Pulse Modulation Techniques Questions and Answers

Answer: C

Non-uniform spacing, which companding implements, puts fine steps where the signal spends most of its time and coarse steps where it rarely goes, improving the perceived ratio at no cost in bits. Doubling the levels and raising the sampling rate both increase the rate. And a regenerator addresses path noise rather than coder quality. Matching a quantiser to the signal's own statistics is free in bandwidth terms and is the first place to look for an improvement.

Answer: D

Speech has a large dynamic range but the amplitude distribution is heavily weighted towards small values, so fine steps down there are used far more often than fine steps at the top. Its bandwidth is modest, silence suppression is a separate technique, and no signal tolerates aliasing. A coder tuned to the statistics of the particular signal always beats a general-purpose one, and speech's amplitude statistics are unusually well suited to the trade.

Answer: A

A uniform quantiser's step is fixed, so its ratio falls decibel for decibel as the signal falls:

\[ \begin{aligned} \text{at full scale} &: 38 \ \mathrm{dB} \\ \text{40 dB below} &: -2 \ \mathrm{dB} \end{aligned} \]

so it reaches 38 dB only at full amplitude. Companding's achievement is holding the ratio roughly constant across the range instead, which is what a listener actually notices.

Answer: B

The slope limit is the product of step and clock rate:

\[ \begin{aligned} \text{before} &: \delta f_{s} \\ \text{after} &: \tfrac{\delta}{2} \times 2f_{s} = \delta f_{s} \end{aligned} \]

so it is unchanged, while the granular noise improves because the steps are smaller. That is precisely the trade a faster clock buys: the same tracking ability with finer resolution, at the cost of bit rate. Recognising which combination of parameters a limit depends on shows immediately which changes cancel.

Answer: C

Rate per channel times the number of channels:

\[ \begin{aligned} 8000 \times 7 &= 56 \ \mathrm{kbit/s} \\ 16 \times 56 &= 896 \ \mathrm{kbit/s} \end{aligned} \]

giving 896 kbit/s before overhead. The 3 kHz figure is a distractor: the sampling rate is given directly and the signal bandwidth plays no further part. Spotting which of the quoted numbers the calculation does not need is half the work in a question of this shape. Working one numerical case through both schemes side by side fixes the distinction better than any rule of thumb.

Answer: D

Both statements are defensible but the reason does not explain the assertion: a higher rate eases the anti-alias filter and pushes the spectral repeats further apart, but quantisation noise depends on the step size and hence on the word length, not on the rate. Rate and resolution are independent levers on quality, and confusing them leads to spending bit rate where word length was wanted.

Answer: A

Both are true and the second is the reason: identical short pulses mean the average transmitted power is low and independent of the modulation, whereas a width-modulated train's power rises with its average duty cycle. The efficiency is bought with a stricter timing requirement, which is the standard shape of these trades. Where a scheme's average power does not depend on the message, the transmitter can be sized for the average rather than the worst case.

Answer: B

The first two are correct and the third is the reverse of the truth: because the channels take turns rather than coexisting, non-linearity has nothing to mix and the linearity requirement is relaxed. It is frequency division that needs excellent linearity, because intermodulation there puts every channel into every other. The two schemes trade linearity against timing accuracy, and knowing which way round that trade runs is the practical consequence.

Answer: C

The first three are correctly paired and the fourth is not, so three match. The number of channels is limited by the frame length and the slot width, both time quantities; pulse amplitude affects the received signal-to-noise ratio and nothing about channel capacity. Each design parameter governs one property, and pairing them correctly is what lets a specification be traced back to the choice that produced it.

Answer: D

Companding redistributes the eight bits so that small signals get finer steps, which improves the ratio on quiet passages without touching the bit rate, the bandwidth limit or the frame capacity - those follow from the sampling rate and word length alone. Reading a specification table means asking which line each other line depends on, and here the companding line stands apart from the two that set the rate.