Accuracy error of a D/A converter is the difference between the
analog output that is expected when a given digital code is
applied and the output that is actually measured with that code
applied to the converter. Accuracy error can be caused by gain
error, zero error, linearity error, or any combination of the three.
Of these three specifications, the linearity error specification is
the most important since it cannot be corrected. Linearity error
is specified over its entire temperature range. This means that
the analog output will not vary by more than its maximum
specification, from an ideal straight line drawn between the
end points (inputs all “1”s and all “0”s) over the specified
Differential linearity error of a D/A converter is the deviation
from an ideal 1 LSB voltage change from one adjacent output
state to the next. A differential linearity error specification of
± 1/2 LSB means that the output voltage step sizes can range
from 1/2 LSB to 1 1/2 LSB when the input changes from one
adjacent input state to the next.
A measure of the change in the full scale range output over
temperature expressed in parts per million of full scale range
per °C (ppm of FSR/°C). Gain drift is established by: 1) testing
the end point differences for each ADDAC80 model at the
lowest operating temperature, 25°C and the highest operating
temperature; 2) calculating the gain error with respect to the
25°C value and; 3) dividing by the temperature change.
A measure of the actual change in output with all “1”s on the
input over the specified temperature range. The maximum
change in offset is referenced to the offset at 25°C and is
divided by the temperature range. This drift is expressed in
parts per million of full scale range per °C (ppm of FSR/°C).
Settling time for each model is the total time (including slew
time) required for the output to settle within an error band
around its final value after a change in input.
Voltage Output Models
Three settling times are specified to 0.01% of full scale range
(FSR); two for maximum full scale range changes of 20 V, 10 V
and one for a 1 LSB change. The 1 LSB change is measured at
the major carry (0 1 1 1 . . . 1 1 to 1 0 0 0 . . . 0 0), the point at
which the worst case settling time occurs. The settling time
characteristic depends on the compensation capacitor selected,
the optimum value is 25 pF as shown in Figure 3a.
Current Output Models
Two settling times are specified to ± 0.01% of FSR. Each is given
for current models connected with two different resistive loads:
10 Ω to 100 Ω and 1000 Ω to 1875 Ω. Internal resistors are provided
for connecting nominal load resistances of approximately 1000 Ω
to 1800 Ω for output voltage ranges of ±1 V and 0 V to –2 V.
Figure 3a. Voltage Model Settling Time Circuit
Figure 3b. Voltage Model Settling Time CF = 25 pF
POWER SUPPLY SENSITIVITY
Power supply sensitivity is a measure of the effect of a power
supply change on the D/A converter output. It is defined as a
percent of FSR per percent of change in either the positive or
negative supplies about the nominal power supply voltages.
All models are supplied with an internal 6.3 V reference voltage
supply. This voltage (Pin 24) is accurate to ± 1% and must be
connected to the Reference Input (Pin 16) for specified opera-
tion. This reference may also be used externally with external
current drain limited to 2.5 mA. An external buffer amplifier is
recommended if this reference is to be used to drive other sys-
tem components. Otherwise, variations in the load driven by the
reference will result in gain variations. All gain adjustments
should be made under constant load conditions.
ANALYZING DEVICE ACCURACY OVER THE
For the purposes of temperature drift analysis, the major device
components are shown in Figure 4. The reference element and
buffer amplifier drifts are combined to give the total reference
temperature coefficient. The input reference current to the
DAC, IREF, is developed from the internal reference and will
show the same drift rate as the reference voltage. The DAC
output current, IDAC, which is a function of the digital input
codes, is designed to track IREF; if there is a slight mismatch in
these currents over temperature, it will contribute to the gain
T.C. The bipolar offset resistor, RBP, and gain setting resistor,
RGAIN, also have temperature coefficients that contribute to
system drift errors. The input offset voltage drift of the output
amplifier, OA, also contributes a small error.