Of fyp and sigma delta modulator

Now that I am in my final year, one of the hot topics is none other than the final year project. Some of the conversations during gatherings, be it with ex-classmates, platoon mates or current school mates, have been centered around the opening of asking what each other’s FYP is about. More often than not, the responses from the other party usually make not much sense to me, unless of course the person is doing something general or something related to my area of specialization. Similarly, when I tell people I am doing a design of sigma delta modulator for my FYP, they go like wth is that?

So, I shall now present a brief overview of what sigma delta modulator is all about, as laymen as I can get. That is of course if you are even slightly interested in what I am doing, if not then just treat the rest of this entry as just some gibberish.

Now, in the world of electronics, there are 2 kinds of signals in general. They are either analog or digital. The signals from the outside world, sound for example, are analog in nature. These are the signals humans can understand. Yet, the computer or other microprocessor cannot. In order to process the analog data from the outside world, an interface is needed to bridge the 2 signals of different nature together. This is done using the Analog to Digital Converter (ADC).

How the ADC works is really, to put it very simply, approximating the analog data to a corresponding discrete level determined by the resolution. Due to this approximation, there will be some errors as the final digital output does not equal to the analog equivalent exactly. This error, or quantization noise, is inevitable but can be minimized. A sigma delta modulator is one such architecture of ADC to attempt to reduce this noise as much as possible.

The basic idea of a sigma delta modulator is to use the techniques of oversampling as well as noise shaping.

Oversampling is sampling at a frequency more than the nyquist rate, thereby reducing the amplitude of the quantization noise spectral density and spreading it over a wider range of frequency band while the total noise power is unaffected. Thus, by applying appropriate filters, the quantization noise out of the band of interest can be reduced.

Noise shaping further enhances the performance by means of filtering the quantization noise in the band of interest. This can be done by making a distinction between the noise and the input signal by means of a feedback loop filtering technique. Hence, by designing appropriate filters, the quantization noise inside the band of interest can also be reduced.

Improving the quantization noise is in essence improving the resolution of the ADC, and this is the objective of my project.

My FYP will start off with system level design using MATLAB, to design and determine the filter transfer function, order and coefficient and stuff, and also the oversampling ratio. After optimizing at the system level, I will then have to move into schematic level using the cadence ic design tools, where the modulator will be realized in transistors' level circuits. The technology that i will be using is 0.18um CMOS process technology. And if time allows, the layout design of the circuit will be implemented as well.

Now thats what i will be doing for the next 2 semesters.