A DM system is designed to operate at 3 times the Nyquist rate for a signal with a 3 kHz bandwidth. The quantizing step size is 250 mV.
(i) Determine the maximum amplitude of a 1 kHz input sinusoid for which the delta modulator does not show slope overload.
(ii) Determine the post-filtered output signal-to-quantizing noise ratio for the signal of part (i).
10 marks · directive: Determine
The core of a good answer
With fₛ = 3 × 6 kHz = 18 kHz, slope overload is avoided for A ≤ δfₛ/(2πfₘ) = 0.25 × 18 000/(2π × 1000) ≈ 0.716 V, and the post filtered SNR is 3fₛ³/(8π²fₘ²W) ≈ 73.9, that is about 18.7 dB.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
The question tests the two noise mechanisms of linear delta modulation: slope overload (signal slope exceeding δfₛ) and granular quantising noise. The method is to find the sampling rate, apply the slope overload condition to a sinusoid, then compute signal power over the in-band quantising noise power after the reconstruction filter.
Answer outline
Introduction
Given W = 3 kHz, Nyquist rate = 2W = 6 kHz, fₛ = 3 × 6 kHz = 18 kHz, step δ = 0.25 V, input m(t) = A cos 2πfₘt with fₘ = 1 kHz.
Step 1: Sampling rate
- Nyquist rate = 2 × 3 kHz = 6 kHz
- fₛ = 3 × 6 = 18 kHz, Tₛ = 55.6 μs
Step 2: Slope overload condition
- Maximum slope of the staircase = δ/Tₛ = δfₛ
- Maximum slope of the input = 2πfₘA
- No slope overload if 2πfₘA ≤ δfₛ
Step 3: Maximum amplitude, part (i)
- A_max = δfₛ/(2πfₘ) = 0.25 × 18 000/(2π × 1000)
- A_max ≈ 0.716 V
Step 4: Signal power
- S = A_max²/2 = 0.7162²/2 ≈ 0.2565 V²
Step 5: Quantising noise after the low pass filter
- Granular error is taken as uniform in (−δ, δ), total power δ²/3 spread roughly uniformly over 0 to fₛ
- In band noise N = (δ²/3)(W/fₛ) = (0.0625/3)(3/18) ≈ 0.003472 V²
Step 6: Output SNR, part (ii)
- SNR = S/N = 0.2565/0.003472 ≈ 73.9
- Equivalent closed form: SNR = 3fₛ³/(8π²fₘ²W) = 3 × 18 000³/(8π² × 1000² × 3000) ≈ 73.9
- SNR ≈ 10 log₁₀ 73.9 ≈ 18.7 dB
Conclusion
(i) A_max ≈ 0.716 V; (ii) post filtered SNR ≈ 73.9 ≈ 18.7 dB. Check: the closed form formula and the direct S/N calculation give the same value. The modest SNR at three times the Nyquist rate shows why DM needs heavy oversampling; SNR rises 9 dB per doubling of fₛ.
Examples and data to cite
- Adaptive delta modulation varies δ to reduce both slope overload and granular noise; ADM was selected as the standard for NASA communications between mission control and spacecraft.
- Doubling fₛ to 36 kHz would raise the SNR by a factor of 8, to about 27.7 dB.
Facts to remember
- Slope overload is avoided when |dm/dt|max ≤ δfₛ; for a sinusoid, 2πfₘA ≤ δfₛ.
- Granular noise power of linear DM ≈ δ²/3, and after a low pass filter of bandwidth W it is (δ²/3)(W/fₛ).
- Output SNR of DM for the maximum sinusoid: 3fₛ³/(8π²fₘ²W).
- Nyquist rate = 2 × bandwidth.
- Numbers recomputed in Python: A_max = 0.71620 V, SNR = 73.86 (18.68 dB).