz; the value is 5.1 nV/√Hz (see Figure 2). Then compute the root mean square (rms) noise as follows:
enrms = enf√ln(10/0.1), (2)
where enf is the noise at 1 Hz.
Using this equation above yields a total rms noise of 10.9 nV; to get the peak-to-peak noise, simply multiply the rms value by 6.6 to get 72.2 nV, which is a pretty good estimate when compared with the value of 78 nV listed in the data sheet spec table.
If the voltage noise density graph in the data sheet does not show the value at 1 Hz, the following simple formula can be used to approximate the value at the interested frequency:
en = enb × √(fce/f), (3)
where enb is the broadband noise (usually the value at 1 kHz), fce is the 1/f corner, and f is the frequency of interest, in this case 1 Hz.
As an example, consider the National LMV851, whose broadband noise is 10 nV/√Hz at 10 kHz. In order to calculate the rms noise, first determine the 1/f corner (fce) graphically. Using the voltage noise density data sheet graph, find the fce of approximately 300 Hz. The above equation then yields
en = 10 × √(300/1) = 173 nV√Hz.
This is the value of the voltage noise at 1 Hz. Then, plugging it into Equation 2 and multiplying the result by 6.6 gives a peak-to-peak value of 2.4 µV.
Another consideration is current noise. Generally, if the source impedance isn't too large (it should be >100 kΩ), it is possible to ignore current noise and still get a very good approximation as seen in the example above. If, however, the source impedance is high, then the same technique can be used to estimate current noise and add the voltage and current noises in rms fashion.
Determine the Speed Requirements
Just as the op amp noise is important for the resolution of the analog-to-digital convertor (ADC), bandwidth is essential to maintain system accuracy. In order for the error to be within ˝ least significant bit (LSB), a quick check is needed to determine whether the amplifier bandwidth is wide enough. Rather than using complex, tedious derivations, the assessment can be made very quickly using the ADC resolution, which allows the bandwidth needed to be determined. To do that, simply use ˝(N/2) and then multiply the result by the –3 dB frequency of the amplifier.2
Using this shortcut and a 14-bit ADC, this example states
feff = 0.007813 × f –3 dB.
) |
(click to enlarge)Figure 2. A block biagram of a portable ECG.
|
In Figure 3, the op amp is National's LMP7711. With a configured gain of 10, the –3 dB is 1.7 MHz, and the maximum bandwidth (to stay within ˝ LSB error) is
0.007813 × 1.7E6 = 13.3 kHz.
Monitoring Devices and Communication Devices
Many newer diagnostic devices include wireless communication. Modern electrocardiograms can transmit patient information to the practitioner's office or hospital via a PDA or other PC peripheral within minutes. Despite the advantage wireless data transmission offers, such equipment can cause serious interference to a medical device, resulting in erroneous readings. To avoid such interference, filters should be implemented. However, adding filters will likely increase the size and the cost of the design. A more economical and quicker approach is to use components (which include filters) to reject radio-frequency noise.
Conclusion
The trend in the medical device industry is to provide high value to its customers by offering affordable, quality products with speedy diagnostics for patient care. It is also imperative that designers find cost-efficient ways to address these requirements. As technology progresses, more medical equipment will use computer interfaces to allow data to be instantaneously transmitted from the patient's home to the practitioner's office. Choosing components for an optimal signal path will be an essential part of the design process. As the demand for multifunctionality increases, portable medical devices will require more precision to achieve better accuracy. Such feats will require innovation, long-term development, and commitment to total solutions.
References
- Sergio Franco, Design with Operational Amplifiers and Analog Integrated Circuits, (New York: McGraw-Hill, 1998).
- Walter Bacharowski, "Decompensating Amplifiers Improve Performance," Electronic Design News [online] December 3, 2007 [cited 1 August 2008]; available from Internet: www.edn.com/article/CA6505576.html.
Soufiane Bendaoud is a marketing manager at National Semiconductor Corp. (Santa Clara, CA).
