Difference between revisions of "Aufgaben:Exercise 5.1: Sampling Theorem"
From LNTwww
m (Oezdemir moved page Aufgabe 5.1: Zum Abtasttheorem to Exercise 5.1: Sampling Theorem) |
m (Text replacement - "Time Discrete" to "Discrete-Time") |
||
(6 intermediate revisions by 3 users not shown) | |||
Line 3: | Line 3: | ||
}} | }} | ||
− | [[File:P_ID1126__Sig_A_5_1.png|right|frame| | + | [[File:P_ID1126__Sig_A_5_1.png|right|frame|Sampling of an analog signal x(t)]] |
− | + | Given is an analog signal x(t) according to the sketch: | |
− | * | + | *It is known that this signal does not contain any frequencies higher than BNF=4 kHz. |
− | * | + | *By sampling with the sampling rate fA , the signal xA(t) sketched in red in the diagram is obtained. |
− | * | + | *For signal reconstruction a low-pass filter is used, for whose frequency response applies: |
:$$H(f) = \left\{ \begin{array}{c} 1 \\ | :$$H(f) = \left\{ \begin{array}{c} 1 \\ | ||
Line 18: | Line 18: | ||
\end{array}$$ | \end{array}$$ | ||
− | + | The range between the frequencies f1 and f2>f1 is not relevant for the solution of this task. | |
− | + | The corner frequencies f1 and f2 are to be determined in such a way that the output signal y(t) of the low-pass filter exactly matches the signal x(t) . | |
Line 30: | Line 30: | ||
− | '' | + | ''Hints:'' |
− | * | + | *This task belongs to the chapter [[Signal_Representation/Time_Discrete_Signal_Representation|Discrete-Time Signal Representation]]. |
− | * | + | *There is an interactive applet for the topic dealt with here: [[Applets:Sampling_of_Analog_Signals_and_Signal_Reconstruction|Sampling of Analog Signals and Signal Reconstruction]] |
− | === | + | ===Questions=== |
<quiz display=simple> | <quiz display=simple> | ||
− | { | + | {Determine the underlying sampling rate from the graph. |
|type="{}"} | |type="{}"} | ||
fA = { 10 3% } kHz | fA = { 10 3% } kHz | ||
− | { | + | {At which frequencies does the spectral function XA(f) have <u>no components</u> with certainty? |
|type="[]"} | |type="[]"} | ||
- f=2.5 kHz, | - f=2.5 kHz, | ||
Line 50: | Line 50: | ||
+ f=34.5 kHz. | + f=34.5 kHz. | ||
− | { | + | {What is the minimum size of the lower cut-off frequency f1 that the signal is perfectly reconstructed? |
|type="{}"} | |type="{}"} | ||
f1, min = { 4 3% } kHz | f1, min = { 4 3% } kHz | ||
− | { | + | {What is the maximum size of the upper corner frequency f2 that the signal is perfectly reconstructed? |
|type="{}"} | |type="{}"} | ||
f2, max = { 6 3% } kHz | f2, max = { 6 3% } kHz | ||
Line 61: | Line 61: | ||
</quiz> | </quiz> | ||
− | === | + | ===Solution=== |
{{ML-Kopf}} | {{ML-Kopf}} | ||
− | '''(1)''' | + | '''(1)''' The distance between two adjacent samples is TA=0.1 ms. Thus, for the sampling rate fA=1/TA=10 kHz_is obtained. |
− | [[File:P_ID1127__Sig_A_5_1_b.png|450px|right|frame| | + | [[File:P_ID1127__Sig_A_5_1_b.png|450px|right|frame|Spectrum XA(f) of the sampled signal <br>(schematic representation)]] |
− | '''(2)''' | + | '''(2)''' Proposed <u>solutions 2 and 4</u> are correct: |
− | * | + | *The spectrum XA(f) of the sampled signal is obtained from X(f) by periodic continuation at a distance of fA=10 kHz. |
− | * | + | *From the sketch you can see that XA(f) can have signal parts at f=2.5 kHz and f=6.5 kHz;. |
