Medical Device Link.

Originally Published MEM Fall 2001

SENSORS

A New Technology for Pressure Transducers

Using microcontrollers rather than operational amplifiers, digitally compensated transducers offer performance, space, and cost benefits.

Duane Tandeske

Pressure transducers are used in medical applications that range from R&D testing to sophisticated respiration-monitoring equipment. Any application that involves measuring a pressure or vacuum will employ a pressure transducer.

Consider the portable ventilator that brings fresh air or oxygen to a patient in need of breathing assistance. Pressure sensors in the oxygen delivery system can determine whether the patient is exhaling or inhaling and thus enables the machine to release oxygen only during the inhalation cycle. In the design of standard ventilators, pressure-sensitive alarms alert the clinician or user to possible ventilation malfunction or changes in pulmonary status. Use of a smaller, lower-cost transducer in these applications would allow designers to reduce equipment size and add more functions.

The development of programmable mixed-signal controllers has led to the emergence of new designs in pressure sensor calibration and compensation. In the past, operational and instrumentation amplifiers exclusively have been used to amplify and condition pressure signals. But now the compensated accuracies of which low-cost pressure sensors are capable have been improved by the calculation and memory benefits offered by the microcontroller. Microcontrollers can compensate parameters such as linearity and nonlinear thermal effects with ease by comparison with compensation using resistors and thermistors. The medical equipment designer is looking for the highest overall accuracy at the lowest possible cost. The new digital technology satisfies these demanding objectives.

Hardware Compensation

Two basic methods are now available for compensating and calibrating a piezoresistive pressure sensor: hardware compensation and software compensation.

Hardware compensation physically changes the bridge characteristics of a transducer by means of external laser-trimmed resistors and thermistors. The compensation technique used will depend on the design of the pressure-sensing element. Some sensors are designed to operate in a constant-voltage circuit and others in a constant-current circuit.

The difference between the two designs has to do with how the wafer is processed in manufacturing the pressure sensor. The value and temperature coefficient (TC) of the bridge resistors are determined by the power used when the resistors are ion-implanted.

In contrast, the software compensation technique does not alter the transducer's bridge characteristics. Rather, it modifies the sensing element's analog output signal by digitizing it and then using a microcontroller to mathematically compensate the pressure signal.

Voltage-Sensor Compensation Technique. The design of a voltage-type compensation sensor includes an off-chip temperature-sensitive element—generally, a thermistor—as part of a compensation network on a laser-trimmed compensation ceramic.

In this configuration, the bridge TC is insufficient to compensate span temperature errors, which are always negative and approximately 2000 ppm, and the resistive compensation network has a large negative TC. Adding the compensation network increases the bridge-voltage TC to the desired positive 2200 ppm/ºC.

In the voltage compensation technique, each transducer has to be characterized over the compensated temperature range. This involves measuring bridge impedance, offset, and span at a minimum of three temperatures: 25ºC, the minimum calibrated temperature, and the maximum calibrated temperature.

Figure 1 diagrams the signal path of the constant-voltage hardware-compensated transducer. Resistors R1 through R10 are thick-film laser-trimmed thermistors and resistors that are used to compensate the span and offset thermal effect of the sensor as well as set its offset and sensitivity. Resistors R1 and R2 compensate for span thermal errors, and R3 adjusts span calibration. R9 and R10 are trimmed to set the offset to zero. R4 through R8 compensate for offset thermal errors. The amplifier gain sets the pressure range and output voltage swing. Each transducer is characterized over temperature, and the value of each compensation resistor is calculated and laser trimmed. From Figure 1 it can be seen that all four adjustments interact. For instance, changing the value of R10 not only changes the offset but also changes the offset TC.

Figure 1. The signal path of a constant-voltage hardware-compensated pressure transducer. An asterisk indicates a thermistor.

Constant-Current Compensation Technique. Constant-current compensated sensors use the positive bridge-impedance temperature coefficient to control the bridge voltage and compensate for the negative span-temperature error. This compensation uses the sensor's TC to compensate span thermal errors, eliminating any chance of temperature gradients from component to component. When the sensor is designed, the bridge TC is forced to be larger than the span TC; during the compensation process, a resistor is placed in parallel with the bridge and selected to force the bridge TC to equal the span TC. Each transducer has to be characterized over temperature as with the constant-voltage technique, and all the adjustments interact with each other.

