AD669
REV. A
–9–
Unipolar coding is straight binary, where all zeros (0000H) on
the data inputs yields a zero analog output and all ones
(FFFFH) yields an analog output 1 LSB below full scale.
Bipolar coding is offset binary, where an input code of 0000H
yields a minus full-scale output, an input of FFFFH yields an
output 1 LSB below positive full scale, and zero occurs for an
input code with only the MSB on (8000H).
The AD669 can be used with twos complement input coding if
an inverter is used on the MSB (DB15).
DIGITAL INPUT CONSIDERATIONS
The threshold of the digital input circuitry is set at 1.4 volts.
The input lines can thus interface with any type of 5 volt logic.
The AD669 data and control inputs will float to indeterminate
logic states if left open. It is important that
CS and L1 be con-
nected to DGND and Chat LDAC be tied to V
LL
if these pins
are not used.
Fanout for the AD669 is 40 when used with a standard low
power Schottky gate output device.
16-BIT MICROPROCESSOR INTERFACE
The 16-bit parallel registers of the AD669 allow direct interfac-
ing to 16-bit general purpose and DSP microprocessor buses.
The following examples illustrate typical AD669 interface
configurations.
AD669 TO ADSP-2101 INTERFACE
The flexible interface of the AD669 minimizes the required
“glue” logic when it is connected in configurations such as the
one shown in Figure 8. The AD669 is mapped into the ADSP-
2101’s memory space and requires two wait states using a 12.5
MHz processor clock.
In this configuration, the ADSP-2101 is set up to use the inter-
nal timer to interrupt the processor at the desired sample rate.
The
WR pin and data lines D8–D23 from the ADSP-2101 are
tied directly to the
L1 and DB0 through DB15 pins of the
AD669, respectively. The decoded signal
CS1 is connected to
both
CS and LDAC. When a timer interrupt is detected, the
ADSP-2101 automatically vectors to the appropriate service
routine with minimal overhead. The interrupt routine then in-
structs the processor to execute a data memory write to the ad-
dress of the AD669.
The
WR pin and CS1 both go low causing the first 16-bit latch
inside the AD669 to be transparent. The data present in the first
rank is then latched by the rising edge of
WR. The rising edge
of
CS1 will cause the second rank 16-bit latch to become
transparent updating the output of the DAC. The length of
WR is extended by two wait states to comply with the timing
requirements of t
LOW
shown in Figure 1b. It is important to
latch the data with the rising edge of
WR rather than the decoded
CS1. This is necessary to comply with the t
DH
specification of
the AD669.
A0
D8
ADSP-2101
DGND
+5V
DECODER
ADDRESS BUS
LDAC
AD669
DGNDDB0
DATA BUS
DMS
WR
CS1
CS
L1
A13
DB15
D23
V
LL
V
LL
V
OUT
a. ADSP-2101 to AD669 Interface
A13
A12
A11
DMS
CS1
b. Typical Address Decoder
Figure 8. ADSP-2101 to AD669 Interface
Figure 8b shows the circuitry a typical decoder might include.
In this case, a data memory write to any address in the range
3000H to 3400H will result in the AD669 being updated. These
decoders will vary greatly depending on the number of devices
memory-mapped by the processor.
AD669 TO DSP56001 INTERFACE
Figure 9 shows the interface between the AD669 and the
DSP56001. Like the ADSP-2101, the AD669 is mapped into
the DSP56001’s memory space. This application was tested
with a processor clock of 20.48 MHz (t
CYC
= 97.66 ns) although
faster rates are possible.
An external clock connected to the
IRQA pin of the DSP56001
interrupts the processor at the desired sample rate. If ac perfor-
mance is important, this clock should be synchronous with the
DSP56001 processor clock. Asynchronous clocks will cause jit-
ter on the latch signal due to the uncertainty associated with the
acknowledgment of the interrupt. A synchronous clock is easily
generated by dividing down the clock from the DSP crystal. If
ac performance is not important, it is not necessary for
IRQA to
be synchronous.
