Signal Conditioning Methods

Signal Conditioning with Methods:

The data that will be acquired in a system will not be from identical sources. Therefore signal conditioning is necessary. Signal conditioning may involve some of the following operations.

Attenuation to scale down the input signals may be necessary to match them to the input levels of the converter's full scale range.

Linearization of data or linear approximation alters conversion.
Analog differentiation precision rectification etc

Signal conditioning can be done in two methods for data acquisition systems. They are:

1. Ratiometric conversion
2. Logarithmic compression

The two methods mentioned above are explained here under:

(1) Ratiometric Conversion:

Ratiometric conversion is explained with reference to a transducer having four strain gauges in a Wheatstone bridge network. In such a bridge the output voltage will be a function of the change of resistance of each a ram and the excitation voltage of the bridge. 

Let the strain gauges be under maximum constant unbalance. Now if the excitation voltage changes by ± X %, the output of the bridge also changes by ± x %.

If we can condition the output voltage of the bridge such that the output of the signal amplifier is proportional to the strain only and independent of excitation voltage the system accuracy improves. This is due to the fact that the fluctuations in the excitation voltage do not affect the sensitivity of the system.

In an analog method of obtaining this result an analog divider will be incorporated. To this analog divider the amplifier's output and the excitation voltage are supplied. Now the output of the divider is a ratio of the amplifiers output to the excitation voltage.

Ratiometric Conversion
Another method is illustrated in Figure. In this method the bridge excitation voltage is supplied as an external reference voltage to the analog to digital converter.

In the A/D converter the conversion factor is proportional to the reference voltage. When such an arrangement is made the system sensitivity is independent of the fluctuations in bridge excitation voltage.

(2) Logarithmic Conversion:

Logarithmic conversion circuit permits measurement of fractional changes in the input as a percentage of the input magnitude rather than a percentage of a range.

As an example if we take an input range of 100 µV to 100 mV, the output voltage may correspond to 0 for 100 µV and 3 V for 100 mV, if the logarithm conversion gain is 1% per decade. Consider a change of 1% that is if the input changes from 100 mV to 101 mV. The output of the logarithm amplifier will change by

Δ V = [log 101 mV/100 mV] x 1 V = 4.3 mV

As the output change is related to the ratio of the input, it is clear that the change in output is the same i.e. 4.3 mV. That is whether the input changes from 10.0 mV to 10.1 mV or from 100 µV to 101 mV the change in the output will be only 4.3 mV.

The output of the logarithm amplifier can be converted into digital form. This can be done using a 12 bit BCD counter. Then the resolution of the counter will be 3 V/1000 = 3 mV, for a 3 V full scale. This can be achieved by properly scaling the logarithm amplifier. 

It is possible to monitor and record changes as low as 1 mV, for an input of 100 µV or 10 µV for 1 mV with this resolution of the converter. It is to be noted that in the absence of the logarithm amplifier the resolution would be 100 µV. Therefore we conclude saying that a 100 to 1 improvement is possible using a logarithm amplifier.

The logarithm amplifier can enhance the resolutions at low inputs. However at the high inputs it offers a poor resolution. The logarithm conversion distributes the resolution on a "percentage of reading" basis against a "percentage of full scale" as with A / D conversion. This type of conditioning will be advantageous in systems having an output relationship involving the logarithm of the measured. Further it will be advantageous where a moderate accuracy measurement around 1% is required over a large range of 1:105.

The log function is inherently unipolar. Therefore other types of compression will be used when handling bipolar inputs.

No comments:

Post a Comment