ELEC 4123
Design Tasks: Signal Processing, 2021-T1
created by Prof. D. Taubman
Signal Processing Topic: Design Tasks Preliminary Notes
The Signal Processing Design topic in this subject is spread over 2 weeks, with 4 formal laboratory sessions across Weeks 1 and 2. This handout introduces all 5 design tasks within this topic.
Your performance in these design tasks will be assessed in the laboratory, whether you are participant in person on online. The laboratory component of your assessment for the electronics design tasks as a whole is worth 20% of the marks for the course. As indicated in the course overview, this is composed of 8% for design outcomes and 12% for understanding. To simplify the awarding of marks, these percentages are multiplied by a factor of 5. This means that there are a total of 40 marks for design outcomes in electronics (your “O mark”) and 60 marks for design principles and design understanding in electronics (your “U mark”).
An individual breakdown of the maximum marks available for each task is provided in the following, with marks contributing to the “O mark” and “U mark” components identified separately. These individual figures may add up to more than the respective “O mark” and “U mark” totals of 40 and 60 respectively. Where this happens, your “O mark” is capped at 40 and your “U mark” is separately capped at 60, but we will retain all of the uncapped original marks, which might be used to offset unexpected poor performance in later design topics. This is done deliberately to provide a means for you to recover from any initial setback in your O or U marks by putting extra effort into other tasks.
Working in Pairs and Communication in the Lab
As described for the Electronics Design Topic in Weeks 1-2.
Assessment Procedure
By and large, the assessment procedures are the same as those described for the Electronics Design Topic in Weeks 1-2. However, much of this topic is done using Matlab and the computer.
Students who are not able to physically enter the lab can develop solutions to computer-based tasks largely independently, although the solutions to some tasks still need to be tested within the laboratory.
Working together with your lab partner, you should ensure that all computer-based solutions can all be demonstrated using the lab computers, even if they are developed on your own computer either within or outside the lab.
Signal Processing Design Task 1
In this design task, you are to design, build and test and analog band-pass filter (an electronic circuit) whose gain is within ±1.5dB of unity within the passband, which extends from fL to fH and is at most -17dB at frequencies smaller than fL/10 or larger than 10fH. The values of fL and fH will be given to you at the start of the laboratory.
It is important that you make sure you understand the requirements for this simple design task very well before you get to the laboratory.
For this particular design task, your soft objective is to design a solution which is as robust as possible to manufacturing variations associated with the electronic components you use. You should be able to comment on the impact of variations in capacitor and resistor values, within their respective tolerances, and why your design should meet the requirements in spite of these variations.
Available Electronic Components (on hand with your lab demonstrator):
Transistors: BC549, BC559
Analog IC’s: LM324, LM348, LM741
Resistors and capacitors, as found in the laboratories
Assessment for this task:
Marks for this task are as follows:
• O: Achievement of requirements: (____/8)
• U: Thorough understanding of the design: (____/8)
• U: Able to show why requirements are met over tolerance range: (____/4)
Signal Processing Design Task 2
In this design task, you are to design and implement an analog filter which is able to extract the third harmonic from a waveform whose frequency f and shape (square or triangular) will be given to you at the start of the laboratory session. Your design is intended to produce a sinusoid whose frequency is 3f and whose total harmonic distortion (THD) is smaller than X dB, where the value of X will be given to you at the start of the laboratory session. The amplitude of the output signal from your filter should vary by no more than ±50% as the frequency f varies by ±10% of its stated value. You can assume ahead of time that f will be no smaller than 1kHz and no larger than 10kHz.
NB: Students who are not physically able to come into the lab would do well to implement this task themselves, since it is a building block for Task 3 and will allow you to perform other tasks in this design topic without the need for laboratory test equipment – all you will need is a battery-powered analog circuit and a PC or laptop with line-input audio and Matlab.
With this in mind, if you are located physically overseas, you are strongly recommended to purchase the components for these tasks online ASAP, along with a couple of breadboards, some hookup wire and a selection of common resistors and ceramic capacitors. This is not a requirement for the course – it is just a suggestion that could improve your experience.
It should be apparent to you that both the shape of the waveform and the value of X are parameters which may affect the fundamental structure of the design. You should make sure you understand what exactly is meant by THD and you should also think about how you might confirm that your design meets the requirements. You are strongly encouraged to discuss such matters during your tutorial session, but be sure to do your own research and thinking beforehand and be sure also to research any ideas which come up during tutorials before going to the laboratory.
For this task, your soft design objective is to come up with a simpler solution, wherever this is possible. By simple, we mean that fewer electronic components are preferred.
Available Electronic Components (on hand with your lab demonstrator):
Transistors: BC549, BC559
Analog IC’s: LM324, LM348, LM741
Resistors and capacitors, as found in the laboratories
Assessment for this task:
Marks for this task are as follows:
• O: Achievement of requirements: (____/10)
• U: Design elegance (i.e., simplicity): (____/2)
• U: Thorough understanding of the design: (____/12)
Signal Processing Design Task 3
For this task you will need to build a small electret microphone amplifier on your breadboard, along with a bandpass filter. Design parameters are chosen so that you can re-use the bandpass filter from Task 2, so as to save time. In particular, the input content you are interested in will have frequencies concentrated around the 3f value from Task 2.
NB: Students who are not physically able to come into the lab would do well to implement this task themselves, as explained earlier for Task 2. In addition to the components you will need a small soldering iron and flux-core solder.
