Signal conditioning

Signal conditioning is one of the most important technologies in measurement and automation system. It provides the interface between signals/sensors and measurement system. To measure signals from transducers, it must convert them into a form a DAQ device can accept. For example, the output voltage of most sensors in this triaxial testing system is very small and susceptible to noise. Therefore, it might need to be amplified the sensor output before digitizing it. This amplification is a form of signal conditioning. Common types of signal conditioning include amplification, linearization, transducer excitation, and isolation.

4.1.3.1 Amplification

Amplification is the most common type of signal conditioning. Amplifying electrical signals improves accuracy in the resulting digitized signal and reduces the effects of the noise. If amplifying the signal at DAQ device, the signal is measured and digitized with noise that might enter the lead wire, which decreases the SNR.

However, if amplifying the signal close to the signal source with an amplifier, noise has a less destructive effect on the signal, and the digitized representation is a better reflection of the original low-level signal. The SNR is a measure of how much noise exists in a signal compared to the signal itself. SNR is defined as the voltage level of the signal divided by the voltage level of the noise.

The amplification used in this testing system was followed the suggestion mentioned above. That is to amplify the electrical signals measured before them entering the DAQ device. Fig. 4.2 shows the pictures of amplifiers and case.

4.1.3.2 Filtering

Signal conditioning systems can include filters to reject unwanted noise within a certain frequency range. Almost all DAQ applications are subject to some degree of

50 or 60 Hz noise picked up from the power lines or machinery. Therefore, most signal conditioning systems include lowpass filter designed specifically to provide maximum rejection of 50 or 60 Hz noise.

Filters are generally grouped into one of five classifications—lowpass, highpass, bandpass, bandstop, and all-pass. These classifications refer to the frequency range (the passband) of signals that the filter is intended to pass from the input to the output without attenuation. The lowpass filter is the most common used filter in signal conditioning system as the result of the simple and effective. An ideal lowpass filter does not attenuate any input signal frequency components in the passband, which is defined as all frequencies below the cutoff frequency. An ideal lowpass filter completely attenuates all signals components in the stopband, which includes all frequencies above the cutoff frequency. The ideal lowpass filter also has a phase shift that is linear with respect to frequency. This linear phase property means that signal components of all frequency are delayed by a constant time, independent of frequency, thereby preserving the overall shape of the signal. Real filters subject input signals to mathematical transfer functions that approximate the characteristics of an ideal filter.

The Fig. 4.3 compares the attenuation of transfer functions of a real filter and an ideal filter. As shown in Fig. 4.2, a real filter has ripple (an uneven variation in attenuation versus frequency) in the passband, a transition region between the passband and the stopband, and a stopband with finite attenuation and ripple.

In addition, real filters have some nonlinearity in their phase response, which causes signal components at higher frequencies to be delayed by longer times than signal components at lower frequencies, resulting in an overall shape distortion of the signal. This can be observed when a square wave or step input is sent through a lowpass filter. An ideal filter smoothes the edges of the input signal. A real filter causes some ringing in the total signal because the higher-frequency components of the signal are delayed. The Fig. 4.4 shows examples of these responses to a step input.

4.1.3.3 Anti-aliasing filters

Another common use of filter is to prevent signal aliasing—a phenomenon that arises when a signal is undersampled (sampled too slow). The Nyquist theorem states that when you sample an analog signal, any signal components at frequencies greater

than one-half the sampling frequency appear in sampled data as a lower frequency signal. This signal distortion can be avoided only by removing any signal components above one-half the sampling frequency with lowpass filter before the signal is sampled.

Fig. 4.5 shows a sine wave signal sampled at the indicated points. When these sampled data points are used to reconstruct the waveform, as shown by the dotted line, the signal appears to have a lower frequency than the original sine wave. The ways to avoid the signal aliasing are to increase the sampling rate or pass the signal through a lowpass filter to remove high-frequency components. And only analog filters can prevent aliasing. Digital filters cannot remove aliased signals because it is impossible to remove aliasing after the signal has been sampled.

4.1.3.4 Other types of signal conditioning

z Linearization—Many transducers have a nonlinear response to changes in the physical phenomenon measured. For example, a change in voltage of 10 mV for a thermocouple is usually not a change of 10 degrees. Most transducers have linearization tables, or so-called calibration data sheets, that map out how to scale the transducer. The linearization of the transducer can be done either in hardware or software.

z Transducer excitation—signal conditioning system can generate excitation, which some transducers require for operation. Strain gauges and RTDs (Resistive temperature detectors) require external voltage and currents, respectively, to excite their circuitry into measuring physical phenomenon.

The type of excitation is similar to a radio that needs power to receive and decode audio signal.

z Isolation—improper grounding of the system is one of the most common causes of measurement problem, noise, and damaged DAQ devices. When the signal monitored contains large voltage spikes that could damage the computer or harm the operator, so do not connect the signal directly to the DAQ device without some type of isolation. And signal conditioning system with isolation can prevent most of these problems.