Chapter III – Measurement systems with electrical signals

- Signal attenuation It happens that the output voltage of one component in the system is very high compared to the voltage limit of the next component. In this case, attenuation of the signal is required. For this purpose, a voltage-dividing network can be used.

Figure 3.10. Attenuation using dividing network.

Vo = Vi

R2 R1 + R2

However, such network can lead to input and/or output loading problems. Example (textbook p. 49) The voltage used to power a heater in an experiment is nominally 120 V. To record this voltage, it must first be attenuated using a voltage divider. The attenuator reduces the voltage by a factor of 15 and the sum of the resistors R1 and R2 is 1000 Ω. a- Find R1 and R2 and the ideal voltage output (neglecting loading effects). b- If the source resistance Rs is 1 Ω, find the actual divider output Vo and the resulting loading error in Vo. c- If the divider output is connected to a recorder that has an input impedance of 5000 Ω, what will be the voltage output (input to the recorder) and the resulting loading error.

Instrumentation and Measurements \ LK\ 2009

26

Chapter III – Measurement systems with electrical signals

Instrumentation and Measurements \ LK\ 2009

27

Chapter III – Measurement systems with electrical signals

- Signal filtering A real signal includes several frequencies. However, some of them are simply noise and others are non-physical frequencies (i.e. frequencies that normally do not exist in the physical phenomena; Exp: under normal conditions, the cardiovascular system does not include frequencies higher than 12 Hz). It is important, therefore, to try to remove undesirable frequencies: this is performed through filtering. Filtering Spurious noise suppression

Aliasing suppression

- Categories of filters

Figure 3.11. Different categories of filters (a) lowpass; (b) highpass;(c) bandpass; (d) bandstop.

- Class of filters - Butterworth - Chebyshev - Elliptic - Bessel - Butterworth filters The main characteristic of Butterworth filters is that they maintain the gain maximally constant in the passband. For low passband Butterworth filters, the gain as a function of frequency can be written as:

Instrumentation and Measurements \ LK\ 2009

28

Chapter III – Measurement systems with electrical signals

1

G=

⎛ f ⎞ 1 + ⎜⎜ ⎟⎟ ⎝ fc ⎠

2n

fc is the corner frequency n is the order of the filter. Note:

f >> 1 if fc For n=1

n

⎛ f ⎞ G → ⎜⎜ ⎟⎟ ⎝ fc ⎠ → f double → G halve

In general the decrease in the stopband is 6n dB/octave.

Figure 3.12. Butterworth filter.

Higher order Butterworth filters do not have a large change in slope after fc. To solve this problem, a Chebychev filter can be used. - Chebychev filter Chebychev filters have a large change in slope after fc, however, the price to pay to achieve this characteristic is a less constant gain in the passband region. G=

1 ⎛ f ⎞ 1 + ε 2Tn2 ⎜⎜ ⎟⎟ ⎝ fc ⎠

ε is the ripple factor. Tn is a Chebyshev polynomial of order n. The transition between pass and stop bands can be improved even better with an elliptic filter. However, this filter induces ripples in both pass and stop band regions.

Instrumentation and Measurements \ LK\ 2009

29

Chapter III – Measurement systems with electrical signals

Figure 3.13. Chebychev filter.

Introducing a filter in the system does not only affect the gain, it also affects the phase, introducing sometimes a very significant phase-shift. This phase shift depends also on the order of the filter. A very efficient class of filters in limiting phase shift is Bessel filters (linear relationship between phase angle and f/fc). However, the change in slope after fc is less steep.

Figure 3.14. Bessel filter. (left) phase angle; (right) gain.

- Selecting a filter In order to get the appropriate filter, you will have to specify:

-

The type: lowpass, highpass, bandpass, … The class: Butterworth, Chebychev, … The order: n

Instrumentation and Measurements \ LK\ 2009

30

Chapter III – Measurement systems with electrical signals

-

The corner frequency (ies)

For Chebychev and elliptic filters other parameters describing the ripple must be given. - Building a Butterworth filter using op-amp Op-amp can be used to construct active filters. The use of op-amp avoids using inductors (not ideal components).

Figure 3.15. Lowpass Butterworth using op-amp. (left) sketch; (right) frequency response.

- Determination of the frequency response of a lowpass Butterworth filter using opamp: We will feed the circuit with a single frequency sine wave signal ( Vmi e j 2πft ). Where Vmi is

the amplitude of the signal and f is its frequency. Since we are in the frequency domain, the gain will be defined using impedance instead of resistance: Z G=− 2 Z1 Z1 consist of R1 Z2

consist of R2 and C in //:

R2 1 + j 2 π fCR

2

1 R2 R1 1 + j 2πfCR2 R Let us put: G0 = − 2 , representing the gain at low frequency for the simple inverting opR1 amp (remove C), then: Then, G = −

G = G0 And

1 V = o 1 + j 2 π fCR 2 Vi G 1 1 = = 2 G0 1 + j 2πfCR2 1 + (2πfCR2 )

Instrumentation and Measurements \ LK\ 2009

31

Chapter III – Measurement systems with electrical signals

To get the corner frequency of this filter, we need to remember that fc is defined as the frequency where the gain decreased by 3 dB (or a fraction of 0.707). Then;

G 0.707G0 1 = = 2 G0 G0 1 + (2πf cCR2 )

1 2πCR2 Knowing the cutoff frequency, we can determine the roll-off rate for frequencies higher then fc. At higher frequencies 2πf c CR2 >> 1; then: Giving: f c =

G 1 = G0 2πf cCR2

Now if the frequency is doubled (1 octave) the ratio

G is divided by 2 which G0

correspond to -6 dB. - Phase shift To get phase shift induced by the lowpass Butterworth filter using an op-amp, we have to write the gain in terms of fc: 1 G = G0 ⎛ f ⎞ 1 + j ⎜⎜ ⎟⎟ ⎝ fc ⎠ ⎛ f ⎞ ⎟⎟ ⎝ fc ⎠

φ = − tan −1 ⎜⎜ In the same way; 1st order highpass filter:

fc =

1 2πR1C

Figure 3.16. Highpass Butterworth using op-amp. (left) sketch; (right) frequency response.

Instrumentation and Measurements \ LK\ 2009

32

Chapter III – Measurement systems with electrical signals

1st order bandpass filter: f c1 =

1 2πR1C1

fc2 =

1 2πR2C2

Figure 3.17. Bandpass Butterworth using op-amp. (left) sketch; (right) frequency response.

NOTE: low order filters suffer from insufficient attenuation. To construct higher order filters, a cascade of lower order filters can be used. Example A transducer measures a sinusoidal signal with an amplitude of ±5 V and a frequency up to 10 Hz. Superimposed on this signal is 60 Hz noise with an amplitude of 0.1 V. It is desired to attenuate the 60 Hz signal to less than 10% of its value using a Butterworth filter. Select a filter order to perform this task if the corner frequency is 10 Hz. - Circuits for integration differentiation and comparison t

Vo (t ) = −

1 Vi (t )dt + V0 (0) RC ∫0

Vo (t ) = − RC

dVi (t ) dt

Figure 3.18. (left) integrator; (right) differentiator.

Instrumentation and Measurements \ LK\ 2009

33