Cirrus Logic CS5529 - Manual (Page 3)

CS5525/6/9 FAQ
DS202TB1 3
12) How can the gain be calibrated if a full-scale
signal is not available?
The CS5525/26 can be gain calibrated with
some input signal other than full scale. For ex-
ample, when the converter is reset, the gain reg-
ister’s calibration word is 1.0 (decimal). If a
signal representing ten percent of full scale
reads three percent less than it should, the value
in the gain register can be scaled up by three
percent. Gain accuracy can be improved if out-
put words are averaged while using this tech-
nique. Use caution when a calibration signal
less than full scale is being used. If the transfer
function of the transducer being used to gener-
ate the ten percent signal happens to have a ma-
jor nonlinearity at the point at which calibration
is being performed, it will cause the rest of the
transfer function to be incorrect.
13) Why does the offset move when the CS5525/26
with a 2.5 V reference, is calibrated several
times? What can be done to prevent this?
The CS5526 is a 20-bit ADC with inherent
Gaussian thermal and quantization noise asso-
ciated with each conversion. Therefore, every
time the converter is calibrated, a different off-
set calibration output has a chance of occurring.
By averaging conversions, the peak-to-peak
noise can be reduced by a factor of 1/sqrt(n)
(where n is the number of samples taken). The
offset register can be accessed after calibration,
and the offset uncertainty of a converter can al-
most be eliminated (to 1 code) by averaging.
The CS5525 (16-bits) always has 1 count of
variability, even if averaged, because the noise
and calibration can occur at a boundary be-
tween two codes. If the calibration code is on
the boundary the random noise could toggle the
offset between the two codes.
14) Is a different calibration required for each gain
getting?
For maximum accuracy, calibrations should be
performed for offset and gain for each gain set-
ting. If a factory calibration is performed using
a system calibration, the offset and gain register
contents can be read by the system microcon-
troller and stored in EEPROM. These same cal-
ibration words can then be uploaded into the
offset and gain registers of the converter when
power is re-applied to the system, or when the
gain range is changed.
15) What is the advantage of performing calibra-
tions at lower output word rates?
Calibrations are performed at the output word
rate selected by the WR2-WR0 bits of the con-
figuration register. Since higher word rates re-
sult in conversion words with more peak-to-
peak noise, it is better to calibrate at lower out-
put word rates. To minimize the digital noise
near the device, the user should wait for each
calibration step to be completed before reading
or writing to the serial port.
16) How can I get the best noise performance from
the CS5525/26?
Use the bipolar mode or increase the reference
voltage, since each of these increase the size of
the LSB.
17) If the charge pump is engaged, how do I ensure
that the converter and its external components
are intrinsically safe?
Intrinsic safety prohibits the use of electrolytic
(or bipolar) capacitors thus limiting the use of
certain size capacitors. Although a 10
µ F cap. is
recommended for the charge pump, two
0.47
µ F ceramic caps in parallel can be used.