Pulse Modulation Techniques Questions and Answers

Answer: A

The companding law changes how the available levels are distributed without changing how many bits describe them, so the rate is untouched. The other three multiply together to give the aggregate rate exactly. That is what makes companding such an attractive improvement: it is the one quality lever in the list that is free in bandwidth terms, and it is why every telephony standard uses it.

Answer: B

The technique is companding, a contraction of compressing and expanding, and it gives small signals finer effective resolution at no cost in bit rate. Aliasing is the folding of high frequencies by under-sampling. Regeneration is the reconstruction of pulses along a route. And multiplexing is the sharing of one path among channels. Each term belongs to a different stage of the chain, and placing them correctly is most of understanding a coder's block diagram.

Answer: C

This is time division multiplexing, which exploits the idle time sampling leaves behind. Frequency division stacks channels in spectrum instead, code division distinguishes them by orthogonal codes while they share both time and frequency, and wavelength division is the optical form of frequency division. Every multiplexing scheme is named after the resource it divides, which makes the family easy to keep straight. The shared properties explain why these schemes form a family, and the differences explain why more than one exists.

Answer: D

The Nyquist interval is the time between samples at the Nyquist rate, so it is simply that rate's reciprocal - a quarter of a millisecond for a four kilohertz signal. Frame intervals, hold times and guard bands are all separate quantities that happen to be measured in the same units. Where a rate has a standard name, its reciprocal usually does too, and keeping the pair straight avoids a factor that is easy to invert. Checking the units of each quantity catches a wrong formula before the arithmetic begins.

Answer: A

The filter must pass the whole signal band while rejecting the first spectral repeat, so its cut-off sits between the signal's top frequency and half the sampling rate. That window is exactly what oversampling widens, which is why a higher rate makes the filter cheaper. Cutting off above half the rate would admit the repeat, and cutting below the signal band would discard part of the signal. A requirement bracketed from both sides always has a window whose width is itself a design variable.

Answer: B

Sampling well above Nyquist moves half the sampling rate far above the signal band, so the filter has a wide transition region and need not be steep. Word length is a separate matter, quantisation noise per unit bandwidth is unchanged by the rate alone, and the signal's bandwidth is a property of the signal. Buying an easier filter with a faster clock is a trade that became attractive as digital circuitry got cheap and precision analogue filters did not.

Answer: C

A hiss present at all levels and unaffected by them points at quantisation noise, which is generated inside the coder and rides along with the signal. Lost synchronisation would render the speech unintelligible rather than merely hissy. Path noise would be removed by the regenerators unless it were severe enough to cause errors, which sound like clicks. And a low filter cut-off would dull the speech rather than add noise. Asking whether an impairment varies with signal level separates coder faults from channel faults at once.

Answer: D

Clicks that track path quality are bit errors: a corrupted bit changes a sample's value abruptly and the decoder reproduces that jump faithfully. Quantisation noise is a steady hiss unaffected by the path. Slope overload distorts loud, fast passages rather than producing isolated clicks. And aliasing produces spurious tones. Whether a symptom follows the path or follows the signal is the single most informative observation about a digital link. The parameter that varies and the parameter that does not are worth naming before any substitution is attempted.

Answer: A

Everything wrong and nothing missing is the signature of a slot count offset: the receiver is decoding each slot as its neighbour. A failed sensor would corrupt one channel only. Narrow filters would dull every reading without misrouting any. And a word length error would scale the values rather than swap them. A fault that affects every channel equally and in the same way almost always lies in the framing rather than in the channels. Most errors on this topic come from carrying a result across from a neighbouring scheme where it does not hold.

Answer: B

Loud and high-pitched together means steep, so the input's slope is exceeding the step size times the clock rate. Granular noise is the opposite failure and shows up on quiet or steady passages. Aliasing would produce spurious tones rather than distortion tracking loudness and pitch. And bit errors would not correlate with the signal at all. The two delta modulator defects can be told apart purely by which passages they spoil, which is why the symptom description is enough to name the cure.