Week 5 - PCM, analog-digital conversion Flashcards
What is digitization?
Converting analog signal to digital data.
PCM is a common method.
What are the 3 main elements of Pulse Code Modulation (PCM)?
Sampling
Quantizing
Encoding
What are the three sampling methods (PAM - Pulse Amplitude Modulation)?
Ideal Sampling (discrete point)
Natural Sampling (sampling with finite width pulse/high speed switch)
Flat-top Sampling (Similar to natural sampling but the sample value is discrete and held over the duration of pulse).
What is the restriction on sampling rate?
To reproduce the original analog signal, we must sample at x2 the highest frequency.
Why? To capture the peak and trough of each cycle.
f, 2f, 4f Which of these would be oversampling and undersampling? (f = highest frequency of analog wave)
f - under-sampling
4f- over-sampling
2f - required sampling rate
Why can’t we tell what the minimum sampling rate is for a bandpass signal given the bandwidth?
Because the lowest frequency is non-zero, we cannot assume the highest frequency = bandwidth.
Sampling rate = 2*highest frequency
What is Quantization?
Mapping signal amplitudes to discrete values. (Approximation)
How do we determine quantization levels?
(Max voltage - min voltage) / levels
What is the relationship between normalized amplitude and normalized quantized values?
The amplitudes are divided into quantization levels. Therefore, each quantization level represents a range of amplitudes.
How do we determine normalized quantized values?
What is the relationship between normalized quantized values, normalized PAM values, and normalized error?
Say we have 8 levels. Our quantization levels would range from 4 to -4 (to represent a sine wave)
The midpoint of each level would be a normalized quantized value.
Any normalized PAM values within that level will assume the normalized quantized value.
The difference in value between the two is the normalized error.
What is a quantization code?
Refers to each quantization level starting from 0.
How is quantization error related to the number of levels?
More levels means less error and more accurate signal recovery.
Quantization error changes the SNR of a signal (T or F)
True
What is the issue with uniform quantization?
Quantization levels are evenly spaced, this may not be practical for music for example where amplitudes are concentrated near zero.
What is PCM Encoding?
Converting each quantization level into an encoded word (1010)