SM-SX100
– 12 –
INTRODUCTION OF CIRCUIT OUTLINE
1. Technical background
1-1) Concept of coding technology
Though the PCM system is the most popular method to convert the analog audio signals into the digital codes, the basic
concept of the PCM system is that the signals are sampled with the frequencies whose band is two or more times of the
frequency band to be transferred and is quantized into the multi-bits.
In the other words, "sampling frequency Fs" and "quantized bit quantity" determine the frequency band (Fs/2) and the
dynamic range respectively for the information to be encoded.
As another concept against the PCM encoding which principally determine the transfer range, Shannon *1) establishes
the theory of the coding system which sets the transfer range with "average information quantity per unit time" from the
viewpoint of the information theory.
On method of this system is "high-speed sampling 1-bit coding" system which uses the
modulation. The quantized
bit quantity has only 2 values of 1 bit but the sampling frequency is sufficiently increased to make it possible to gain the
transfer range which assures the dynamic range.
As two systems are compared with each other, it is said that the PCM system determines the dynamic range with the
quantized bit quantity, "voltage dissolution power" but the 1-bit coding system" increases time separation power" to assure
the purposed dynamic range.
On the PCM signal, when the quantized bit quantity is designated b, "signal to noise ratio (S/N)" of the quantized noise
to the signal is expressed as follows.
(S/N) dB = 6.02b + 1.76 dB
It shows that the ratio of signal to noise is improved in proportion to the increase of the quantized bit quantity.
When the quantized bit quantity is 16 and the signal is the sine wave, the ratio of signal to noise is practically gained as
follows.
(S/N) dB = 6.02 X 16 + 1.76 = 98.08 dB
On the other hand, Fig. 12-1 shows if "7th order feedback high-speed 1-bit quantizer" is used, the quantized noise
distribution can be controlled by timely setting the part feedback coefficient b1 to b3 in the algorithm.
Fig. 12-2 shows that the quantized noise monotonously increases toward the high range in the quantized noise distribution
in case of b1 to b3 = 0. Figs. 12-3 and 12-4 show the same sampling frequency as shown in Fig. 12-2 but show respectively
the quantized noise distributions which gain "wide D range" and and "wide frequency range" respectively by changing the
values of b1 to b3.
As described above, even on the 1-bit signal of the same sampling frequency, "D range" and "frequency band" of the
transferred signal can be selected at the degree of freedom with the design of the the above algorithm coefficient.
Though the 1-bit signal which is used at SACD and 1-bit amplifier is the 1-bit signal of "64fs", it is designed with the algorithm
which can assure the wide transfer band (D range: 120 dB (<20 Hz), Frequency band: 100 kHz).
*1) C.E.Shannon, "A mathematical theory of communication, "Bel Syst. Tech.J.27, (1948).
Figure 12-1
Input
Multiplier
Adder
Delay unit
Quantizer
Output
Quantization
Figure 12-2
Figure 12-3
Figure 12-4
0dB
–50
–100
–150
100
1k
10k
20k
100kMz
0dB
–50
–100
–150
100
1k
10k
20k
100kMz
Wide D range
0dB
–50
–100
–150
100
1k
10k
20k
100kMz
Wide frequency band
