REV. 0
Information furnished by Analog Devices is believed to be accurate and
reliable. However, no responsibility is assumed by Analog Devices for its
use, nor for any infringements of patents or other rights of third parties
which may result from its use. No license is granted by implication or
otherwise under any patent or patent rights of Analog Devices.
a
AD7894
One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A.
Tel: 781/329-4700 World Wide Web Site: http://www.analog.com
Fax: 781/326-8703 © Analog Devices, Inc., 1998
5 V, 14-Bit Serial, 5 s
ADC in SO-8 Package
FUNCTIONAL BLOCK DIAGRAM
SIGNAL
SCALING*
TRACK/
HOLD
14-BIT
ADC
AD7894
OUTPUT
REGISTER
REF IN VDD
VIN
CONVST
GND BUSY SCLK SDATA
*AD7894-10, AD7894-3
FEATURES
Fast 14-Bit ADC with 5 s Conversion Time
8-Lead SOIC Package
Single 5 V Supply Operation
High Speed, Easy-to-Use, Serial Interface
On-Chip Track/Hold Amplifier
Selection of Input Ranges
10 V for AD7894-10
2.5 V for AD7894-3
0 V to +2.5 V for AD7894-2
High Input Impedance
Low Power: 20 mW Typ
Pin Compatible Upgrade of 12-Bit AD7895
GENERAL DESCRIPTION
The AD7894 is a fast, 14-bit ADC that operates from a single
+5 V supply and is housed in a small 8-lead SOIC. The part
contains a 5 µs successive approximation A/D converter, an on-
chip track/hold amplifier, an on-chip clock and a high speed
serial interface.
Output data from the AD7894 is provided via a high speed,
serial interface port. This two-wire serial interface has a serial
clock input and a serial data output with the external serial clock
accessing the serial data from the part.
In addition to the traditional dc accuracy specifications such as
linearity, full-scale and offset errors, the AD7894 is also speci-
fied for dynamic performance parameters including harmonic
distortion and signal-to-noise ratio.
The part accepts an analog input range of ±10 V (AD7894-10),
±2.5 V (AD7894-3), 0 V to +2.5 V (AD7894-2), and operates
from a single +5 V supply consuming only 20 mW typical.
The AD7894 features a high sampling rate mode and, for low
power applications, a proprietary automatic power-down mode
where the part automatically goes into power-down once conver-
sion is complete and “wakes up” before the next conversion
cycle.
The part is available in a small outline IC (SOIC).
PRODUCT HIGHLIGHTS
1. Fast, 14-Bit ADC in 8-Lead Package
The AD7894 contains a 5␣ µs ADC, a track/hold amplifier,
control logic and a high speed serial interface, all in an 8-lead
package. This offers considerable space saving over alterna-
tive solutions.
2. Low Power, Single Supply Operation
The AD7894 operates from a single +5 V supply and con-
sumes only 20 mW. The automatic power-down mode,
where the part goes into power-down once conversion is
complete and “wakes up” before the next conversion cycle,
makes the AD7894 ideal for battery powered or portable
applications.
3. High Speed Serial Interface
The part provides high speed serial data and serial clock lines
allowing for an easy, two-wire serial interface arrangement.
–2– REV. 0
AD7894–SPECIFICATIONS
(VDD = +5 V 5%, GND = 0 V, REF IN = +2.5 V. All specifications TMIN to TMAX unless
otherwise noted.)
Parameter A Versions
l
B Versions
1
Units Test Conditions/Comments
DYNAMIC PERFORMANCE
2
Signal to (Noise + Distortion) Ratio
3
@ +25°C 78 78 dB min f
IN
= 70 kHz Sine Wave, f
SAMPLE
= 160 kHz
T
MIN
to T
MAX
77 77 dB min See Figure 14
Total Harmonic Distortion (THD)
3
–86 –86 dB max f
IN
= 10 kHz Sine Wave, f
SAMPLE
= 160 kHz,
Typically –87 dB. See Figure 15
Peak Harmonic or Spurious Noise
3
–92 –92 dB typ f
IN
= 10 kHz Sine Wave, f
SAMPLE
= 160 kHz
Intermodulation Distortion (IMD)
3
fa = 9 kHz, fb = 9.5 kHz, f
SAMPLE
= 160 kHz
2nd Order Terms –92 –92 dB typ
3rd Order Terms –92 –92 dB typ
DC ACCURACY
Resolution 14 14 Bits
Minimum Resolution for Which
No Missing Codes Are Guaranteed 14 14 Bits
Relative Accuracy
3
±2±1.5 LSB max
Differential Nonlinearity
3
–1 to +1.5 –1 to +1.5 LSB max
AD7894-2
Positive Gain Error
3
±12 ±10 LSB max
Unipolar Offset Error ±8±6 LSB max
AD7894-10, AD7894-3 Only
Positive Gain Error
3
±8±6 LSB max
Negative Gain Error
3
±8±6 LSB max
Bipolar Zero Error ±10 ±8 LSB max
ANALOG INPUT
AD7894-10
Input Voltage Range ±10 ±10 V
Input Current 2 2 mA max See Analog Input Section
AD7894-3
Input Voltage Range ±2.5 ±2.5 V
Input Current 1.5 1.5 mA max See Analog Input Section
AD7894-2
Input Voltage Range 0 to +2.5 0 to +2.5 V
Input Current 500 500 nA max
REFERENCE INPUT
REF IN Input Voltage Range 2.375/2.625 2.375/2.625 V min/V max 2.5 V ± 5%
Input Current 1 1 µA max
Input Capacitance
4
10 10 pF max
LOGIC INPUTS
Input High Voltage, V
INH
2.4 2.4 V min V
DD
= 5 V ± 5%
Input Low Voltage, V
INL
0.8 0.8 V max V
DD
= 5 V ± 5%
Input Current, I
IN
±10 ±10 µA max V
IN
= 0 V to V
DD
Input Capacitance, C
IN4
10 10 pF max
LOGIC OUTPUTS
Output High Voltage, V
OH
4.0 4.0 V min I
SOURCE
= 400 µA
Output Low Voltage, V
OL
0.4 0.4 V max I
SINK
= 1.6 mA
Output Coding
AD7894-10, AD7894-3 Twos Complement
AD7894-2 Straight (Natural) Binary
CONVERSION RATE
Conversion Time
Mode 1 Operation 5 5 µs max
Mode 2 Operation
5
10 10 µs max
Track/Hold Acquisition Time
3
0.35 0.35 µs max
SAMPLE AND HOLD
–3 dB Small Signal Bandwidth 7.5 7.5 MHz typ
Aperture Jitter 50 50 ps typ
–3–REV. 0
AD7894
Parameter A Versions
l
B Versions
1
Units Test Conditions/Comments
POWER REQUIREMENTS
V
DD
+5 +5 V nom ±5% for Specified Performance
I
DD
5.5 5.5 mA max Digital Inputs @ V
DD
, V
DD
= 5 V ± 5%
Power Dissipation 27.5 27.5 mW max Typically 20 mW
Power-Down Mode
I
DD
@ T
MIN
to T
MAX
20 20 µA max Digital Inputs @ GND, V
DD
= 5 V ± 5%
Power Dissipation T
MIN
to T
MAX
100 100 µW max
Typ 15 µW
NOTES
1
Temperature ranges are as follows: A, B Versions: –40°C to +85°C.
