REV. 0
–40–
AD7708/AD7718
Calibration
The AD7708/AD7718 provides four calibration modes that can
be programmed via the mode bits in the mode register. One of
the major benefits of the AD7708/AD7718 is that it is factory-
calibrated with chopping enabled as part of the final test process
with the generated coefficients stored within the ADC. At power-
on, the factory gain calibration coefficients are automatically
loaded to the gain calibration registers on the AD7708/AD7718.
This gives excellent offset and drift performance and it is
envisaged that in the majority of applications the user will not
need to perform any field calibrations.
Also, because factory
gain calibration coefficients (generated at 25°C ambient) are
automatically present at power-on, an internal full-scale calibration
will only be required if the part is being operated at temperatures
significantly different from 25°C.
When chopping is disabled (CHOP =1) the AD7708/AD7718
requires an offset calibration or new calibration coefficients on
range changing or when significant temperature changes occur
as the signal chain is no longer chopped and offset and drift
errors are no longer removed as part of the conversion process.
The factory-calibration values for any one channel will be over-
written if any one of the four calibration options is initiated.
The AD7708/AD7718 offers “internal” or “system” calibration
facilities. For full calibration to occur, the calibration logic must
record the modulator output for two different input conditions.
These are “zero-scale” and “full-scale” points. These points
are derived by performing a conversion on the different input
voltages provided to the input of the modulator during calibration.
The result of the “zero-scale” calibration conversion is stored in
the Offset Calibration Registers for the appropriate channel.
The result of the “full-scale” calibration conversion is stored
in the Gain Calibration Registers. With these readings, the
calibration logic can calculate the offset and the gain slope for
the input-to-output transfer function of the converter. During an
“internal” zero-scale or full-scale calibration, the respective
“zero” input and “full-scale” input are automatically connected
to the ADC input pins internally to the device. A “system” cali-
bration, however, expects the system zero-scale and system
full-scale voltages to be applied to the external ADC pins
before the calibration mode is initiated. In this way external
ADC errors are taken into account and minimized as a result
of system calibration. It should also be noted that to optimize
calibration accuracy, all AD7708/AD7718 ADC calibrations
are carried out automatically at the slowest update rate with
chop enabled. When chop mode is disabled calibrations are
carried out at the update rate defined by the SF word in the
filter register.
Internally in the AD7708/AD7718, the coefficients are normalized
before being used to scale the words coming out of the digital
filter. The offset calibration coefficient is subtracted from the
result prior to the multiplication by the gain coefficient. With
chopping disabled AD7708/AD7718 ADC specifications will
only apply after a zero-scale calibration at the operating point of
interest. From an operational point of view, a calibration should
be treated like another ADC conversion. A zero-scale calibration
(if required) should always be carried out before a full-scale
calibration. System software should monitor the RDY bit in the
STATUS register to determine end of calibration via a polling
sequence or interrupt driven routine.
Grounding and Layout
Since the analog inputs and reference inputs are differential,
most of the voltages in the analog modulator are common-mode
voltages. The excellent common-mode rejection of the part will
remove common-mode noise on these inputs. The analog and
digital supplies to the AD7708/AD7718 are independent and
separately pinned out to minimize coupling between the analog
and digital sections of the device. The AD7708/AD7718 can be
operated with 5 V analog and 3 V digital supplies or vice versa.
The digital filter will provide rejection of broadband noise on
the power supplies, except at integer multiples of the modulator
sampling frequency. The digital filter also removes noise from
the analog and reference inputs provided these noise sources do
not saturate the analog modulator. As a result, the AD7708/
AD7718 is more immune to noise interference than a conventional
high-resolution converter. However, because the resolution of the
AD7708/AD7718 is so high and the noise
levels from the
converter so low, care must be taken with regard to grounding
and layout.
The printed circuit board that houses the ADC should be designed
so the analog and digital sections are separated and confined to
certain areas of the board. This facilitates the use of ground planes
that can be easily separated. A minimum etch technique is
generally best for ground planes as it gives the best shielding.
Although the AD7708/AD7718 has separate pins for analog and
digital ground, the AGND and DGND pins are tied together
internally via the substrate. Therefore, the user must not tie
these two pins to separate ground planes unless the ground
planes are connected together near the AD7708/AD7718.
In systems where the AGND and DGND are connected some-
where else in the system, i.e., the systems power supply, they
should not be connected again at the AD7708/AD7718 or a
ground loop will result. In these situations it is recommended that
ground pins of the AD7708/AD7718 be tied to the AGND plane.
In any layout it is implicit that the user keep in mind the flow of
currents in the system, ensuring that the paths for all currents
are as close as possible to the paths the currents took to reach
their destinations. Avoid forcing digital currents to flow through
the AGND.
Avoid running digital lines under the device as these will couple
noise onto the die. The analog ground plane should be allowed
to run under the AD7708/AD7718 to prevent noise coupling.
The power supply lines to the AD7708/AD7718 should use as
wide a trace as possible to provide low impedance paths and
reduce the effects of glitches on the power supply line. Fast
switching signals like clocks should be shielded with digital
ground to avoid radiating noise to other sections of the board
and clock signals should never be run near the analog inputs.
Avoid crossover of digital and analog signals. Traces on opposite
sides of the board should run at right angles to each other. This will
reduce the effects of feedthrough through the board. A microstrip
technique is by far the best, but is not always possible with a
double-sided board. In this technique, the component side of
the board is dedicated to ground planes while signals are placed
on the solder side.