Figure 2 shows the signal path of a constant-current compensated pressure transducer. In this configuration, resistors R1 through R6 are laser trimmed or selected to temperature-compensate and calibrate the transducer. Resistors R1 and R2 are adjusted to compensate for offset thermal errors. R3 and R4 are adjusted to set the output-offset voltage, R5 is adjusted to compensate for span thermal-effect errors, and R6 is adjusted to set the sensor span.

Figure 2. The signal path of a constant-current hardware-compensated pressure transducer.

An examination of the two hardware compensation techniques reveals that they both force the bridge voltage to have a positive temperature coefficient equal in magnitude to the negative temperature coefficient of the gauge factor. Paralleling one resistor in the bridge with a large low-TC resistor compensates the offset TC. Adding the thick-film resistor reduces the effective TC of the bridge resistor. The end result is the same with both techniques.

Software Compensation

Programmable mixed-signal controllers have made a new compensation technique practical and cost-effective. Offset, offset TC, and linearity are just a few of the parameters that can be compensated digitally. The new controllers are application-specific integrated circuits (ASICs) that include multiple analog input channels, analog-to-digital (A/D) convertors, digital-to-analog (D/A) convertors, and internal temperature-sensing circuits—all the hardware necessary to monitor and adjust the various sensor parameters. The individual characteristics of each sensor have to be measured and the calibration constants stored in the microcontroller memory. The microcontroller measures the bridge output as well as the temperature sensor, and calculates the ideal transfer curve using the calibration constants stored in memory.

Figure 3 is a basic block diagram of the software-compensated pressure-transducer signal path. The input multiplexer selects the bridge output or temperature signal as the input signal to be sent to the A/D convertor. The charge-balancing A/D convertor then digitizes the analog signals for the 16-bit microcontroller.

Figure 3. The signal path of a software-compensated pressure transducer.

The microcontroller in turn performs calibration and temperature-compensation of the pressure signal, using calibration constants that have been stored in the electrically erasable programmable read-only memory (EEPROM). Offset, offset TC, span, span TC, linearity, and offset and span calibration are compensated. Finally, the output of the microcontroller is converted to a 0.5- to 4.5-V analog signal by the 11-bit D/A convertor.

Software and Hardware Compensation Compared

Compare the block diagrams in Figures 1–3. Note that in both the constant-voltage and constant-current devices (Figures 1 and 2) there is a resistor in series with the top of the pressure sensor. This resistor will drop the bridge voltage by 20 to 60% in some constant-voltage devices. Since the output signal from the pressure sensor is directly proportional to the bridge voltage, the smaller bridge voltage reduces the sensitivity and resolution of the device by 20 to 70%.

Digital compensation, however, does not change the bridge voltage as a function of temperature. Thus, the pressure sensor is operated at maximum bridge voltage, taking full advantage of its capabilities with respect to sensitivity and resolution.

The larger signal has a positive effect on the signal-to-noise (S/N) ratio at the pressure sensor level and reduces the amount of gain necessary to obtain the desired output signal.

Another advantage of software compensation for a pressure transducer is that sensor characteristics are not modified during the compensation procedure. Therefore, there is no interaction among parameters. If a nonstandard specification is needed, it is much easier to modify software than to lay out a new compensation ceramic.

Response time differs significantly between hardware-compensated transducers and software-compensated devices. The sensing element is very fast in both cases, exhibiting a rise time of less than 100 microseconds, so the distinction has to relate to differences between the amplifier and the microcontroller.

The instrumentation amplifier has a typical rise time of 1 V/µs, or 5 microseconds for a 5-V swing. This is significantly faster than the pressure-sensing element. The sensor has a self-resonant frequency in the range of 30–50 kHz, which corresponds to a response time of 20–100 microseconds. Therefore, in the hardware-compensated transducer, the sensor is the limiting factor.

The software-compensated transducer uses the same pressure sensor as the hardware-compensated transducer. It connects the analog output of the sensor to the ASIC, where the output is digitized by an A/D convertor and processed by a microcontroller, then converted back to analog output by a D/A convertor.

This process takes something like 10 milliseconds, depending on the clock frequency. Software-compensated transducers therefore would not be a good choice for an application in which rapidly changing pressures need to be measured. Their response times, which are 50 to 100 readings per second, are suitable for measuring static pressures or pressures that change only moderately quickly.

Resolution—that is, the smallest change in pressure the transducer can detect—is another parameter that will be different in hardware- and software-compensated transducers. The resolution of the hardware-compensated transducer is determined by its S/N ratio. Because the amplifier generates most of the white noise in the trans-ducer, selecting an amplifier with low noise specifications will improve the S/N ratio.