Following the datasheet provided on the course web-site, you can construct an electret microphone amplifier by tying the B- terminal of the microphone to ground, and attaching a 1kW resistor between the B+ terminal and VCC=6V (or anywhere around there). Pass the B+ terminal through a bypass capacitor to a simple opamp buffer with 10kW input impedance. The output from the amplifier should pass to the bandpass filter from Task 2. Please note that you will need to solder wires to the electret microphones provided. This should be done using the soldering workstations in your laboratory. Take care not to leave the soldering iron in contact with the terminals of the microphone for more than a couple of seconds, so as to avoid destroying the device.
The output from your bandpass filter should be passed to the audio input of the laboratory computers (or your home computer) via a small protection circuit – you can insert your own opamp buffer first if you feel the need to do this, to isolate your filter from the load applied by the protection circuit and cabling that go to the computer.
The protection circuit which you need to construct uses 4 diodes and a 1kW resistor to limit the voltage which can be passed to the computer’s audio-in jack to around ±1.3V in the worst case. Make sure you know how to do this – discussed in the Week 3 lecture.
When working in the lab it is a good idea to measure all output signals using the CRO first to ensure that no out-of-range signals can be injected into the computer.
The labs are equipped with breakout box that can be (or is) connected to the computer’s audio-in and audio-out jacks and also to your breadboard, but if you are doing this yourself at home you may need to buy a cable with a 3mm audio jack and strip the wires at one end of the cable to connect to your protection circuit.
Use Matlab’s “audiorecorder” function to record the waveform at 48kHz. The audio input for this task comes from a simple HTML sound source (a web page) that plays a sound from any web browser. You can play it on your phone or another computer. The HTML sound source will be provided via Moodle.
The objective of this task is to determine the frequency of the audio signal from the sampled audio waveform, reporting this as a numerical output from a digital signal processing algorithm. Please note that there are only a limited number of frequencies that can be produced via the HTML sound source and this may help you develop an elegant solution.
The principle soft objective here is to design a computationally solution. For this, you will need to be able to comment on the numerical complexity of a solution rather and
justify your approach using something other than just running time under Matlab – i.e., you should have fundamental reasons for believing your solution is good.
Available Electronic Components (on hand with your lab demonstrator):
Analog IC’s: LM324, LM348, LM741
Diodes: 1n4148
Microphones: AM4010 electret, or equivalent Resistors and capacitors, as found in the laboratories
Assessment for this task:
Marks for this task are as follows:
• O: Achievement of requirements: (____/10)
• U: Understanding of the analog design: (____/4)
• U: Understanding of algorithm along with complexity issues: (____/10)
Signal Processing Design Task 4
This task re-uses the analog circuit from Task 3, but has a different objective. In this task, the audio input still comes from the HTML file and you still sample the received signals using Matlab’s “audiorecorder” function at 48kHz. However, the goal is to sub- sample and then reconstruct the 48kHz signal at a rate R that is no larger than a given bound Rmax that will be given to you in Week 2.
The main objective of the task is to reconstruct a high quality replica of the original waveform. You can compare the reconstructed and input waveforms at 48kHz numerically, computing the MSE (Mean Squared Error) distortion D and the power (average squared amplitude) P of the original input waveform, reporting the signal-to- noise ratio (SNR) P/D in dB as your fidelity metric. At least 30dB is expected, but much higher values should be achievable.
The soft objective for this task is to make R as small as possible. In particular, you can get significant bonus marks for coming up with a working solution in which R is smaller than the largest frequency produced by the HTML sound source. You should immediately recognize that this involves aliasing and that your reconstruction would need to remove the aliasing, using knowledge of the limited bandwidth of the sound source and your analog bandpass filter.
Available Electronic Components (on hand with your lab demonstrator):
Nothing new – you are re-using the analog design from Task 3
Assessment for this task:
Marks for this task are as follows:
• O: Achievement of requirements: (____/8)
• U: Thorough understanding of reconstruction for R > Nyq Rate: (____/8)
• U: Demonstration & understanding of sub-Nyquist reconstruction: (____/6)
Marking for this task cannot commence until Week 2
Signal Processing Design Task 5
The setup for this task is the same as Tasks 3 and 4, except that you will need two microphones, each with its own amplifier, bandpass filter and protection circuit.
If you were not able to get the bandpass filter of Task 3 implemented or it is too burdensome to construct two identical analog circuits, you are allowed to undertake Task 5 using only an opamp amplifier (no analog filter), but in this case your solution may need additional digitial signal processing.
The design objective for this task is to locate the direction of arrival of the HTML sound source (the “target”) using the stereo microphone input. You should be aware of the fact that the frequency of the target may be any of the values achievable by the HTML sound source. Your design is to be implemented in Matlab using digital signal processing. Specifically, your solution should print estimates of the direction of the sound source (over the range -90 to +90 degrees) at intervals of approximately 1 second, relative to the edge of the laboratory bench.
It should be apparent that you will need to carefully select the positions of your microphones, taking into account physical properties of the propagation of sound waves. If you are not already familiar with the concept of correlation, you would do well to revise this important signal processing concept.
For this design task, there are two soft objectives. The first soft objective is to maximize the accuracy with which you determine the angle of the incident sound source. The second soft objective is to be able to robustly detect the location of the target sound source even when the target is at some distance and in the presence of other interfering sources of sound – e.g., other students talking in the laboratory.
Available Electronic Components (on hand with your lab demonstrator):
Analog IC’s: LM324, LM348, LM741
Diodes: 1n4148
Microphones: AM4010 electret, or equivalent Resistors and capacitors, as found in the laboratories
Assessment for this task:
Marks for this task are as follows:
• O: Achievement of requirements: (____/10)
• U: Thorough understanding of the design: (____/8)
• U: Achievement & understanding of features to minimise interference: (____/6)
Marking for this task cannot commence until Week 2