the capacitor can acquire the signal in 65ns. The low-bias-
current output buffer is then required to settle to only the
resolution (8 bits) of the first (MSB) flash encoder in 65ns,
while an additional 85ns is allowed for settling to the
resolution (14 bits) of the second (LSB) flash encoder.
Sample/Hold droop appears as only an offset error and does
not effect linearity.
THEORY OF OPERATION
The ADC614 is a two-step subranging analog-to-digital
converter. This architecture is shown in Figure 1. The major
system building blocks are: sample/hold amplifier, MSB
flash encoder, DAC and error amplifier, LSB flash encoder,
digital error corrector, and timing circuits. The ADC614
uses hybrid technology with laser-trimmed integrated cir-
cuits mounted in a multilayer ceramic package to integrate
this complex circuit into a complete analog-to-digital con-
verter subsystem with state-of-the-art performance.
Both the MSB and the LSB flash encoder (ADC) functions
are performed by one 8-bit flash encoder. The DAC voltage
reference is also used to generate reference voltages for the
MSB and LSB encoder to compensate drift errors. Buffering
and scaling amplifiers are laser-trimmed to minimize volt-
age offset errors and optimize gain (input full-scale range)
symmetry.
Conceptually, the subranging technique is simple: sample
and hold the input signal, convert to digital with a coarse
ADC, convert back to analog with a coarse-resolution (but
high-accuracy) DAC, subtract this voltage from the S/H
output, amplify this “remainder,” convert to digital with a
second coarse ADC, and combine the digital output from the
first ADC (MSB) with the digital output from the second
ADC (LSB). In practice, however, achieving high conver-
sion speed without sacrificing accuracy is a difficult task.
The subtraction DAC is an ECL 8-bit resolution monolithic
DAC with 14-bit accuracy. Laser-trimmed thin-film nichrome
resistors and high-speed bipolar circuitry allow the DAC
output to settle to 14-bit accuracy in only 35ns.
A “remainder” or coarse conversion-error voltage is gener-
ated by resistively subtracting the DAC output from the
output of the sample/hold amplifier. Before the second
(LSB) conversion, the “remainder” is amplified by a wideband
fast-settling two-input amplifier with a gain of 32V/V. To
prevent overload on large amplitude transients, the active
input is switched off to blank the amplifier input from the
beginning of the S/H acquisition time to the end of the MSB
encoder update time.
The analog input signal is sampled by a high-speed sample/
hold amplifier with low distortion, fast acquisition time and
very low aperture uncertainty (jitter). A diode bridge sam-
pling switch is used to achieve an acceptable compromise
between speed and accuracy. The diode bridge switching
transients are buffered from the analog input by a high input
impedance buffer amplifier. Since the hold capacitor does
not appear in the feedback of the diode bridge output buffer,
R Strobe
L Strobe
M Strobe
Data
Valid
Convert
Command
T/H
Command
8-Bit
Reg
Timing
Digital
Output
DAC
Latch
MUX
8-Bit
Reg
Encoder
Strobe
Amp
Enable
T/H
Output
Error
Amp
+
Signal
Input
Digital/
Analog
+
–
–
Converter
+
–
X16
Reference
Track/Hold
A/D Converter
Gain
Adjust
FIGURE 1. ADC614 Block Diagram—A Two-Step Subranging Architecture.
7
®
ADC614