Descriptores Funcionales de los Mecanos

When Instrumentation Engineers discuss signal conditioning, they are usually talking about analog signal conditioning. Traditionally, data acquisition systems have acquired analog data in the form of temperatures, accelerations, strains, positions, etc. This type of data has always required analog signal conditioning in order for the data system to accept it as an input source. For example, the full-scale output of a transducer may be in the range of 0-20mVDC where the input range to the data system is 0-5Vdc. In this case, it is obvious that voltage amplification is required to take advantage of the dynamic range of the instrumentation system.

Another example could be the case where the output of the transducer is a high level signal in the range of 0-5VDC and the data system input is +2.5Vdc. For this case, the signal conditioning required would be a zero shift.

Analog signal conditioning types can be lumped into several categories: amplification (and its opposite attenuation), signal conversion (frequency-to-DC, AC-DC, etc.), and zero shifting. Filtering can also be lumped into this category but a separate section has been dedicated for this topic.

Amplifiers are by far the most common piece of signal conditioning because of the wide range of uses, such as amplification, attenuation, DC-shifting, impedance matching, common-mode rejection, isolation, and others. Many operational amplifiers exist for very specific requirements such as voltage-in/voltage-out, current-in/voltage-out, and charge-in/voltage-out (commonly referred to as a “charge amp”). Determining the type of amplification is the key in providing accurate measurements through the data acquisition system.

Voltage amplifiers suitable for instrumentation system use typically provide high input

this is accomplished by the amplifier requiring very little current from the transducer and insuring minimal intrusion into the measured quantity. This minimizes source loading thus increasing measurement accuracy and insuring minimal intrusion into the measured quantity.

This is true for transducer signals referenced to ground (single-ended) and transducers that output difference signals (differential). Because the transducers (such as in Wheatstone bridge implementations) in this second group are very common, an operational amplifier was optimized for this type of signal. The amplifier is called the instrumentation amplifier or “in-amp.”

The instrumentation amplifier amplifies the difference between two signals. This difference is usually in the millivolts range. When this number is compared to the common mode voltage, (the voltage on each terminal of the instrumentation amp referenced to ground) it is obvious that these instrumentation amps must have very large common mode rejection values.

This value is called the common mode rejection ratio (CMMR), which in decibel (dB) is defined as:

Gdiff is the differential gain of the amplifier

Vcm is the common mode voltage at the input of the in-amp Vout is the output resulting from the presence of Vcm at the input

This is a very important concept in the world of measurements. Common mode voltages commonly exist in an instrumentation system. It should also be noted that CMMR is tied directly to frequency with less common mode rejection occurring at higher frequencies. The graph at Figure 4-1 illustrates this fact.

Figure 4-1. CMR versus frequency.

Errors due to stray common mode voltages are very common in every instrumentation system. Any common mode noise not rejected will be amplified and introduced as measurement error. Relying strictly on the instrumentation amp’s rejection capability is not a good design practice. Pay close attention to cable routing and cable shielding to minimize stray common mode voltages.

Isolation or buffer amplifiers also play a key role in an instrumentation system.

Typically, these amplifiers are characterized by a high isolation resistance (~10¹²Ω), very high input impedance, high CMRR, and gains on the order from 1-10. These are commonly used to isolate critical system parameters, such as flight control signals, from the instrumentation system. Opto-isolators also provide a level of isolation and provide another means of isolating signals from the data system.

Sometimes the transducer output is at a level too high for the data system to accurately capture. When this is the case, attenuation of the signal must occur. Sometimes this is as simple as providing a resistor dividing network provided the transducer has the drive capability so the impedance of the network does not load the output (see Figure 4-2).

Figure 4-2. Resistor divider network.

Another method, shown at Figure 4-3, is to use an amplifier configured to provide a fractional gain.

Figure 4-3. Fractional gain inverting amplifier.

Current amplifiers also exist and most commonly are the current-to-voltage amplifiers (which usually reside near the data acquisition system input). Another method used is to pass the current through a precision resistor and amplify the voltage across the resistor with a

on the self-generating, piezoelectric effect of either quartz crystals or ceramic materials to produce an electrical output signal proportional to the physical input. Some of these transducers have built-in microelectronic amplifiers, which convert the high impedance charge signal from the crystals into a low impedance voltage output signal. Piezoelectric transducers that do not contain additional circuitry are known as charge mode or high impedance output transducers therefore requiring a charge-in/voltage-out amplifier or “charge amplifier” (see Figure 4-4).

