signal (shown in Figure 6(d)) is amplified to produce a full-scale input to the second ADC.
nonlinearity is defined as the deviation of any quantization step in the ADC from q, the
Figure 6. (a) Ideal ADC transfer function; (b) input ramp signal; (c) output (quantized) ramp
18
Subranging ADCs are becoming very popular since they can achieve high-speed
operation with
high resolution. They require far fewer comparators for a given resolution than flash ADCs.
While the internal ADCs within a subranging ADC have traditionally been flash ADCs, other
types of ADCs may be used. For example, a new architecture, the cascaded magnitude amplifier,
has been used in the Analog Devices AD9042 (a 12-bit, 41-Msamples/s subranging ADC). This
architecture provides a very high-speed conversion and greatly reduces the number of
comparators required in the internal ADCs.
The cascaded magnitude amplifier (MA) ADC consists of B−1 MAs in series and a single
comparator placed in series after the last MA. A diagram showing the operation of the cascaded
MA ADC is given in Figure 7. Referring to this figure, the first MA compares the input signal to
a voltage level V
fs
/2 where V
fs
is the full-scale input voltage of the ADC. If the input signal is
greater than V
fs
/2, the bit representing this MA is set to 1. If the input signal is less than V
fs
/2, the
bit representing this MA is set to 0. Therefore, this first MA divides
the full-scale voltage into
two regions and the bit representing this MA is set according to the region in which the input
voltage falls.
The next (second) MA uses the output of the first MA as its input. As shown in Figure 7, the
second MA divides each of the two regions defined by the first MA into two additional regions.
If the input voltage to the first MA is between V
fs
and V
fs
/2, the second MA determines if the
input signal is between V
fs
and 3V
fs
/4 or between 3V
fs
/4 and V
fs
/2. The bit representing this MA
then is set to 1 or 0, respectively. Conversely, if the input voltage to the first MA is between 0
and V
fs
/2, the second MA determines if the input signal is between V
fs
/2 and V
fs
/4 or between
V
fs
/4 and 0. The bit representing this MA then is set to 0 or 1, respectively.
Each subsequent MA (and the final comparator) further subdivides the regions in a similar
manner, providing all of the necessary quantization levels (Figure 7 shows operation up to the
third MA). Because of the way that the bit representing an MA is set (as seen in Figure 7), the
output bits from each MA form a Gray code that represents the input signal voltage. In the Gray
code, only one bit changes in the code word from one quantization level to another.
Determination of the output bits is dependent upon the input signal propagating through the
cascaded MAs only. Very high-speed operation is achieved because the response time of the
amplifiers is very fast.
19
Figure 7. Operation of the cascaded magnitude amplifier ADC.
7V
fs
8
3V
fs
4
V
fs
5V
fs
8
V
fs
2
3V
fs
8
V
fs
4
V
fs
2
V
fs
8
V
fs
4
MA3
3V
fs
4
V
fs
V
fs
4
0
V
fs
2
MA2
bit = 1
bit = 0
bit = 0
bit = 1
bit = 1
bit = 0
bit = 0
bit = 1
bit = 1
bit = 0
bit = 1
bit = 0
0
V
fs
2
0
V
fs
bit = 1
bit = 0
MA1
MA1 = First Magnitude Amplifier
MA2 = Second Magnitude Amplifier
MA3 = Third Magnitude Amplifier
V
fs
2
3V
fs
4
20
Integrating ADCs are another category of converters. They convert the analog input signal
amplitude into a time interval that is measured subsequently. The most popular methods within
this category of ADCs are the dual slope and charge balancing methods. While these types of
ADCs are highly linear and are good at rejecting input noise, they are quite slow.
A relatively new type of ADC is the ΣΔ converter. The first-order ΣΔ converter is the most basic
ΣΔ converter (Figure 8). It consists of a ΣΔ modulator, a digital filter, and a decimator. To
understand how this converter works, an understanding of oversampling, noise shaping, digital
filtering, and decimation is required.
Figure 8. First-order ΣΔ ADC.
The operation of the ΣΔ converter relies upon the effects of oversampling. ΣΔ converters use a
very low-resolution quantizer (typically a 1-bit quantizer) and sample at a rate much greater than
2f
Chia sẻ với bạn bè của bạn: