Application Hints (Continued)
advantage of this method. Compensation involves a capaci-
tor to V
EE
on pin 16, using the values of the previous
paragraph. The negative reference voltage must be at least
4V above the V
EE
supply. Bipolar input signals may be
handled by connecting R14 to a positive reference voltage
equal to the peak positive input level at pin 15.
When a DC reference voltage is used, capacitive bypass to
ground is recommended. The 5V logic supply is not recom-
mended as a reference voltage. If a well regulated 5V supply
which drives logic is to be used as the reference, R14 should
be decoupled by connecting it to 5V through another resistor
and bypassing the junction of the 2 resistors with 0.1 µF to
ground. For reference voltages greater than 5V, a clamp
diode is recommended between pin 14 and ground.
If pin 14 is driven by a high impedance such as a transistor
current source, none of the above compensation methods
apply and the amplifier must be heavily compensated, de-
creasing the overall bandwidth.
OUTPUT VOLTAGE RANGE
The voltage on pin 4 is restricted to a range of −0.55 to 0.4V
when V
EE
= −5V due to the current switching methods
employed in the DAC0808.
The negative output voltage compliance of the DAC0808 is
extended to −5V where the negative supply voltage is more
negative than −10V. Using a full-scale current of 1.992 mA
and load resistor of 2.5 kΩbetween pin 4 and ground will
yield a voltage output of 256 levels between 0 and −4.980V.
Floating pin 1 does not affect the converter speed or power
dissipation. However, the value of the load resistor deter-
mines the switching time due to increased voltage swing.
Values of R
L
up to 500Ωdo not significantly affect perfor-
mance, but a 2.5 kΩload increases worst-case settling time
to 1.2 µs (when all bits are switched ON). Refer to the
subsequent text section on Settling Time for more details on
output loading.
OUTPUT CURRENT RANGE
The output current maximum rating of 4.2 mA may be used
only for negative supply voltages more negative than −8V,
due to the increased voltage drop across the resistors in the
reference current amplifier.
ACCURACY
Absolute accuracy is the measure of each output current
level with respect to its intended value, and is dependent
upon relative accuracy and full-scale current drift. Relative
accuracy is the measure of each output current level as a
fraction of the full-scale current. The relative accuracy of the
DAC0808 is essentially constant with temperature due to the
excellent temperature tracking of the monolithic resistor lad-
der. The reference current may drift with temperature, caus-
ing a change in the absolute accuracy of output current.
However, the DAC0808 has a very low full-scale current drift
with temperature.
The DAC0808 series is guaranteed accurate to within ±
1
⁄
2
LSB at a full-scale output current of 1.992 mA. This corre-
sponds to a reference amplifier output current drive to the
ladder network of 2 mA, with the loss of 1 LSB (8 µA) which
is the ladder remainder shunted to ground. The input current
to pin 14 has a guaranteed value of between 1.9 and 2.1 mA,
allowing some mismatch in the NPN current source pair. The
accuracy test circuit is shown in
Figure 4
. The 12-bit con-
verter is calibrated for a full-scale output current of 1.992
mA. This is an optional step since the DAC0808 accuracy is
essentially the same between 1.5 and 2.5 mA. Then the
DAC0808 circuits’ full-scale current is trimmed to the same
value with R14 so that a zero value appears at the error
amplifier output. The counter is activated and the error band
may be displayed on an oscilloscope, detected by compara-
tors, or stored in a peak detector.
Two 8-bit D-to-A converters may not be used to construct a
16-bit accuracy D-to-A converter. 16-bit accuracy implies a
total error of ±
1
⁄
2
of one part in 65,536 or ±0.00076%, which
is much more accurate than the ±0.019% specification pro-
vided by the DAC0808.
MULTIPLYING ACCURACY
The DAC0808 may be used in the multiplying mode with
8-bit accuracy when the reference current is varied over a
range of 256:1. If the reference current in the multiplying
mode ranges from 16 µA to 4 mA, the additional error
contributions are less than 1.6 µA. This is well within 8-bit
accuracy when referred to full-scale.
A monotonic converter is one which supplies an increase in
current for each increment in the binary word. Typically, the
DAC0808 is monotonic for all values of reference current
above 0.5 mA. The recommended range for operation with a
DC reference current is 0.5 to 4 mA.
SETTLING TIME
The worst-case switching condition occurs when all bits are
switched ON, which corresponds to a low-to-high transition
for all bits. This time is typically 150 ns for settling to within
±
1
⁄
2
LSB, for 8-bit accuracy, and 100 ns to
1
⁄
2
LSB for 7 and
6-bit accuracy. The turn OFF is typically under 100 ns.These
times apply when R
L
≤500Ωand C
O
≤25 pF.
Extra care must be taken in board layout since this is usually
the dominant factor in satisfactory test results when measur-
ing settling time. Short leads, 100 µF supply bypassing for
low frequencies, and minimum scope lead length are all
mandatory.
DAC0808
www.national.com 10