2
Applies to Mode 1 operation. See Operating Modes section.
3
See Terminology.
4
Sample tested @ +25°C to ensure compliance.
5
This 10 µs includes the “wake-up” time from standby. This “wake-up” time is timed from the rising edge of CONVST, whereas conversion is timed from the falling
edge of CONVST, for narrow CONVST pulsewidth the conversion time is effectively the “wake-up” time plus conversion time, hence 10 µs. This can be seen from
Figure 3. Note that if the CONVST pulsewidth is greater than 5 µs, the effective conversion time will increase beyond 10 µs.
Specifications subject to change without notice.
TIMING CHARACTERISTICS
1, 2
Parameter A, B Versions Units Test Conditions/Comments
t
1
40 ns min CONVST Pulsewidth
t
2
31.25
2
ns min SCLK High Pulsewidth
t
3
31.25
2
ns min SCLK Low Pulsewidth
t
4
60
3
ns max Data Access Time after Falling Edge of SCLK
V
DD
= 5 V ± 5%
t
5
10 ns min Data Hold Time after Falling Edge of SCLK
t
6
20
4
ns max Bus Relinquish Time after Falling Edge of SCLK
NOTES
1
Sample tested at +25°C to ensure compliance. All input signals are measured with tr = tf = 1 ns (10% to 90% of +5 V) and timed from a voltage level of +1.6 V.
2
The SCLK maximum frequency is 16 MHz. Care must be taken when interfacing to account for the data access time, t
4
, and the setup time required for the user’s
processor. These two times will determine the maximum SCLK frequency with which the user’s system can operate. See Serial Interface section for more information.
3
Measured with the load circuit of Figure 1 and defined as the time required for an output to cross 0.8 V or 2.0 V.
4
Derived from the measured time taken by the data outputs to change 0.5 V when loaded with the circuit of Figure 1. The measured number is then extrapolated back
to remove the effects of charging or discharging the 50 pF capacitor. This means that the time, t
6
, quoted in the timing characteristics is the true bus relinquish time
of the part and as such is independent of external bus loading capacitances.
Specifications subject to change without notice.
(VDD = +5 V 5%, GND = 0 V, REF IN = +2.5 V)
ABSOLUTE MAXIMUM RATINGS*
(T
A
= +25°C unless otherwise noted)
V
DD
to GND . . . . . . . . . . . . . . . . . . . . . . . . . . –0.3␣ V to +7 V
Analog Input Voltage to GND
AD7894-10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ±17 V
AD7894-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ±7 V
AD7894-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . –5 V to +10 V
Reference Input Voltage to GND . . . . –0.3 V to V
DD
+ 0.3 V
Digital Input Voltage to GND . . . . . . . –0.3 V to V
DD
+ 0.3 V
Digital Output Voltage to GND . . . . . –0.3 V to V
DD
+ 0.3 V
Operating Temperature Range
Commercial (A, B Versions) . . . . . . . . . . . –40°C to +85°C
Storage Temperature Range . . . . . . . . . . . –65°C to +150°C
Junction Temperature . . . . . . . . . . . . . . . . . . . . . . . . +150°C
SOIC Package, Power Dissipation . . . . . . . . . . . . . . . 450 mW
␣␣θ
JA
Thermal Impedance . . . . . . . . . . . . . . . . . . . . . 170°C/W
Lead Temperature, Soldering
␣␣␣␣Vapor Phase (60 sec) . . . . . . . . . . . . . . . . . . . . . . . +215°C
␣␣␣␣Infrared (15 sec) . . . . . . . . . . . . . . . . . . . . . . . . . . .+220°C
*Stresses above those listed under Absolute Maximum Ratings may cause perma-
nent damage to the device. This is a stress rating only; functional operation of the
device at these or any other conditions above those listed in the operational
sections of this specification is not implied. Exposure to absolute maximum rating
conditions for extended periods may affect device reliability.
CAUTION
ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily
accumulate on the human body and test equipment and can discharge without detection.
Although the AD7894 features proprietary ESD protection circuitry, permanent damage may
occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD
precautions are recommended to avoid performance degradation or loss of functionality.
WARNING!
ESD SENSITIVE DEVICE
AD7894
–4– REV. 0
ORDERING GUIDE
Temperature Package Package
Model Range INL Input Range SNR Description Option
AD7894AR-10 –40°C to +85°C±2 LSB ±10 V 77 dB 8-Lead Narrow Body SOIC SO-8
AD7894BR-10 –40°C to +85°C±1.5 LSB ±10 V 77 dB 8-Lead Narrow Body SOIC SO-8
AD7894AR-3 –40°C to +85°C±2 LSB ±2.5 V 77 dB 8-Lead Narrow Body SOIC SO-8
AD7894BR-3 –40°C to +85°C±1.5 LSB ±2.5 V 77 dB 8-Lead Narrow Body SOIC SO-8
AD7894AR-2 –40°C to +85°C±2 LSB 0 V to +2.5 V 77 dB 8-Lead Narrow Body SOIC SO-8
PIN FUNCTION DESCRIPTIONS
Pin Pin
No. Mnemonic Description
1 REF IN Voltage Reference Input. An external reference source should be connected to this pin to provide the
reference voltage for the AD7894’s conversion process. The REF IN input is buffered on-chip. The
nominal reference voltage for correct operation of the AD7894 is +2.5␣ V.
2V
IN
Analog Input Channel. The analog input range is ±10 V (AD7894-10), ±2.5 V (AD7894-3) and 0 V to
+2.5␣ V (AD7894-2).
3 GND Analog Ground. Ground reference for track/hold, comparator, digital circuitry and DAC.
4 SCLK Serial Clock Input. An external serial clock is applied to this input to obtain serial data from the AD7894.
A new serial data bit is clocked out on the falling edge of this serial clock. Data is guaranteed valid for
10 ns after this falling edge so data can be accepted on the falling edge when a fast serial clock is used.
The serial clock input should be taken low at the end of the serial data transmission.
5 SDATA Serial Data Output. Serial data from the AD7894 is provided at this output. The serial data is clocked
out by the falling edge of SCLK, but the data can also be read on the falling edge of SCLK. This is pos-
sible because data bit N is valid for a specified time after the falling edge of SCLK (data hold time) (see
Figure 5). Sixteen bits of serial data are provided as two leading zeroes followed by the 14 bits of conver-
sion data. On the 16th falling edge of SCLK, the SDATA line is held for the data hold time and then
disabled (three-stated). Output data coding is twos complement for the AD7894-10 and AD7894-3, and
straight binary for the AD7894-2.
6 BUSY The BUSY pin is used to indicate when the part is doing a conversion. The BUSY pin will go high on
the falling edge of CONVST and will return low when the conversion is complete.
7CONVST Conversion Start. Edge-triggered logic input. On the falling edge of this input, the track/hold goes into its
hold mode and conversion is initiated. If CONVST is low at the end of conversion, the part goes into
power-down mode. In this case, the rising edge of CONVST will cause the part to begin waking up.
8V
DD
Positive supply voltage, +5 V ± 5%.
1.6mA
400mA
+1.6V
TO
OUTPUT
PIN 50pF
Figure 1. Load Circuit for Access Time and Bus
Relinquish Time
PIN CONFIGURATION
SOIC (SO-8)
TOP VIEW
(Not to Scale)
8
7
6
5
1
2
3
4
REF IN
V
IN
GND
V
DD
CONVST
BUSY
SDATASCLK
AD7894
AD7894
–5–REV. 0
TERMINOLOGY
Signal to (Noise + Distortion) Ratio
This is the measured ratio of signal to (noise + distortion) at the
output of the A/D converter. The signal is the rms amplitude of
the fundamental. Noise is the rms sum of all nonfundamental
signals up to half the sampling frequency (f
S
/2), excluding dc.
The ratio is dependent upon the number of quantization levels
in the digitization process; the more levels, the smaller the quan-
tization noise. The theoretical signal to (noise + distortion) ratio
for an ideal N-bit converter with a sine wave input is given by:
Signal to (Noise + Distortion) = (6.02␣ N + 1.76) dB
Thus for a 14-bit converter, this is 86.04 dB.
Total Harmonic Distortion
Total harmonic distortion (THD) is the ratio of the rms sum of
harmonics to the fundamental. For the AD7894, it is defined as:
THD dB VVVVV
V
( ) log=++++
20
2
2
3
2
4
2
5
2
6
2
1
where V
1
is the rms amplitude of the fundamental and V
2
, V
3
,
V
4
, V
5
and V
6
are the rms amplitudes of the second through the
sixth harmonics.
Peak Harmonic or Spurious Noise
Peak harmonic or spurious noise is defined as the ratio of the
rms value of the next largest component in the ADC output
spectrum (up to f
S
/2 and excluding dc) to the rms value of the
fundamental. The value of this specification is normally deter-
mined by the largest harmonic in the spectrum, but for parts
where the harmonics are buried in the noise floor, it will be a
noise peak.
Intermodulation Distortion
With inputs consisting of sine waves at two frequencies, fa and
fb, any active device with nonlinearities will create distortion
products at sum and difference frequencies of mfa ± nfb where
m, n = 0, 1, 2, 3, etc. Intermodulation terms are those for which
neither m nor n is equal to zero. For example, the second order
terms include (fa + fb) and (fa – fb), while the third order terms
include (2 fa + fb), (2 fa – fb), (fa + 2 fb) and (fa – 2 fb).
The AD7894 is tested using two input frequencies. In this case,
the second and third order terms are of different significance.
The second order terms are usually distanced in frequency from
the original sine waves, while the third order terms are usually at
a frequency close to the input frequencies. As a result, the second
and third order terms are specified separately. The calculation
of the intermodulation distortion is as per the THD specification
where it is the ratio of the rms sum of the individual distortion
products to the rms amplitude of the fundamental expressed
in dBs.
Relative Accuracy
Relative accuracy or endpoint nonlinearity is the maximum
deviation from a straight line passing through the endpoints of
the ADC transfer function.
Differential Nonlinearity
This is the difference between the measured and the ideal 1␣ LSB
change between any two adjacent codes in the ADC.
Positive Gain Error (AD7894-10)
This is the deviation of the last code transition (01 . . . 110 to
01 . . . 111) from the ideal (4 × VREF – 1 LSB) after the
Bipolar Zero Error has been adjusted out.
Positive Gain Error (AD7894-3)
This is the deviation of the last code transition (01 . . . 110 to
01 . . . 111) from the ideal (VREF – 1 LSB) after the Bipolar
Zero Error has been adjusted out.
Positive Gain Error (AD7894-2)
This is the deviation of the last code transition (11 . . . 110 to
11 . . . 111) from the ideal (VREF – 1 LSB) after the Unipolar
Offset Error has been adjusted out.
Bipolar Zero Error (AD7894-10, AD7894-3)
This is the deviation of the midscale transition (all 0s to all 1s)
from the ideal 0 V (GND).
Unipolar Offset Error (AD7894-2)
This is the deviation of the first code transition (00 . . . 000 to
00 . . . 001) from the ideal 1 LSB.
Negative Gain Error (AD7894-10)
This is the deviation of the first code transition (10 . . . 000 to
10 . . . 001) from the ideal (–4 × VREF + 1 LSB) after Bipolar
Zero Error has been adjusted out.
Negative Gain Error (AD7894-3)
This is the deviation of the first code transition (10 . . . 000 to
10 . . . 001) from the ideal (– VREF + 1 LSB) after Bipolar
Zero Error has been adjusted out.
Track/Hold Acquisition Time
Track/Hold acquisition time is the time required for the output
of the track/hold amplifier to reach its final value, within
±1/2␣ LSB, after the end of conversion (the point at which the
track/hold returns to track mode). It also applies to situations
where there is a step input change on the input voltage applied
to the V
IN
input of the AD7894. This means that the user must
wait for the duration of the track/hold acquisition time after the
end of conversion or after a step input change to V
IN
before
starting another conversion, to ensure that the part operates to
specification.
AD7894
–6– REV. 0
CONVERTER DETAILS
The AD7894 is a fast, 14-bit single supply A/D converter. It
provides the user with signal scaling, track/hold, A/D converter
and serial interface logic functions on a single chip. The A/D
converter section of the AD7894 consists of a conventional
successive-approximation converter based around an R-2R
ladder structure. The signal scaling on the AD7894-10 and
AD7894-3 allows the part to handle ±10 V and ±2.5 V input
signals respectively while operating from a single +5␣ V supply.
The AD7894-2 accepts an analog input range of 0 V to +2.5 V.
The part requires an external +2.5 V reference. The reference
input to the part is buffered on-chip. The AD7894 has two
operating modes, the high sampling mode and the “auto-sleep”
mode where the part automatically goes into sleep after the end
of conversion. These modes are discussed in more detail in the
Timing and Control Section.
A major advantage of the AD7894 is that it provides all of the
above functions in an 8-lead SOIC package. This offers the user
considerable space saving advantages over alternative solutions.
The AD7894 typically consumes only 20␣ mW, making it ideal
for battery powered applications.
Conversion is initiated on the AD7894 by pulsing the CONVST
input. On the falling edge of CONVST, the on-chip track/hold
goes from track-to-hold mode and the conversion sequence is
started. The conversion clock for the part is generated internally
using a laser-trimmed clock oscillator circuit. Conversion time for
the AD7894 is 5␣ µs in the high sampling mode (10 µs for the auto
sleep mode), and the track/hold acquisition time is 0.35␣ µs. To
obtain optimum performance from the part, the read operation
should not occur during the conversion or during 250 ns prior
to the next conversion. This allows the part to operate at through-
put rates up to 160 kHz and achieve data sheet specifications.
CIRCUIT DESCRIPTION
Analog Input Section
The AD7894 is offered as three part types, the AD7894-10,
which handles a ±10 V input voltage range, the AD7894-3,
which handles input voltage range ±2.5 V and the AD7894-2,
which handles a 0␣ V to +2.5␣ V input voltage range.
TRACK/
HOLD
AD7894-10/AD7894-3
REF IN
V
IN
GND
R1
R2
R3
TO ADC
REFERENCE
CIRCUITRY
TO INTERNAL
COMPARATOR
Figure 2. AD7894-10/AD7894-3 Analog Input Structure
Figure 2 shows the analog input section for the AD7894-10 and
AD7894-3. The analog input range of the AD7894-10 is ±10 V
and the analog input range for the AD7894-3 is ±2.5 V. This
input is benign, with no dynamic charging currents as the resis-
tor stage is followed by a high input impedance stage of the
track/hold amplifier. For the AD7894-10, R1 = 8 k, R2 = 2 k
and R3 = 2 k. For the AD7894-3, R1 = R2 = 2 k and R3
is open circuit. The current flowing in the analog input is di-
rectly related to the analog input voltage. The maximum input
current flows when the analog input is at negative full scale.
For the AD7894-10 and AD7894-3, the designed code transi-
tions occur on successive integer LSB values (i.e., 1 LSB, 2 LSBs,
3 LSBs . . .). Output coding is twos complement binary with
1 LSB = FS/16384. The ideal input/output transfer function for
the AD7894-10 and AD7894-3 is shown in Table I.
Table I. Ideal Input/Output Code Table for the AD7894-10/
AD7894-3
Digital Output
Analog Input
l
Code Transition
+FSR/2 – 1 LSB
2
011 . . . 110 to 011 . . . 111
+FSR/2 – 2 LSBs 011 . . . 101 to 011 . . . 110
+FSR/2 – 3 LSBs 011 . . . 100 to 011 . . . 101
GND + 1 LSB 000 . . . 000 to 000 . . . 001
GND 111 . . . 111 to 000 . . . 000
GND – 1 LSB 111 . . . 110 to 111 . . . 111
–FSR/2 + 3 LSBs 100 . . . 010 to 100 . . . 011
–FSR/2 + 2 LSBs 100 . . . 001 to 100 . . . 010
–FSR/2 + 1 LSB 100 . . . 000 to 100 . . . 001
NOTES
1
FSR is full-scale range = 20 V (AD7894-10) and = 5 V (AD7894-3) with
REF IN = +2.5 V.
2
1 LSB = FSR/16384 = 1.22 mV (AD7894-10) and 0.3 mV (AD7894-3) with
REF IN = +2.5 V.
The analog input section for the AD7894-2 contains no biasing
resistors and the V
IN
pin drives the input directly to the track/
hold amplifier. The analog input range is 0 V to +2.5 V into a
high impedance stage with an input current of less than 500 nA.
This input is benign, with no dynamic charging currents. Once
again, the designed code transitions occur on successive integer
LSB values. Output coding is straight (natural) binary with
1 LSB = FS/16384 = 2.5 V/16384 = 0.15 mV. Table II shows
the ideal input/output transfer function for the AD7894-2.
Table II. Ideal Input/Output Code Table for AD7894-2
Digital Output
Analog Input
1
Code Transition
+FSR – 1 LSB
2
111 . . . 110 to 111 . . . 111
+FSR – 2 LSB 111 . . . 101 to 111 . . . 110
+FSR – 3 LSB 111 . . . 100 to 111 . . . 101
GND + 3 LSB 000 . . . 010 to 000 . . . 011
GND + 2 LSB 000 . . . 001 to 000 . . . 010
GND + 1 LSB 000 . . . 000 to 000 . . . 001
NOTES
1
FSR is full-scale range and is 2.5 V for AD7894-2 with VREF = +2.5 V.
2
1 LSB = FSR/16384 and is 0.15 mV for AD7894-2 with VREF = +2.5 V.
AD7894
–7–REV. 0
Track/Hold Section
The track/hold amplifier on the analog input of the AD7894
allows the ADC to accurately convert an input sine wave of full-
scale amplitude to 14-bit accuracy. The input bandwidth of the
track/hold is greater than the Nyquist rate of the ADC, even
when the ADC is operated at its maximum throughput rate of
160 kHz (i.e., the track/hold can handle input frequencies in
excess of 100 kHz).
The track/hold amplifier acquires an input signal to 14-bit accu-
racy in less than 0.35␣ µs. The operation of the track/hold is
essentially transparent to the user. With the high sampling
operating mode the track/hold amplifier goes from its tracking
mode to its hold mode at the start of conversion (i.e., the falling
edge of CONVST). The aperture time for the track/hold (i.e.,
the delay time between the external CONVST signal and the
track/hold actually going into hold) is typically 15␣ ns. At the
end of conversion (on the falling edge of BUSY) the part re-
turns to its tracking mode. The acquisition time of the track/
hold amplifier begins at this point. For the auto shutdown mode,
the rising edge of CONVST wakes up the part and the track
and hold amplifier goes from its tracking mode to its hold mode
5 µs after the rising edge of CONVST (provided that the
CONVST high time is less than 5 µs). Once again the part re-
turns to its tracking mode at the end of conversion when the
BUSY signal goes low.
Reference Input
The reference input to the AD7894 is buffered on-chip with a
maximum reference input current of 1␣ µA. The part is specified
with a +2.5 V reference input voltage. Errors in the reference
source will result in gain errors in the AD7894’s transfer func-
tion and will add to the specified full-scale errors on the part.
Suitable reference sources for the AD7894 include the AD780
and AD680 precision +2.5 V references.
Timing and Control Section
Figure 3 shows the timing and control sequence required to
obtain optimum performance from the AD7894. In the se-
quence shown, conversion is initiated on the falling edge of
CONVST and new data from this conversion is available in the
output register of the AD7894 5␣ µs later. Once the read opera-
tion has taken place, a further 250␣ ns should be allowed before
the next falling edge of CONVST to optimize the settling of the
track/hold amplifier before the next conversion is initiated.
With the serial clock frequency at its maximum of 16␣ MHz, the
achievable throughput rate for the part is 5␣ µs (conversion
time) plus 1.0␣ µs (read time) plus 250␣ ns (quiet time). This
results in a minimum throughput time of 6.25␣ µs (equivalent to
a throughput rate of 160 kHz). A serial clock of less than 16 MHz
can be used, but this will in turn mean that the throughput
time will increase.
The read operation consists of 16 serial clock pulses to the
output shift register of the AD7894. After 16 serial clock pulses
the shift register is reset and the SDATA line is three-stated. If
there are more serial clock pulses after the 16th clock, the shift
register will be moved on past its reset state. However, the shift
register will be reset again on the falling edge of the CONVST
signal to ensure that the part returns to a known state every
conversion cycle. As a result, a read operation from the output
register should not straddle across the falling edge of CONVST
as the output shift register will be reset in the middle of the
read operation and the data read back into the microprocessor
will appear invalid.
OPERATING MODES
Mode 1 Operation (High Sampling Performance)
The timing diagram in Figure 3 is for optimum performance in
operating Mode 1 where the falling edge of CONVST starts
conversion and puts the Track/Hold amplifier into its hold
mode. This falling edge of CONVST also causes the BUSY
signal to go high to indicate that a conversion is taking place.
The BUSY signal goes low when the conversion is complete,
which is 5 µs max after the falling edge of CONVST and new
data from this conversion is available in the output register of
the AD7894. A read operation accesses this data. This read
operation consists of 16 clock cycles and the length of this read
operation will depend on the serial clock frequency. For the
fastest throughput rate (with a serial clock of 16 MHz) the read
operation will take 1.0 µs. The read operation must be com-
plete at least 250 ns before the falling edge of the next CONVST
and this gives a total time of 6.25 µs for the full throughput
time (equivalent to 160 kHz). This mode of operation should
be used for high sampling applications.
250ns MIN
t
1
= 40ns MIN
t
CONVERT
= 5ms
BUSY
SCLK
CONVST
CONVERSION IS
INITIATED;
TRACK/HOLD
GOES INTO HOLD
CONVERSION
ENDS
5ms LATER
SERIAL READ
OPERATION READ OPERATION
SHOULD END
250ns PRIOR TO
NEXT FALLING
EDGE OF CONVST
OUTPUT
SERIAL
SHIFT
REGISTER
IS RESET
Figure 3. Mode 1 Timing Operation Diagram for High Sampling Performance
AD7894
–8– REV. 0
250ns MIN
BUSY
SCLK
CONVST
CONVERSION
IS INITIATED;
TRACK/HOLD
GOES INTO
HOLD
CONVERSION
ENDS
10ms LATER
SERIAL READ
OPERATION READ OPERATION
SHOULD END 250ns
PRIOR TO NEXT
RISING EDGE OF
CONVST
OUTPUT
SERIAL SHIFT
REGISTER
IS RESET
t
CONVERT
= 10ms
PART
WAKES
UP
Figure 4. Mode 2 Timing Diagram Where Automatic Sleep Function is Initiated
t2t3
t4t5t6
2 LEADING
ZEROS
THREE-STATE THREE-
STATE
1 2 3 4 15 16
DB13 DB12 DB0
SCLK (I/P)
DOUT (O/P)
t2 = t3 = 31.25ns MIN, t4 = 60ns MAX, t5 = 10ns MIN, t6 = 20ns MAX @ 5V, A, B, VERSIONS
Figure 5. Data Read Operation
Mode 2 Operation (Auto Sleep After Conversion)
The timing diagram in Figure 4 is for optimum performance in
operating Mode 2, where the part automatically goes into sleep
mode once BUSY goes low, after conversion and “wakes up”
before the next conversion takes place. This is achieved by keep-
ing CONVST low at the end of conversion, whereas it was high
at the end of conversion for Mode 1 Operation. The rising edge
of CONVST “wakes up” the AD7894. This wake-up time is
typically 5 µs and is controlled internally by a monostable cir-
cuit. While the AD7894 is waking up there is some digital activ-
ity internal to the part. If the falling edge of CONVST (putting
the track/hold amplifier into hold mode) should occur during
this digital activity, noise will be injected into the track/hold
amplifier resulting in a poor conversion. For optimum results
the CONVST pulse should be between 40 ns and 2 µs or greater
than 6 µs in width. The narrower pulse will allow a system to
instruct the AD7894 to begin waking up and perform a conver-
sion when ready, whereas the pulse greater than 6 µs will give
control over when the sampling instant takes place. Note that
the 10 µs wake-up time shown in Figure 4 is for a CONVST pulse
less than 2 µs. If a CONVST pulse greater than 6 µs is used, the
conversion will not complete for a further 5 µs after the falling edge
of CONVST. Even though the part is in sleep mode, data can still
be read from it. The read operation consists of 16 clock cycles as in
Mode 1 Operation. For the fastest serial clock of 16 MHz, the read
operation will take 1.0 µs and this must be complete at least 250 ns
before the falling edge of the next CONVST, to allow the track/
hold amplifier to have enough time to settle. This mode is very
useful when the part is converting at a slow rate, as the power
consumption will be significantly reduced from that of Mode 1
Operation.
Serial Interface
The serial interface to the AD7894 consists of just three wires, a
serial clock input (SCLK) and the serial data output (SDATA)
and a conversion status output (BUSY). This allows for an
easy-to-use interface to most microcontrollers, DSP processors
and shift registers.
Figure 5 shows the timing diagram for the read operation to the
AD7894. The serial clock input (SCLK) provides the clock
source for the serial interface. Serial data is clocked out from
the SDATA line on the falling edge of this clock and is valid on
both the rising and falling edges of SCLK. The advantage of
having the data valid on both the rising and falling edges of the
SCLK is to give the user greater flexibility in interfacing to the
part and so a wider range of microprocessor and microcontrol-
ler interfaces can be accommodated. This also explains the two
timing figures, t
4
and t
5
, that are quoted on the diagram. The
time t
4
specifies how long after the falling edge of the SCLK the
next data bit becomes valid, whereas the time t
5
specifies for
how long after the falling edge of the SCLK the current data bit
is valid. The first leading zero is clocked out on the first rising
edge of SCLK. Note that the first zero will be valid on the first
falling edge of SCLK even though the data access time is speci-
fied at 60 ns for the other bits. The reason for this is that the
first bit will be clocked out faster than the other bits is due to
the internal architecture of the part. Sixteen clock pulses must
be provided to the part to access to full conversion result. The
AD7894 provides two leading zeros followed by the 14-bit
conversion result starting with the MSB (DB13). The last data
bit to be clocked out on the penultimate falling clock edge is the
LSB (DB0). On the 16th falling edge of SCLK the LSB (DB0)
will be valid for a specified time to allow the bit to be read on
the falling edge of the SCLK and then the SDATA line is dis-
abled (three-stated). After this last bit has been clocked out,
the SCLK input should return low and remain low until the
next serial data read operation. If there are extra clock pulses
after the 16th clock, the AD7894 will start over again with
outputting data from its output register and the data bus will
no longer be three-stated even when the clock stops. Provided
the serial clock has stopped before the next falling edge of
AD7894
–9–REV. 0
CONVST, the AD7894 will continue to operate correctly
with the output shift register being reset on the falling edge of
CONVST. However, the SCLK line must be low when CONVST
goes low in order to reset the output shift register correctly.
The serial clock input does not have to be continuous during the
serial read operation. The 16 bits of data (two leading zeros and
14-bit conversion result) can be read from the AD7894 in a
number of bytes.
The AD7894 counts the serial clock edges to know which bit
from the output register should be placed on the SDATA out-
put. To ensure that the part does not lose synchronization, the
serial clock counter is reset on the falling edge of the CONVST
input provided the SCLK line is low. The user should ensure
that the SCLK line remains low until the end of the conversion.
When the conversion is complete, BUSY goes low, the output
register will be loaded with the new conversion result and can be
read from with 16 clock cycles of SCLK.
MICROPROCESSOR/MICROCONTROLLER INTERFACE
The AD7894 provides a two-wire serial interface that can be
used for connection to the serial ports of DSP processors and
microcontrollers. Figures 6 through 9 show the AD7894
interfaced to a number of different microcontrollers and DSP
processors. The AD7894 accepts an external serial clock and
as a result, in all interfaces shown here, the processor/controller
is configured as the master, providing the serial clock, with
the AD7894 being the slave in the system. The BUSY signal
need not be used for a two-wire interface if the read can be
timed to occur 5 µs after the start of conversion (assuming
Mode 1 operation).
AD7894 to 8X51/L51 Interface
Figure 6 shows an interface between the AD7894 and the
8X51/L51 microcontroller. The 8X51/L51 is configured for its
Mode 0 serial interface mode. The diagram shows the simplest
form of the interface where the AD7894 is the only part con-
nected to the serial port of the 8X51/L51 and, therefore, no
decoding of the serial read operations is required.
To select the AD7894 in systems where more than one device is
connected to the 8X51/L51’s serial port, a port bit, configured
as an output from one of the 8X51/L51’s parallel ports, can be
used to gate on or off the serial clock to the AD7894. A simple
AND function on this port bit and the serial clock from the
8X51/L51 will provide this function. The port bit should be
high to select the AD7894 and low when it is not selected.
The end of conversion can be monitored by using the BUSY
signal, which is shown in the interface diagram of Figure 6.
With the BUSY line from the AD7894 connected to the Port
P1.2 of the 8X51/L51 the BUSY line can be polled by the
8X51/L51. The BUSY line can be connected to the INT1 line
of the 8X51/L51 if an interrupt driven system is preferred.
These two options are shown on the diagram.
Note also that the AD7894 outputs the MSB first during a read
operation while the 8X51/L51 expects the LSB first. Therefore,
the data that is read into the serial buffer needs to be rearranged
before the correct data format from the AD7894 appears in the
accumulator.
The serial clock rate from the 8X51/L51 is limited to signifi-
cantly less than the allowable input serial clock frequency with
which the AD7894 can operate. As a result, the time to read
data from the part will actually be longer than the conversion
time of the part. This means that the AD7894 cannot run at its
maximum throughput rate when used with the 8X51/L51.
P1.2 OR INT1
P3.0
P3.1
SDATA
BUSY
SCLK
AD7894
8X51/L51
Figure 6. AD7894 to 8X51/L51 Interface
AD7894 to 68HC11/L11 Interface
An interface circuit between the AD7894 and the 68HC11/L11
microcontroller is shown in Figure 7. For the interface shown,
the 68L11 SPI port is used and the 68L11 is configured in its
single-chip mode. The 68L11 is configured in the master mode
with its CPOL bit set to a logic zero and its CPHA bit set to a
logic one. As with the previous interface, the diagram shows the
simplest form of the interface where the AD7894 is the only part
connected to the serial port of the 68L11 and therefore no de-
coding of the serial read operations is required.
Once again, to select the AD7894 in systems where more than
one device is connected to the 68HC11’s serial port, a port bit,
configured as an output from one of the 68HC11’s parallel
ports, can be used to gate on or off the serial clock to the AD7894.
A simple AND function on this port bit and the serial clock
from the 68L11 will provide this function. The port bit should
be high to select the AD7894 and low when it is not selected.
The end of conversion is monitored by using the BUSY signal,
which is shown in the interface diagram of Figure 7. With the
BUSY line from the AD7894 connected to the Port PC2 of the
68HC11/L11 the BUSY line can be polled by the 68HC11/L11.
The BUSY line can be connected to the IRQ line of the 68HC11/
L11 if an interrupt driven system is preferred. These two op-
tions are shown in the diagram.
The serial clock rate from the 68HC11/L11 is limited to signifi-
cantly less than the allowable input serial clock frequency with
which the AD7894 can operate. As a result, the time to read
data from the part will be longer than the conversion time of the
part. This means that the AD7894 cannot run at its maximum
throughput rate when used with the 68HC11/L11.
PC2 OR IRQ
SCK
MISO SDATA
BUSY
SCLK
AD7894
68HC11/L11
Figure 7. AD7894 to 68HC11/L11 Interface
AD7894
–10– REV. 0
AD7894 to ADSP-2101/5 Interface
An interface circuit between the AD7894 and the ADSP-2101/5
DSP processor is shown in Figure 8. In the interface shown, the
RFS1 output from the ADSP-2101/5s SPORT1 serial port is
used to gate the serial clock (SCLK1) of the ADSP-2101/5
before it is applied to the SCLK input of the AD7894. The
RFS1 output is configured for active high operation. The BUSY
line from the AD7894 is connected to the IRQ2 line of the
ADSP-2101/5 so that at the end of conversion an interrupt is
generated telling the ADSP-2101/5 to initiate a read operation.
The interface ensures a noncontinuous clock for the AD7894’s
serial clock input, with only 16 serial clock pulses provided and
the serial clock line of the AD7894 remaining low between data
transfers. The SDATA line from the AD7894 is connected to
the DR1 line of the ADSP-2101/5’s serial port.
The timing relationship between the SCLK1 and RFS1 outputs
of the ADSP-2101/5 are such that the delay between the rising
edge of the SCLK1 and the rising edge of an active high RFS1
is up to 30␣ ns. There is also a requirement that data must be set
up 10␣ ns prior to the falling edge of the SCLK1 to be read cor-
rectly by the ADSP-2101/5. The data access time for the AD7894
is 60␣ ns (A, B versions) from the rising edge of its SCLK input.
Assuming a 10␣ ns propagation delay through the external AND
gate, the high time of the SCLK1 output of the ADSP-2105
must be (30 + 60 + 10 + 10)␣ ns, i.e., 110 ns. This means
that the serial clock frequency with which the interface of Figure
8 can work is limited to 4.5␣ MHz.
Another alternative scheme is to configure the ADSP-2101/5
such that it accepts an external noncontinuous serial clock. In
this case, an external noncontinuous serial clock is provided that
drives the serial clock inputs of both the ADSP-2101/5 and the
AD7894. In this scheme, the serial clock frequency is limited to
the processor’s cycle rate, up to a maximum of 13.8 MHz.
IRQ2
SCLK1
DR1 SDATA
BUSY
AD7894
ADSP-2101/5
SCLK
RFS1
Figure 8. AD7894 to ADSP-2101/5 Interface
AD7894 to DSP56002/L002 Interface
Figure 9 shows an interface circuit between the AD7894 and the
DSP56002/L002 DSP processor. The DSP56002/L002 is
configured for normal-mode asynchronous operation with gated
clock. It is also set up for a 16-bit word with SCK as gated
clock output. In this mode, the DSP56002/L002 provides 16
serial clock pulses to the AD7894 in a serial read operation.
The DSP56002/L002 assumes valid data on the first falling
edge of SCK so the interface is simply three-wire as shown in
Figure 9.
The BUSY line from the AD7894 is connected to the MODA/
IRQA input of the DSP56002/L002 so that an interrupt will be
generated at the end of conversion. This ensures that the read
operation will take place after conversion is finished.
MODA/IRQA
SCK
SDR SDATA
BUSY
SCLK
AD7894
DSP56002/L002
Figure 9. AD7894 to DSP56002/L002 Interface
AD7894 PERFORMANCE
Linearity
The linearity of the AD7894 is determined by the on-chip
14-bit D/A converter. This is a segmented DAC which is laser
trimmed for 14-bit integral linearity and differential linearity.
Typical relative accuracy numbers for the part are ±1/2␣ LSB
while the typical DNL errors are ±1/3␣ LSB.
Noise
In an A/D converter, noise exhibits itself as code uncertainty in
dc applications and as the noise floor (in an FFT, for example)
in ac applications. In a sampling A/D converter like the AD7894,
all information about the analog input appears in the baseband
from dc to 1/2 the sampling frequency. The input bandwidth of
the track/hold exceeds the Nyquist bandwidth, so an antialiasing
filter should be used to remove unwanted signals above f
S
/2 in
the input signal in applications where such signals exist.
Figure 10 shows a histogram plot for 8192 conversions of a dc
input using the AD7894. The analog input was set at the center
of a code transition. It can be seen that almost all the codes
appear in the one output bin indicating very good noise perfor-
mance from the ADC.
ADC CODE
6000
010397
COUNTS
98 99 100 101 102
5000
4000
3000
2000
1000
Figure 10. Histogram of 8192 Conversions of a DC Input
AD7894
–11–REV. 0
Dynamic Performance (Mode 1 Only)
With a conversion time of 5 µs, the AD7894 is ideal for wide
bandwidth signal processing applications. These applications
require information on the ADC’s effect on the spectral con-
tent of the input signal. Signal to (Noise + Distortion), Total
Harmonic Distortion, Peak Harmonic or Spurious Noise and
Intermodulation Distortion are all specified. Figure 11 shows a
typical FFT plot of a 10 kHz, ±10␣ V input after being digitized
by the AD7894-10 operating at a 160 kHz sampling rate. The
signal to (noise + distortion) ratio is 80.24 dB and the total
harmonic distortion is –96.35 dB.
The formula for signal to (noise + distortion) ratio (see Ter-
minology section) is related to the resolution or number of bits
in the converter. Rewriting the formula, below, gives a mea-
sure of performance expressed in effective number of bits (N):
N=(SNR –1.76)
6.02
where SNR is Signal to (Noise + Distortion) Ratio.
FREQUENCY – kHz
0
–14008010
dBs
20 30 40 50 60 70
–20
–60
–80
–100
–120
–40
f
S
= 160kHz
F
IN
= 10kHz
SNR = 80.24dB
THD = –96.35dB
Figure 11. AD7894 FFT Plot
The effective number of bits for a device can be calculated from
its measured signal to (noise + distortion) ratio. Figure 12
shows a typical plot of effective number of bits versus frequency
for the AD7894 from dc to f
SAMPLING
/2. The sampling fre-
quency is 160 kHz. The plot shows that the AD7894 converts
an input sine wave of 10␣ kHz to an effective numbers of bits of
13.00, which equates to a signal to (noise + distortion) level of
80.02 dB.
FREQUENCY – kHz
14
13
910 1000100
ENOBs
12
11
10
Figure 12. Effective Number of Bits vs. Frequency
Power Considerations
In the automatic power-down mode the part may be operated at
a sample rate that is considerably less than 160 kHz. In this
case, the power consumption will be reduced and will depend
on the sample rate. Figure 13 shows a graph of the power con-
sumption versus sampling rates from 1 Hz to 100 kHz in the
automatic power-down mode. The conditions are 5 V supply
+25°C. The SCLK pin was held low and no data was read from
the part.
SAMPLING FREQUENCY – Hz
100
0.1 1 10000010
POWER – mW
100 1000 10000
10
1
Figure 13. Power vs. Sampling Rate in Automatic Power-
Down Mode
82
81
78
–40 80–20 020 40 60
80
79
f
S
= 160kHz
F
IN
= 10kHz
TEMPERATURE – 8C
SNR+D – dB
Figure 14. SNR + D vs. Temperature
FREQUENCY – kHz
100
80
10 1000100
THD – dB
90
70
60
50
40
30
20
10
0
Figure 15. THD vs. Frequency
AD7894
–12– REV. 0
OUTLINE DIMENSIONS
Dimensions shown in inches and (mm).
C3342–8–9/98
PRINTED IN U.S.A.
8-Lead Narrow Body SOIC
(SO-8)
0.1968 (5.00)
0.1890 (4.80)
85
41
0.2440 (6.20)
0.2284 (5.80)
PIN 1
0.1574 (4.00)
0.1497 (3.80)
0.0688 (1.75)
0.0532 (1.35)
SEATING
PLANE
0.0098 (0.25)
0.0040 (0.10)
0.0192 (0.49)
0.0138 (0.35)
0.0500
(1.27)
BSC 0.0098 (0.25)
0.0075 (0.19) 0.0500 (1.27)
0.0160 (0.41)
88
08
0.0196 (0.50)
0.0099 (0.25) 3 458