Design engineers and those specifiers with a high degree of technical knowledge will recognise that there is no standard definition of accuracy for pressure sensors. To some degree this is understandable; manufacturers don’t want to confuse a potentially simple buying decision with additional – and some might argue unnecessary – technical data. As a result, there can be a wide variation in the way sensor performance is presented in a data sheet.
It is not unheard of, for example, for a manufacturer to provide only typical specifications or to greatly limit the number of error factors represented in their accuracy specification. Specifications, therefore, may exclude errors, or may be calculated over a very narrow range, at only one point in the range, or at a point where the absolute best accuracy is represented. Either way, the ‘true’ performance of the sensor may be difficult to discern.
If all manufacturers used the same parameters with which to measure their sensors, then the raw numbers in data sheets would be a fair way of comparison. The designer could then find the detailed information needed to fully understand the performance of each sensor under the specific conditions required for his application. Unfortunately, life is never that simple.
So what are the major error terms inherent in a MEMS pressure sensor against which each product should be measured, and how can the design engineer maximise his chances of selecting the right product for the right purpose?
A definition
A pressure sensor is a device that is used to convert a mechanical pressure value into an electrical signal. The behaviour between pressure input and output signal is described with the help of a linear equation known as the 'transfer function’.
Mathematically, the ideal transfer function is a straight line, which is independent of temperature, passing through the ideal Offset with a slope equal to the ideal slope triangle of full scale span over the operating pressure range.
Variations in slope, offset and deviation of transfer function from linearity result in variances of the output value from the theoretical sensor equation. These variances provide us with the error of the sensor.
Calculating the error is only one part of the story. What is really required is the total error band for fully integrated (compensated, calibrated and amplified) sensors that enable customers to choose what they want quickly and easily without having to calculate how an individual error that might be encountered in their application will impact the sensor’s performance.
The total error band specifies maximum deviation in output from the ideal transfer function over the entire compensated temperature and pressure range. It includes all errors due to: offset, full scale span, pressure non-linearity, pressure hysteresis, non-repeatability, thermal effect on offset, thermal effect on span and thermal hysteresis – indeed all eight errors that we believe should be included in a performance calculation.
Error values of the total error band are given in maximum values. It means that customers do not have to assess typical value definition and probabilities of mismatch rate because all products are designed to remain within a total error band after exposure to the specified environmental conditions, including the solder reflow process.
Errors in offset and span are variations in ideal parameters of the output function caused by manufacturing tolerances and drifts over time. Manufacturing variations are corrected in an additional calibration process performed on every sensor to reduce these variations to the barest minimum. Long term drift influences are guaranteed by design over lifetime of the sensors.
The next group of errors in the total error band centre on the accuracy of the sensor. Accuracy describes how the sensor’s output follows the ideal output function, which is described by offset and slope. Deviations of the output function from a straight line are summarised in pressure non-linearity.
The difference between output readings when the same pressure is applied consecutively, under the same operating conditions, with pressure approaching from same direction is given as a ‘non-repeatability’ figure, and from opposite directions it is captured in ‘pressure hysteresis’ errors.
The last group within our total error band calculation comprises thermal influences. These lead to changes in offset and span of the sensor. Thermal hysteresis is the difference between output readings when the same temperature is reached consecutively, under the same operating conditions, with temperature approaching from opposite directions within the specified temperature range.
Conclusions
Specifying the right pressure sensor for the right application is a challenging task. Manufacturers, however, could make that task easier by being more transparent and consistent in the way in which their performance data is presented. Standardising that data around the concept of a total error band, and giving that band a ‘value’, is certainly one way of ensuring that design engineers are comparing like for like.
Jochen Schiffner is the EMEA Applications Engineer for pressure sensors at Honeywell Sensing and Control