− | * | + | *In contrast, there are no components at f=5.5 kHz. |
− | * | + | *Also at f=34.5 kHz will be valid XA(f)=0. |
<br clear=all> | <br clear=all> | ||
− | '''(3)''' | + | '''(3)''' It must be ensured that all frequencies of the analog signal are weighted with H(f)=1. |
− | * | + | *From this follows according to the sketch: |
:f1, min=BNF=4 kHz_. | :f1, min=BNF=4 kHz_. | ||
− | '''(4)''' | + | '''(4)''' Likewise, it must be guaranteed that all spectral components of XA(f), that are not contained in X(f) are removed by the low-pass filter. |
− | * | + | *According to the sketch, the following must apply: |
:f_{2, \ \text{max}} = f_{\rm A} – B_{\rm NF} \;\underline{= 6 \ \text{kHz}}. | :f_{2, \ \text{max}} = f_{\rm A} – B_{\rm NF} \;\underline{= 6 \ \text{kHz}}. | ||
Line 86: | Line 86: | ||
__NOEDITSECTION__ | __NOEDITSECTION__ | ||
− | [[Category: | + | [[Category:Signal Representation: Exercises|^5.1 Discrete-Time Signal Representation^]] |
Latest revision as of 11:03, 11 October 2021
Given is an analog signal x(t) according to the sketch:
- It is known that this signal does not contain any frequencies higher than B_{\rm NF} = 4 \ \text{kHz}.
- By sampling with the sampling rate f_{\rm A} , the signal x_{\rm A}(t) sketched in red in the diagram is obtained.
- For signal reconstruction a low-pass filter is used, for whose frequency response applies:
- H(f) = \left\{ \begin{array}{c} 1 \\ 0 \\ \end{array} \right.\quad \begin{array}{*{5}c} {\rm{{\rm{f\ddot{u}r}}}} \\ {\rm{{\rm{f\ddot{u}r}}}} \\ \end{array}\begin{array}{*{5}c} |f| < f_1 \hspace{0.05cm}, \\ |f| > f_2 \hspace{0.05cm} \\ \end{array}
The range between the frequencies f_1 and f_2 > f_1 is not relevant for the solution of this task.
The corner frequencies f_1 and f_2 are to be determined in such a way that the output signal y(t) of the low-pass filter exactly matches the signal x(t) .
Hints:
- This task belongs to the chapter Discrete-Time Signal Representation.
- There is an interactive applet for the topic dealt with here: Sampling of Analog Signals and Signal Reconstruction
Questions
Solution
(1) The distance between two adjacent samples is T_{\rm A} = 0.1 \ \text{ms}. Thus, for the sampling rate f_{\rm A} = 1/ T_{\rm A} \;\underline {= 10 \ \text{kHz}}is obtained.
(2) Proposed solutions 2 and 4 are correct:
- The spectrum X_{\rm A}(f) of the sampled signal is obtained from X(f) by periodic continuation at a distance of f_{\rm A} = 10 \ \text{kHz}.
- From the sketch you can see that X_{\rm A}(f) can have signal parts at f = 2.5 \ \text{kHz} and f = 6.5 \ \text{kHz};.
- In contrast, there are no components at f = 5.5 \ \text{kHz}.
- Also at f = 34.5 \ \text{kHz} will be valid X_{\rm A}(f) = 0.
(3) It must be ensured that all frequencies of the analog signal are weighted with H(f) = 1.
- From this follows according to the sketch:
- f_{1, \ \text{min}} = B_{\rm NF} \;\underline{= 4 \ \text{kHz}}.
(4) Likewise, it must be guaranteed that all spectral components of X_{\rm A}(f), that are not contained in X(f) are removed by the low-pass filter.
- According to the sketch, the following must apply:
- f_{2, \ \text{max}} = f_{\rm A} – B_{\rm NF} \;\underline{= 6 \ \text{kHz}}.