You are on page 1of 8

Comm 320 Midterm Review Aliasing: interference to the original signal resuling from sampling at a rate lower than

the nyquist rate. IE: finmax = 3.3khz and fsample = 5khz -> will result in aliasing due to 2finmax !< fnyquist

PAM (Pulse Amplitude Modulation): y y y Most primitive state of digital information Has sever discrete levels, 1 per sample conductive to multiplexing serveral sources over a single tx line

PAM CCT:

PWM (Pulse Width Modulation): y y y Used extensively in digitally controlled systems waveform represents a variable duty cycle Usually converted from PAM Signal

PWM CCT:
     

  

PPM (Pulse Position Modulation) y Pulses of short duration and equal width and hieght are produced corresponding to FALLING edge of the pwm signal. IE: each pulse is positioned proportionally to the amplitude of a sample with respect to time of sampling PPM has the advantage that the power consumed is much lower than PAM/PWM PPM detection through a RS Latch with a RC filter. S I/P is the sample and R I/P is the PPM signal

y y

PPM CCT (2 555 Timers):

PCM (Pulse Code Modulation): Block Diagram:

PCM Requires the following: 1. 2. 3. 4. Filtering/Bandlimiting (anti-aliasing filter) Sampling Quantizing Coding

Filtering: y y Used in Audio Aliasing This is a band pass filter (IE for voice -> between 300hz and 3.3khz is passed)

Sampling: y Must be at the nyquist rate or higher

Quantizing: y y Analog signal can be resolved/quantized by sampling into 15 distinct levels (2^4 - 1 = 15 incl 0) IE: analog i/p = 6v --> 6v 15 = 0.4v / level Quantizing is the process of resolving samples/signals into discrete levels which can be represented by an n-bit code. It includes the approximation of each sample to the nearest discrete value Two Types: o Linear o Non-Linear

Linear Quantization: y Results when no processing is done to the original signal, except that after sampling the quantized level is converted to n-bit code

Non-Linear: y y y y Requires enhancement for the signal to allow for detection of lower amplitude in the original signal The n-bit code corresponding to both the linear and non-linear quantization Quantization Error (Qe) is the difference between quantized levels and actual sample values Maximum quantization error = 1/2 LSB

Performance Parameters: y y y Qe -> 1/2 LSB (Resolution/Step size) Dynamic Range (DR) -> 20 log (2n - 1) where n is the number of bits into which each sample is resolved DR = 20 log (Vmax/Vmin) where Vmax = 2n - 1 and Vmin = LSB

Resolution: The stepsize or LSB -> Its the minimum voltage that can be recognized by the system and converted into binary Dynamic Range Formulas: DRdb = 20 log (Vmax/Vmin) IN GENERAL: Example: Max analog I/P: 4.5kHz Minimum DR: 45db Max Decoded Voltage @ Reciever: +/-2.55V DRdb = 20 log (2n - 1)

DRdb = 20 log 2n = 20n x log2 = 6n (20 x 8 x log 2 = 48.16db)

Sample Rate: fnyquist = 2 x fsample = 4.5kHz x 2 = 9kHz Minimum Bits: 20 log (Vmax/Vmin) = 45db log (Vmax/Vmin) = 45db/20 2.25 = log(Vmax/Vmin) --> 10^2.25 = Vmax/Vmin --> 177.83 = Vmax/Vmin Since Vmax = 2n - 1 -> Vmax => 177.83 + 1 = 2n log (177.83+1) / log 2 = n n = 7.47 Rounding down doesn't meet the DR requirement of 45db (20 log 2^7 = 42.14db < 45db) Rounding up to 8 bits produces the required DR -> 20 log 8^2 = 48.16db > 45db Resolution: Resolution = (Vmax / (2n - 1)) Resolution = (2.55V / 2^8 - 1) Resolution = 0.01V = 10mV Maximum Quantization Error: Qemax = Resolution/2 = LSB / 2 = 10mV/2 = 5mV Companding: COMpression (Tx) and exPANDING (Rx) Two approaches: y y Analog Companding Digital Companding

In both approaches the objective is to make the PCM system more sensitive to low amplitude signals. Standard formula for Companding:
   

Where Vmax = Maximum uncompressed analog input amplitude (in Volts) Where = Parameter used to specify the amount of compression (unitless) Where Vout = Compressed O/P amplitude (Volts) Law used in US & Japan A Law used by ITU-T (International Telecommunications Union; -T specifies Standardization)