The software-compensated transducer, on the other hand, has a resolution specification similar to that of an A/D convertor: the output voltage swing divided by the number of bits (2n, where n is the number of bits) determines the least-significant-bit value. An 11-bit D/A convertor and a 4-V span yields a least significant bit of 1.95 mV (4 V divided by 2048), which would be the smallest change in pressure to which the transducer could react.

Design Benefits

What benefits can the digitally compensated transducer offer a medical equipment design engineer? Cost and sensing accuracy are chief among the many advantages provided by software-compensated transducers. Size and reliability are also issues that come up constantly when a pressure transducer must be selected. Design engineers are always looking for ways to make systems cheaper, smaller, and more reliable. Reducing the parts count of the pressure transducer is one way to make it both smaller and more reliable at the same time.

Cost. The design engineer has to keep the finished product within a price range that makes it competitive in the marketplace while also being careful to design in the desired overall functional performance. When selecting a pressure transducer, the engineer has to consider the expense of assembly and calibration labor as well as component cost. For their part, pressure transducer manufacturers are trying to reduce the selling price of their components while maintaining performance specifications. Microcontrollers have become very cost-effective and are competitive with operational amplifiers. Also, electronically programming a transducer is less labor-intensive than laser-trimming compensation ceramics and installing them into the transducer. Software-compensated transducers, for all these reasons, are an appealing choice for designers looking at performance-to-cost ratios.

Accuracy. Digital compensation improves overall transducer accuracy by compensating sensor linearity. Linearity is not a parameter that can be compensated in any of the hardware techniques.

Software-compensated transducers have much better thermal performance. The compensation technique of trimming thick-film resistors can only provide a linear correction of a sensor's output. However, software-compensated transducers use a calibration microcontroller that incorporates a second-order curve-fitting equation to minimize the nonlinear span and offset temperature coefficients, producing much better thermal errors over a wider temperature range (see Figure 4).

Figure 4. Typical errors from –10° to 85°C. Offset = ± 0.2 (% FSS). Full-scale output = ± 0.3 (% FSS). Offset TC = ± 0.3 (% FSS). Span TC = ± 0.5 (% FSS).

Packaging. Hardware-compensated transducers of both the constant-voltage and constant-current types consist of three basic parts: a pressure sensor, a compensation ceramic, and an amplifier. The compensation ceramic is worth looking at. Whatever the type of transducer, and whatever the configuration of the compensation device, all of the laser-trimmed thick-film resistors are screened onto a ceramic that is either placed inside the package or attached to the outside of it. Thick-film resistors are generally large-geometry resistors because of the necessity of getting the trim range without compromising the resistor. Some compensation ceramics have an area of 1 sq in. or more. They have to be that large to accommodate the number of resistors and thermistors necessary to calibrate and compensate the transducer. The size of the compensation ceramic in turn determines the size and construction of the overall package.

A software-compensated transducer contains no compensation ceramic. It consists merely of a pressure sensor and an ASIC. With no compensation ceramic to accommodate, a digitally compensated transducer can be smaller than a hardware-compensated one. Overall package size is determined by the size of the two silicon chips and the number of electrical leads on the package. Another advantage is that the package does not have to be designed to allow access to the compensation ceramic for laser trimming (see Figure 5).

Figure 5. Three-dimensional assembly drawing of the transducer.

Reliability. Another benefit of removing the compensation ceramic is improved reliability. The ceramic is brittle and can be broken during installation or maintenance. Also, if the compensation ceramic comes in contact with harsh chemicals, the resistors and the metal traces can suffer. The assault can spoil the calibration of the transducer, or cause the device to fail completely. In addition to removing this problem, elimination of the compensation ceramic also reduces the number of electrical connections that have a direct effect on transducer reliability. And finally, a compensation ceramic has a large mass. The presence of one in a transducer limits the amount of shock and vibration the device can tolerate.

Conclusion

Medical equipment design engineers now have more pressure-transducer options from which to select than in the past. Devices offering digital compensation provide significant design advantages for many applications. New software-compensated transducers are smaller, less expensive, and more accurate than their hardware-compensated counterparts. Impressive accuracy specifications allow for greater freedom to devise plug-and-play designs that minimize the calibration time of the finished product.

Duane Tandeske is an applications engineer at SenSym ICT (Milpitas, CA). He can be reached at ddtanapps@aol.com.

Copyright © 2001 Medical Electronics Manufacturing