Piezoelectric transducers output a high impedance level of charge based upon a dynamic input such as shock or vibration. The charge amp senses this by charging a highly stable capacitor in its feedback loop. Making a few assumptions, mainly that the open-loop gain of the amplifier is very large, the low impedance output of the amplifier can be expressed as:

f a

out C

V ≅− q where

qa is the charge sensitivity of the accelerometer in pC/g (in this example) Cf is the feedback capacitor

Figure 4-4. Simplified charge amplifier.

Figure 4-5 shows a typical charge mode sensor system including sensor, low noise cable, and charge amplifier. A piezoelectric transducer has a high output impedance and can be

modeled as a signal source and a capacitor. The piezoelectric sensing element generates an electrical charge signal. A charge amplifier or in-line charge converter (explained above) utilizes high input impedance, low output impedance inverting amplifiers with capacitive

feedback. Adjusting the value of the feedback capacitor alters the transfer function or gain of the charge amplifier. Typically, charge mode transducers are used when high temperature

survivability is required. If the measurement signal must be transmitted over long distances, use an in-line charge converter near the transducer. This minimizes the chance of noise and prevents transducer sensitivity loss.

Because of the high-impedance nature of the output signal generated by charge mode transducers, several important precautionary measures should be followed. Use special low-noise coaxial cable between the transducer and the charge amplifier. This cable is specially treated to reduce triboelectric (motion induced) noise effects. Also, always maintain high insulation resistance of the transducer, cabling, and connectors. To insure high insulation resistance, all components must be kept dry and clean.

Figure 4-5. Typical piezoelectric transducer, cable, and charge amp.

The last type of signal conditioning to be discussed in this section is zero shifting, commonly referred to as offset. This involves providing a DC bias, either positive or negative, usually along with amplification, to the transducer signal. As simplistic as this method is, it is used throughout the instrumentation community to optimize the transducer signal for data acquisition.

4.3 Filtering

Entire textbooks and courses have been dedicated to this vast topic. Filters in this context are used as frequency-selective signal conditioners. Filtering is used to transmit wanted and attenuate unwanted frequency content in the measurement signal. They reduce the amount of noise outside of the data bandwidth and can also be used to “select” certain bands of

frequencies. Filtering is always used for analog signals when digitization is done. By nature, instrumentation systems are band-limited. Filtering controls where the band limiting occurs.

In any instrumentation system, there are two distinct and separate reasons for capturing the data. Either you want the frequency content of the waveform or you want the waveform itself. Filtering for frequency content is much easier. The instrumentation engineer must satisfy two requirements:

a. All frequencies of interest must lie in the flat portion of the filter transfer function (flat amplitude response)

b. All amplitudes must lie in the linear range of the instrumentation system’s transfer function (input/output linearity)

c. In order to reproduce the waveform, a third criteria is added: All frequencies must lie in the linear phase portion of the instrumentation system’s transfer function (linear phase response).

a. Low pass. As their name suggests, low-pass filters pass frequencies below their

cut-off value.

b. High pass. High-pass filters pass frequencies above their cut-of frequency.

c. Band-pass. Band-pass filters pass frequencies within a given band.

d. Band reject. Band reject filters reject frequencies within a given band.

There are ideal filters, which are nice to talk about, and there are practical filters that we have to use. Practical filters will provide attenuation in the stop-band but will subject the signal to attenuation, ripple, phase-shift, or delay in the pass-band. There are two general guidelines when dealing with practical filters:

a. The closer the practical filter’s amplitude response approximates the ideal filter’s amplitude response, the less linear will be the practical filter’s phase response.

b. Conversely, a practical filter with an optimized linear phase response will have a less than ideal (not flat) amplitude response.

Therefore, we have two major categories of filters; there are filters with optimized amplitude response and there are filters with optimized phase response.

Filters lumped in the amplitude category include Butterworth, Bessel, Chebychev and Cauer or elliptical filters. Butterworth filters exhibit no pass-band ripple but have a roll-off characteristic (expressed in dB/decade of frequency) that is not as steep as other filters.

Chebychev filters do have a steeper roll-off characteristic than Butterworth filters but they do introduce pass-band ripple. For steep roll-off characteristics, elliptical filters are the choice at the expense of pass-band ripple and stop-band ripple. Bessel filters are considered to provide linear phase response and little delay distortion in the pass-band. In the end, the characteristic of the signal to be filtered will determine which type of filter to use.