Given the vast array of technologies, value-added electronics, and packaging options available, component selection for pressure sensors can be a daunting task for medical designers. Couple those factors with the increasing pressure designers face to reduce component costs and board size, and the selection process becomes even more complex. While a variety of sensing technologies are certainly available that are adequate for any number of applications, piezoresistive-based pressure sensors may be the smart choice for designers searching for sensors that are particularly well suited for ultra-low-pressure applications. Piezoresistive pressure sensors are small in size, moderately priced, and relatively easy to use. Additionally, the pressure ranges available in piezoresistive sensors are getting lower. Not too many years ago, 0–1 psi was the lowest pressure range available; today, sensors and transducers boast pressure ranges down to 4 and 5 in. of water full scale (27.68 in. H
O/psi). As a result, these sensors are ideal for applications such as respirators, ventilators, spirometers, and other instruments that require measuring very low pressures and in which the media are noncorrosive gases.
A temperature-compensated and -calibrated sensor. A typical transducer.
The two questions most frequently asked by medical designers when selecting a pressure sensor for an application are, "What is the smallest pressure change the sensor can measure?" and "At what accuracy?" Today, pressure sensors and transducers are available in pressure ranges of 1, 4, 5, and 10 in. of water, and 1 psi. If a designer's desired pressure range happens to coincide with one of these ranges, it's simple to determine resolution and accuracy from the data sheet. Unfortunately, however, this is a rare occurrence and, more often than not, the user is forced to determine how the sensor will perform at pressures above or below the specified range. The smallest pressure change a sensor can measure is determined by the signal-to-noise ratio of the sensor. The smallest change the entire system can measure is a function of the entire system and the resolution of the A/D used. For example, if a 5-in.-of-water sensor has an output sensitivity of 5 mV and 2 µV peak to peak of white noise, it gives a signal-to-noise ratio of
The sensor has a 5-mV/V/in. H2O sensitivity. This is the theoretical limit of the sensor alone. To get an idea of the total system noise, all noise generated by amplifiers, other electronics, and the surrounding environment should also be calculated. Filters can be added to improve the system signal-to-noise ratio if necessary.
To answer the second question of how accurately the sensor can measure over the desired pressure range, designers must examine how each parameter changes as a function of working pressure. The relationships of the primary parameters to working pressure are listed in Table I.
|Parameter||Relationship to Working Sensor|
|Linearity error||Varies proportionally to working pressure|
|Varies inversely to working pressure|
|Varies proportionally to working pressure|
|Resolution||Varies inversely to working pressure|
Table I. Summary of the effects of working pressure on sensor specifications.
Many medical applications have working pressures of 1 in. of water or less, and the accuracy of the sensor at 1 in. of water becomes critical. From Table I, one can look at each parameter of a 5-in.-of-water sensor used at 1 in. of water full scale and conclude the following:
Because initial offset and offset TC are the largest error source when operating a sensor at lower pressure ranges, designers should—when possible—design sensors that have automated reference capability, as well as long-term stability and warm-up drift.
A typical transducer.
Bossed and Standard Diaphragms
Understanding the actual mechanical structure of the low-pressure sensor chip can also be helpful to designers when choosing a particular sensor for medical applications in which the instrument or device is portable and when it will be exposed to significant shock and vibration or electromechanical interference. Two approaches for achieving sensitivity at low pressures involve the use of either standard or bossed diaphragms. A bossed diaphragm is one that has been etched with a different thickness within the diaphragm area. With the center of the diaphragm thicker than the outer edges, the strain is concentrated at the sensing resistor locations. This configuration will produce about 10 times more signal for the same size diaphragm. The standard diaphragm, on the other hand, is the same thickness over the entire diaphragm area.
Figure 1. Bossed diaphragm sensitivity distribution.
As would be expected, each of these designs has advantages and disadvantages. The bossed diaphragm gives maximum signal for the same excitation voltage and sensor size. However, the disadvantages are that it is more expensive to manufacture and that the sensor is position-sensitive. The offset will change by the cosine of the angle, and the sensor will measure shock and vibration like an accelerometer. For these reasons, bossed diaphragm sensors shouldn't be used in handheld equipment or for applications in which the sensor will experience a lot of vibration in the vertical axis. Additionally, the sensor noise susceptibility will increase if the device is mounted in a cabinet that has a cooling fan or some other source of random vibration.
Figure 2. Bossed diaphragm offset voltage distribution.
Figure 3. Bossed diaphragm linearity and hysteresis.
Conversely, the standard diaphragm, which is extremely, and uniformly, thin, is not sensitive to position or vibration because of its low mass. The standard diaphragm design will be less expensive to manufacture but will generate less output signal for the same size diaphragm and excitation voltage.
When designing in low-pressure sensors, it's good practice to find out from the vendor what type of sensor is being used rather than searching for the highest output signal available and trying to force it into the design.
Another issue that needs to be addressed when selecting a low-pressure sensor is what sensor configuration to use: full-blown transducers (i.e., those with a high-level output), temperature-compensated and -calibrated sensors, or basic uncompensated sensors. Following is a brief look at each configuration and its advantages.
Transducers. The most expensive of the options mentioned above, transducers with high-level output generally fall within the $20-plus price range (for typical OEM volumes) and are not very flexible. These transducers are assumed to be off-the-shelf standard products. They are calibrated for particular pressure ranges and have a fixed sensitivity. On the positive side, using a transducer reduces or eliminates the need for additional circuitry. It is generally wired into the system and is ready to measure. These factors make a transducer a good choice if a small number of devices need to be built and the pressure range desired is offered off the shelf.
Temperature-Compensated and -Calibrated Sensors. A more versatile selection is the temperature-compensated and -calibrated sensor ($12 to $15 for OEM volumes). Each sensor is temperature compensated (offset and span errors are compensated for) and calibrated so all devices have zero output at no pressure and the same sensitivity. These sensors will be less expensive than transducers, and users can adjust the circuit gain to measure any pressure range.
A big advantage of the compensated sensor is that the customer has control of the electronics in the circuit. If, for example, resolution is very important, users can purchase an amplifier specifically designed for low-noise operation. This alone can improve the signal-to-noise ratio by 10:1 or more when compared with using a general-purpose, low-cost amplifier. Another advantage is that a customer can adjust the amplifier gain to use the full range of an analog-to-digital converter to get maximum system resolution. The compensated sensor requires some analog design, but with the new instrumentation amplifiers on the market, a single 8-pin package and a few external components should be all that are required.
In some applications, more accuracy may be required from the sensor. If a programmable mixed-signal controller is used in the system, some second-order compensating can be done to improve the accuracy of the sensor. For example, the controller can measure the sensor output with no pressure applied and save this reading to do an autoreference on the sensor. Autoreferencing will reduce the calibration and thermal offset errors of the sensor, significantly improving overall accuracy.
Basic Sensors. The least expensive of the sensors listed is the packaged basic sensor ($3 to $10 for OEM volumes). These are sensor elements that have been packaged and functionally tested but have no value-added circuitry or external compensation for temperature effects. Users are left to temperature compensate and calibrate each sensor. Basic sensors are generally used in high-volume applications, where the cost of equipment to calibrate the devices can be amortized over a large number of units. The main advantage to the basic sensor is that users achieve the full range of sensitivities that can have a ratio as high as 2:1. If a low-pressure sensor is desired (1 in. of water or less), having more signal to work with will help significantly.
If the sensor is going to be compensated and calibrated using hardware, the characteristics of the sensor design will dictate whether the sensor is driven by constant voltage or constant current. If the temperature coefficient of the bridge were equal to or larger than the temperature coefficient of span, a constant current excitation would be used. It is also possible to do a digital compensation using a programmable controller. Constant current and constant voltage cost about the same. The application will determine the accuracy required.
Below are terms frequently used in discussions relating to sensor selection. For those needing a quick review, a brief explanation of each term follows.
Sensor versus transducer. The main difference between a sensor and a transducer is that a transducer has a signal-conditioned output (0–5 V dc, 1–6 V dc, 0–10 V dc, etc.).
Linearity. Linearity is the maximum deviation of measured output at a constant temperature from best fit straight line (BFSL). There are many ways to measure linearity; BFSL is the most common among sensor manufacturers.
Sensitivity. This is the ratio of output signal voltage change to corresponding input pressure change.
Span. Arithmetic difference in output signal measured at the specified maximum and minimum operating pressures.
Offset TC. How the voltage will change with temperature when a fixed voltage is applied to the bridge.
Span TC. How the span/sensitivity will change as a result of temperature.
Ratiometric Output. A sensor's span is ratiometric to the applied voltage.
Resolution. The smallest measurable change in pressure.
Contact with Gases and Fluids
Most ultra-low-pressure sensors are by nature nonisolated, meaning that they have wetted materials that are incompatible with media other than dry, ionic gases. These wetted materials generally consist of silicon, gold, RTV, Pyrex, and aluminum.
Common among piezoresistive sensors is an inherent design feature that makes the back side of the chip—which is generally used for gage sensing—more amenable to some limited contact with noncorrosive, nonconductive fluids. In contrast, the front side—where the gold wire bonds and aluminum traces are found—are much more susceptible to sensor failure if contact with fluids occurs. As a result, designers are frequently faced with protecting the sensor from contact with corrosive gases or fluids.
Many manufacturers provide some level of environmental protection that either alone or in conjunction with physical barriers—such as membranes or PTFE filters—can provide the sensor with a limited level of protection from failure due to media contact. A downside, however, to any environmental protection is that offset errors can result from applying any material to an ultrasensitive chip. Consequently, environmental barriers such as silicone gel are better suited to higher-pressure applications. This limitation accentuates the need for a mechanical design of the system that protects the low-pressure sensor from contact with hostile media. Sensor manufacturers are constantly improving the materials used and package designs used in the manufacture of silicon wafers.
As piezoresistive sensor pressure ranges get lower and lower, the medical electronics designer has significantly more options to work with than in the past. The heightened sensitivity and available value-added electronics can provide designers with either a complete plug-and-play solution or, at the very least, the fundamental sensor element that the designers can then augment with other state-of-the-art components to create a completely customized sensor solution.
Duane Tandeske is an applications engineer for SenSym (Milpitas, CA). He can be reached at firstname.lastname@example.org.