DDC101 Datasheet, PDF (19/29 Pages) Burr-Brown (TI) – 20-BIT ANALOG-TO-DIGITAL CONVERTER

Basic Integration Frequency Response
The sin(x)/x basic integration characteristic is controlled by
the digital filter’s measurement time (T_MEAS). The measure-
ment frequency, f_MEAS is l/T_MEAS. The input frequency re-
sponse of the DDC101 is down –3dB at f_MEAS/2.26 with a
null at f_MEAS. Subsequent nulls are at harmonics 2f_MEAS,
3f_MEAS, 4f_MEAS, etc. as shown in the frequency response curve
below. This characteristic is often used to eliminate known
interference by setting f_MEAS or a harmonic to exactly the
frequency of the interference. Table VI illustrates the fre-
quency characteristics of the DDC101 integration function
for various measurement times. As an example, for N =
2272, K = 16, and M = 256: T_MEAS = (N-M-K)/f_CLK = (2272-
256-16)/2MHz = 1ms and f_MEAS = 1kHz. T_INT = 2272/2MHz
= 1.14ms; f_CONV = l/T_INT = 880Hz.
MEASUREMENT TIME
100µs
1ms
10ms
16.66ms
20ms
–3dB FREQUENCY
4.42kHz
442Hz
44.2Hz
26.5Hz
22.1Hz
f
MEAS
10kHz
1kHz
100Hz
60Hz
50Hz
TABLE VI. Basic Integration Frequency Response Examples.
Oversampling Frequency Response
The M oversamples of the initial and the final data points
create an oversampling sin(x)/x type of low pass filter
response. The oversampling function reduces broadband
noise of the input signal and the DDC101. Broadband noise
is reduced approximately in proportion to the square root of
the number of oversamples, M. As an example, a conversion
with 128 oversamples will have approximately 1/2 the noise
of a conversion with 32 oversamples (√32/128 = √1/4 =
1/2) The oversampling low pass filter response creates a null
0
–10
Nyquist
(f_CONV/2)
–20
–30
–20dB/decade
Slope
–40
–50
0.1f_MEAS
f_CONV
f_MEAS
Frequency
10f_MEAS
FIGURE 16. Basic Integration Frequency Response.
at f_OS = 1/T_OS. The oversample time, T_OS, is M/f_CLK. For M =
256 and f_CLK = 2MHz, f_OS is approximately 7.8kHz. Subse-
quent nulls are at harmonics 2f_OS, 3f_OS, 4f_OS, etc. The –3dB
point is at f_OS/2.26. Table VII illustrates the DDC101
oversampling frequency characteristics with approximate
values for f_OS and the –3dB frequency. An oversampling
frequency response graph is shown below in Figure 17. This
figure shows the frequency response for M = 256 oversamples
with an f_CLK of 2MHz . The slope of the attenuation curve
decreases at approximately 20dB/decade.
OVERSAMPLES (M)
256
128
64
16
–3dB FREQUENCY
3.5kHz
6.9kHz
13.9kHz
55kHz
f_OS
7.8kHz
15.6kHz
31.2kHz
125kHz
TABLE VII. Oversample Frequency Response Examples.
Normalized DDC101 Frequency Response
The normalized frequency response, H(f), of the DDC101 that is applied to the input signal consists of the product of the three
frequency response components:
( ( ) ( ₍ )) ( ₍ )) _H(f)
=
sin πf N − M − K)/ f _CLK
πf(N − M − K)/ f _CLK
• ^{sin πfM/ f CLK}
Msin πf / f _CLK
• ^{sin πfLN/ f CLK}
Lsin πfN/ f _CLK
• e^{− jπf (LN − K −1)/ f CLK}
Where:
Basic Integration
Oversampling
Multiple Integrations
Linear Phase
f
is the signal frequency
f_CLK
is the system clock frequency, typically 2MHz
N
is the total number of clock periods in each integration time, T_INT = N/f_CLK, T_INT is the DDC101 CDAC's
integration time
M
is the number of oversamples in one oversampled data point
K
is the number of clocks used in the acquisition time
(N-M-K)/f_CLK is the digital filters measurement time, T_MEAS, (T_MEAS = T_INT –(M+K)/f_CLK)
M/f_CLK
is the oversample time, T_OS
LN/f_CLK
is the total conversion time for multiple integrations, T_CONV
The DDC101's transfer response has a linear phase characteristic as indicated by the exponential term.
®
19
DDC101