After the interrupt is acknowledged, the interrupt routine ini-
tiates a memory write cycle. All of the AD669 control inputs are
AD669
REV. A
–10–
tied together which configures the input stage as an edge trig-
gered 16-bit register. The rising edge of the decoded signal
latches the data and updates the output of the DAC. It is neces-
sary to insert wait states after the processor initiates the write
cycle to comply with the timing requirements t
LOW
shown in
Figure 1b. The number of wait states that are required will vary
depending on the processor cycle time. The equation given in
Figure 9 can be used to determine the number of wait states
given the frequency of the processor crystal.
A0–A15
DSP56001
D0–D23
DGND
+5V
EXTERNAL
CLOCK
ADDRESS
DECODE
74F32
LDAC
AD669
DGND
DB0–DB15
XTAL
CS
L1
DS
X/Y
WR
IRQA
V
LL
CS1
T =
1
2 (XTAL)
t – T + 9ns
2T
LOW
# OF
WAIT STATES =
V
LL
Figure 9. DSP56001 to AD669 Interface
As an example, the 20.48 MHz crystal used in this application
results in T = 24.4 ns which means that the required number of
wait states is about 2.76. This must be rounded to the next
highest integer to assure that the minimum pulse widths comply
with those required by the AD669. As the speed of the proces-
sor is increased, the data hold time relative to
CS1 decreases. As
processor clocks increase beyond 20.48 MHz, a configuration
such as the one shown for the ADSP-2101 is the better choice.
AD669 TO 8086 INTERFACE
Figure 10 shows the 8086 16-bit microprocessor connected to
multiple AD669s. The double-buffered capability of the AD669
allows the microprocessor to write to each AD669 individually
and then update all the outputs simultaneously. Processor
speeds of 6, 8, and 10 MHz require no wait states to interface
with the AD669.
The 8086 software routine begins by writing a data word to the
CS1 address. The decoder must latch the address using the
ALE signal. The decoded
CS1 pulse goes low causing the first
rank latch of the associated AD669 to become transparent.
Simultaneously, the 8086 places data on the multiplexed bus
which is then latched into the first rank of the AD669 with the
rising edge of the
WR pulse. Care should be taken to prevent
excessive delays through the decoder potentially resulting in a
violation of the AD669 data hold time (t
DH
).
The same procedure is repeated until all three AD669s have had
their first rank latches loaded with the desired data. A final write
command to the LDAC address results in a high-going pulse
that causes the second rank latches of all the AD669s to become
transparent. The falling edge of LDAC latches the data from the
first rank until the next update. This scheme is easily expanded
to include as many AD669s as required.
8086
DGND
+5V
ALE
WR
M/
I0
ADDRESS
DECODE
LDAC
CS1 CS2 CS3
DGND
L1
DB0 – DB15
LDAC
AD669
V
OUT
AD0 – AD15
DGND
L1
DB0 – DB15
LDAC
AD669
V
OUT
DGND
L1
DB0 – DB15
LDAC
AD669
CS
V
OUT
V
LL
V
LL
V
LL
V
LL
CS
CS
Figure 10. 8086-to-AD669 Interface
8-BIT MICROPROCESSOR INTERFACE
The AD669 can easily be operated with an 8-bit bus by the ad-
dition of an octal latch. The 16-bit first rank register is loaded
from the 8-bit bus as two bytes. Figure 11 shows the configura-
tion when using a 74HC573 octal latch.
The eight most significant bits are latched into the 74HC573 by
setting the “latch enable” control line low. The eight least sig-
nificant bits are then placed onto the bus. Now all sixteen bits
can be simultaneously loaded into the first rank register of the
AD669 by setting
CS and L1 low.
8-BIT µP
AND
CONTROL
D7
D0
D7
D0
Q7
Q0
74HC573
11
LDAC
MSB
DB8
DB7
LSB
AD669
CS1 L1
Figure 11. Connections for 8-Bit Bus Interface
AD669
REV. A
–11–
NOISE
In high resolution systems, noise is often the limiting factor. A
16-bit DAC with a 10 volt span has an LSB size of 153 µV
(–96 dB). Therefore, the noise floor must remain below this
level in the frequency range of interest. The AD669’s noise
spectral density is shown in Figures 12 and 13. Figure 12 shows
the DAC output noise voltage spectral density for a 20 V span
excluding the reference. This figure shows the l/f corner frequency
at 100 Hz and the wideband noise to be below 120 nV/
Hz.
Figure 13 shows the reference noise voltage spectral density.
This figure shows the reference wideband noise to be below
125 nV/
Hz.
1000
1
11M
100
10
10 100k10k1k100 10M
FREQUENCY – Hz
NOISE VOLTAGE – nV/ Hz
Figure 12. DAC Output Noise Voltage Spectral Density
1000
1
1
1M
100
10
10 100k
10k1k100
10M
FREQUENCY – Hz
NOISE VOLTAGE – nV/ Hz
Figure 13. Reference Noise Voltage Spectral Density
BOARD LAYOUT
Designing with high resolution data converters requires careful
attention to board layout. Trace impedance is the first issue. A
306 µA current through a 0.5 trace will develop a voltage
drop of 153 µV, which is 1 LSB at the 16-bit level for a 10 V
full-scale span. In addition to ground drops, inductive and ca-
pacitive coupling need to be considered, especially when high
accuracy analog signals share the same board with digital sig-
nals. Finally, power supplies need to be decoupled in order to
filter out ac noise.
Analog and digital signals should not share a common path.
Each signal should have an appropriate analog or digital return
routed close to it. Using this approach, signal loops enclose a
small area, minimizing the inductive coupling of noise. Wide PC
tracks, large gauge wire, and ground planes are highly recom-
mended to provide low impedance signal paths. Separate analog
and digital ground planes should also be utilized, with a single
interconnection point to minimize ground loops. Analog signals
should be routed as far as possible from digital signals and
should cross them at right angles.
One feature that the AD669 incorporates to help the user layout
is the analog pins (V
CC
, V
EE
, REF OUT, REF IN, SPAN/BIP
OFFSET, V
OUT
and AGND) are adjacent to help isolate analog
signals from digital signals.
SUPPLY DECOUPLING
The AD669 power supplies should be well filtered, well regu-
lated, and free from high frequency noise. Switching power sup-
plies are not recommended due to their tendency to generate
spikes which can induce noise in the analog system.
Decoupling capacitors should be used in very close layout prox-
imity between all power supply pins and ground. A 10 µF tanta-
lum capacitor in parallel with a 0.1 µF ceramic capacitor
provides adequate decoupling. V
CC
and V
EE
should be bypassed
to analog ground, while V
LL
should be decoupled to digital
ground.
An effort should be made to minimize the trace length between
the capacitor leads and the respective converter power supply
and common pins. The circuit layout should attempt to locate
the AD669, associated analog circuitry and interconnections as
far as possible from logic circuitry. A solid analog ground plane
around the AD669 will isolate large switching ground currents.
For these reasons, the use of wire wrap circuit construction
is not recommended; careful printed circuit construction is
preferred.
GROUNDING
The AD669 has two pins, designated analog ground (AGND)
and digital ground (DGND.) The analog ground pin is the
“high quality” ground reference point for the device. Any exter-
nal loads on the output of the AD669 should be returned to
analog ground. If an external reference is used, this should also
be returned to the analog ground.
If a single AD669 is used with separate analog and digital
ground planes, connect the analog ground plane to AGND and
the digital ground plane to DGND keeping lead lengths as short
as possible. Then connect AGND and DGND together at the
AD669. If multiple AD669s are used or the AD669 shares ana-
log supplies with other components, connect the analog and
digital returns together once at the power supplies rather than at
each chip. This single interconnection of grounds prevents large
ground loops and consequently prevents digital currents from
flowing through the analog ground.

AD669SQ

Mfr. #:
Manufacturer:
Analog Devices Inc.
Description:
Digital to Analog Converters - DAC IC MONO 16-BIT
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union