QUICK START GUIDE FOR DEMONSTRATION CIRCUIT 745
16-BIT DIFFERENTIAL INPUT DELTA SIGMA ADC
4
EXPERIMENTS
INPUT NOISE
One of the characteristics of the LTC2433-1 is that
the 1.45µV input noise floor is far below the quanti-
zation level of 76uV when a 5-volt reference is used.
This means that the output will be stable if the input
noise level is also below 76uV. In this sense, the
LTC2433-1 is a true 16 effective bit part, whereas
many 16-bit SAR converters have several LSBs of
noise.
Solder a short wire from the IN- turret post to the
IN+ turret post. Noise should be below the quanti-
zation level of the LTC2433-1. This will result in a
noise reading of zero on the control software.
Applying an external reference of 100mV results in
a 1.52µV LSB, and the noise reading will be ap-
proximately 250ppm of 100mV, or 2.5µV. This is
slightly higher than the 1.45µV input thermal be-
cause quantization noise is also a factor.
COMMON MODE REJECTION
Tie the two inputs (still connected together) to
ground through a short wire and note the indicated
voltage. Tie the inputs to REF+; the difference
should be less than 1LSB due to the 130 dB CMRR
of the LTC2433-1.
BIPOLAR SYMMETRY
To demonstrate the symmetry of the ADCs transfer
function, connect a stable, low noise, floating volt-
age source (with a voltage less than Vref/2) from
IN+ to IN- and note the indicated voltage. Reverse
the polarity; the indicated voltage will typically be
within one LSB of the first reading multiplied by –1.
One convenient voltage source for this experiment
is a single alkaline battery. While a battery has fairly
low noise, it is sensitive to temperature drift. It is
best to use a large (D-size) battery that is insulated
from air currents. A better source is a battery pow-
ered series reference such as the LT1790. This part
is available with output voltages of 1.25V, 2.048V,
2.5V, 3V, 3.3V, 4.096V and 5V.
INPUT NORMAL MODE REJECTION
The LTC2433-1’s SINC4 digital filter is trimmed to
strongly reject 50 or 60Hz line noise when operated
with the internal conversion clock. To measure in-
put normal mode rejection, connect IN- to a 2.5 volt
source such as an LT1790-2.5 reference or a power
supply. Apply a 10Hz, 2V peak-to-peak sine wave to
IN+ through a 1uF capacitor. No DC bias is required
because the 2-3MΩ input impedance of the
LTC2433-1 tends to self-bias the input to mid-
reference (see datasheet applications information
for details.)
Start taking data. The input noise will be quite large,
and the graph of output vs. time should show large
variations.
Next, slowly increase the frequency to 60Hz. The
noise should be almost undetectable in the graph.
Note that the indicated noise in ppm may still be
above that of the datasheet specification because
the inputs are not connected to a DC source.