Digital Companding: [Sending System] Determine the 8-bit digitally compressed code sent for a sample size of 1.8315V if the resolution is 0.002V. Assume that the ADC is 12-Bit with the MSB as the Sign Bit a) 1.8315 is positive -> Sign Bit = 1 b) Determine the Magnitude: Sample/Resolution = 1.8315v/0.002V = 915.75 Round UP: 916 Binary: s01110010100 -> = 916

c) Qe = 1 - 0.75 (from rounding) -> Qe = 0.25 x Resolution = 0.25 x 0.002 = 0.5mV d) Code: Segment determined from the first 1 found -> S01110010100 matches on chart to Segment 6 Remainder of binary changed to: S01ABCDxxxxx Code that is sent to recieving system: S110ABCD [Receiving System] a) Received Code: S110ABCD -> Translated to S01ABCD10000 (appended automatically) We know ABCD = 1100 -> 12 bit Code becomes: S01110010000 S01110010000 into decimal = 912 Signal level after the Digital to Analog conversion = 912 x LSB = 912 x 2mV = 1.824v b) Signal Loss = Qe + Ce = 0.5mV + [(916-912) x 2mV] = 0.5mV + 8mV = 8.5mV

Digital to Analog Conversion (DAC): y y y y Resolution/Step Size is LSB Max O/P is (2n - 1) x LSB Vanalog Out = # steps x LSB -> binary I/P code x LSB Two Types: o Binary Weighted DACs o R/2R DACs

Binary Weighted DAC:

If Binary I/P = 0001 ->LSB Step Size = Av x 5v -> Av = -1k/8k x 5v = -0.625v

R/2R DAC:

For Bin 1000 (8) -> O/P = -5V -> LSB = -5V/8 = -0.625v

PCM: y y y y y y Codec -> enCOder and DECoder Encode/Decode telephone circuit card found at the central office (CO) Codec consists of Tx/Rx PCM Circuit Combo Card has replaced Codec -> Combo card includes anti-aliasing, filter, receiver, etc PCM Line speed for telephone systems: fsample x N (N = # of bits) Provides voice channel capacity of 8ksps x 8 bits -> 64kbps (64000bits per second)

DPCM (Differential Pulse Code Modulation): y Sends only the difference of successive samples so that the number of bits required to accurately represent the signal is reduced/decreased allowing for increased/enhanced bandwidth (BW)

ADPCM (Adaptive Differential PCM): y y Uses adaptive filter to process adjacent sample differences Normal voice channel is 64kbps -> ADPCM Sends same data at 32kbps if the difference sent is 4 bit code --> 8k samples/sec x 4 bits = 32kbit/s --> 4 bit code consists of 1 sign bit and 3 bit magnitude

DM (Delta Modulation): y y y Does not use/adhere to Nyquist Sends a 1 if the slope is positive and a 0 is the slope is negative uses a clock to determine when bits are sent

Digital Audio Standards/Organizations: y AES/EBU -> Audio Engineering Standards/European Broadcasting Union These support a variety of systems including DAY and CD DAT -> Digital Audio Tape uses 48khz sample rate; CD uses44.1khz SPDIF -> Sony/Philips Digital Interconnection Format Is not compatible with AES/EBU Supports 16 & 20 bit audio, uses same sampling rates

OTHER: A Compressor has = 240 with a maximum I/P range of +/- 2.55v and a minimum I/P range of +/- 0.01v Input Dynamic Range (DRv) = Max I/P Range/ Min I/P Range = 2.55v/0.01v = 255 Input Dynamic Range (DRdb) = 20 log (Max Range/Min Range) = 20 log (2.55/0.01) = 48.13db Output Dynamic Range (DRv) = Voutmax/Voutmin = 2.541/.03083 = 8.24 Output Dynamic Range (DRdb) = 20 log (Voutmax / Voutmin) = 20 log (2.541/0.3083) = 18.32db

Digital Companding Example: Resolution = 0.0025v For Input Voltage = +3.473: Sign Bit = 1 (postive) Magnitude = Vin/Resolution = 3.473v/0.0025v = 1389.2 -> round: 1389 (Binary of 1389 = 10101101101) 12-Bit linear Code: S10101101101 <- segment = S1 (11) and ABCD = 0101 8 Bit Compressed: S111ABCD -> Segment 7 12 Bit Decoded: S1ABCD100000 = S10101100000 Qe = 0.2 x 0.0025 = 0.5mV Compression Error: [10101101101 (1389) 10101100000 (1376)] x 0.0025 = 1101 (13) x 0.0025 = 32.5mV

You might also like