Reduction Of Jitter; Filtering; Digital- / Analogue Conversion - AVM EVOLUTION CD 3.2 Operating Instructions Manual

The characteristic of this noise is that it has an energy which depends on the resolution used to quantize the original
signal and which is continuously spread over the whole range of the sampling frequency bandwidth. It is obvious that this
noise can mask fine details of the originally recorded music.
For physical reasons it is not possible to avoid quantization noise. Also a reduction of the total noise energy is not
possible because the noise has been created when the signal was recorded. An elegant solution of this problem is to
increase sampling frequency when re-converting the signal from digital to analogue. The upsampling converter installed
in the CD3.2 / 5.2 can increase sampling frequency from 44,1 kHz up to 192 kHz.
When re-converting the upsampled signal the upsampling converter produces the same amount of noise energy as a
conventional converter.
The difference is that the noise energy is spread over a much broader frequency band. So the part of noise energy which
is within the audible spectrum decreases. You can imagine that like if You have a certain volume of fluid in a small glass.
If You fill the fluid in a glass which has much more diameter the quantity of fluid doesn't change but the level will be lower
than in the small glass. In the same way the increasing of sampling frequency (called upsampling) broadens the noise
bandwidth and reduces the noise level. Most of the noise energy now is located in a frequency region beyond the audible
range and can easily be filtered out without affecting the music signal.

1.5.2 Reduction of jitter

Jitter means slight, varying deviations in the sampling frequency of a digital signal. These deviations come from
deviations in speed of the CD when it is played back (a natural effect, which can be reduced by mechanical means, but
never fully eliminated). They can additionally come from electronic circuits through which the signal must pass. When
such a signal is converted to analogue the samples arrive sometimes a little bit too early, sometimes a little bit too late at
the DAC. This leads to modulations in the analogue signal which can affect the quality of the reproduced music. The
spatial image is not precise, You cannot exactly locate the instruments, the sound is a bit roughened.
The solution for this problem is upsampling. Upsampling does not only mean multiplying of sampling frequency by a fixed
factor like it is done by the oversampling technique used in former times. Upsampling technique is more similar to
recording the original digital signal anew with a different sampling frequency (re-clocking). That means that the sampling
frequency of the original signal and the upsampled signal are fully independent of each other. Thus if the upsampling
converter has a stable jitter free clock the upsampled signal contains less jitter than the original digital signal.
The musical advantages of re-clocking are the second reason why the AVM CD3.2 / 5.2 are equipped with a brandnew
upsampling circuitry and an additional stable oscillator circuit.

1.5.3 Filtering

If a digital signal is converted to analogue the analogue signal contains not only the original signal, but as well it's mirror
image which lies in the frequency domain beyond one half of the sampling frequency. This mirror image (aliasing) can
cause unwanted interferences with the original signal and thus must be filtered out before passing the signal to the
amplifier.
If the original sampling rate of 44,1 kHz is used the filter slope must be positioned somewhat above 20 kHz and has to be
very sharp in order to let the audio signal pass and to eliminate the aliasing components. Such filters cause a large
phase deviation at the end of the pass band and have often also amplitude deviations. This leads to a harsh reproduction
of music and can also affect the localization of solo instruments and voices.
Upsampling to higher rates makes it possible to set the filter frequency far out of the audio signal range. For example at
192 kHz sampling rate the filter must take effect at 96 kHz. In this frequency region no music signal is present. Thus the
filter can theoretically not affect musical reproduction.
Anyhow the filter frequency and the gradient of the slope – even if out of normal audio range have some subtle, but
audible influence on the musical reproduction. Therefore the CD3.2 / 5.2 offer You five different filter characteristics. So
you can choose your favorite filter upon your own taste.

1.5.4 Digital- / analogue conversion

The CD3.2 / 5.2 use highly precise 24-bit converters to reproduce the analogue signal out of the digital data. The
converters output balanced signals. These signals are fed into a differential amplifier. The difference between the signals
is twice the audio signal (because one of the signals is inverted) and the difference of the inaccuracies of the converters.
As the two converters per channel are on the same chip, their inaccuracy is nearly the same and thus also nearly
eliminated by the differential amplifier.
The second advantage of this differential technique is that the (very low) individual noise coming from the converters is
reduced by 3 dBs.
The result is a clearly audible advantage in dynamic of the music signal and an audibly improved reproduction of the
finest details.
5