SPT
7
2/10/98
SPT7840
TYPICAL INTERFACE CIRCUIT
Very few external components are required to achieve the
stated device performance. Figure 1 shows the typical inter-
face requirements when using the SPT7840 in normal circuit
operation. The following sections provide descriptions of the
major functions and outline critical performance criteria to
consider for achieving the optimal device performance.
Figure 1 - Typical Interface Circuit
SPT7840
VRHF
VRLS
VRLF
VRHS
VIN
CLK
VCAL
DAV
D10
D0
EN
AVDD
AGND
DGND* DVDD
Ref In
(+4 V)
VIN
CLK IN
Enable/Tri-State
(Enable = Active Low)
Interfacing
Logics
+D5
NOTES: 1) FB3 is to be located as closely to the device as possible.
2) There should be no additional connections to the right of FB1 and FB2.
3) All capacitors are 0.1 F surface-mount unless otherwise specified.
4) FB1, FB2 and FB3 are 10 H inductors or ferrite beads.
FB1
FB2
+A5
AGND
+
10 F
+5 V
Analog
+5 V
Analog
RTN
+A5
FB3
DGND
+
10 F
+5 V
Digital
+5 V
Digital
RTN
+D5
*To reduce the possibility of latch-up, avoid
connecting the DGND pins of the ADC to the
digital ground of the system.
POWER SUPPLIES AND GROUNDING
SPT suggests that both the digital and the analog supply
voltages on the SPT7840 be derived from a single analog
supply as shown in figure 1. A separate digital supply should
be used for all interface circuitry. SPT suggests using this
power supply configuration to prevent a possible latch-up
condition on power up.
OPERATING DESCRIPTION
The general architecture for the CMOS ADC is shown in the
block diagram. The design contains eight identical succes-
sive approximation ADC sections, all operating in parallel, a
16-phase clock generator, an 11-bit 8:1 digital output multi-
plexer, correction logic, and a voltage reference generator
which provides common reference levels for each ADC
section.
The high sample rate is achieved by using multiple SAR ADC
sections in parallel, each of which samples the input signal in
sequence. Each ADC uses 16 clock cycles to complete a
conversion. The clock cycles are allocated as follows:
Table II - Clock Cycles
Clock
Operation
1
Reference zero sampling
2
Auto-zero comparison
3
Auto-calibrate comparison
4
Input sample
5-15
11-bit SAR conversion
16
Data transfer
The 16 phase clock, which is derived from the input clock,
synchronizes these events. The timing signals for adjacent
ADC sections are shifted by two clock cycles so that the
analog input is sampled on every other cycle of the input clock
by exactly one ADC section. After 16 clock periods, the timing
cycle repeats. The sample rate for the configuration is one-
half of the clock rate, e.g., for a 20 MHz clock rate, the input
sample rate is 10 MHz. The latency from analog input sample
to the corresponding digital output is 12 clock cycles.
Since only eight comparators are used, a huge power
savings is realized.
The auto-zero operation is done using a closed loop
system that uses multiple samples of the comparator's
response to a reference zero.
The auto-calibrate operation, which calibrates the gain of
the MSB reference and the LSB reference, is also done with
a closed loop system. Multiple samples of the gain error are
integrated to produce a calibration voltage for each ADC
section.
Capacitive displacement currents, which can induce sam-
pling error, are minimized since only one comparator
samples the input during a clock cycle.
The total input capacitance is very low since sections of the
converter which are not sampling the signal are isolated
from the input by transmission gates.
VOLTAGE REFERENCE
The SPT7840 requires the use of a single external voltage
reference for driving the high side of the reference ladder. It
must be within the range of 3 V to 5 V. The lower side of the
ladder is typically tied to AGND (0.0 V), but can be run up to
2.0 V with a second reference. The analog input voltage
range will track the total voltage difference measured be-
tween the ladder sense lines, VRHS and VRLS.
Force and sense taps are provided to ensure accurate and
stable setting of the upper and lower ladder sense line
voltages across part-to-part and temperature variations. By
using the configuration shown in figure 2, offset and gain
errors of less than
±2 LSB can be obtained.