MAX4460/MAX4461/MAX4462
SOT23, 3V/5V, Single-Supply, Rail-to-Rail
Instrumentation Amplifiers
______________________________________________________________________________________
15
The rotational nature of gain error, and the fact that it is
pivoted around point E in Figure 5, shows that gain-
error contribution to the total output voltage error is
directly proportional to the input voltage. At zero input
voltage, the error contribution of gain error is zero, i.e.,
the total deviation from the origin (the expected zero
output value) is only due to end-points OE and nonlin-
earity error at zero value of input (segment EZ on the
vertical axis).
The nonlinearity is the maximum deviation from a
straight line, and the end-point nonlinearity is the devia-
tion from the end-point line. As shown in Figure 5, it is
likely that two nonlinearities are encountered, one posi-
tive and the other a negative nonlinearity error, shown
as NL+ and NL- in Figure 5.
Generally, NL+ and NL- have different values and this
remains the case if the device is calibrated (trimmed)
for end-points errors (which means changing the gain
of the instrumentation amplifier in such a way that the
slope of line AB becomes equal to that of CD, and the
offset becomes trimmed such that OE vanishes to
zero). This is an undesirable situation when nonlinearity
is of prime interest.
The straight line shown in Figure 6 is in parallel to end-
points line AB and has a Y intercept of OS on the verti-
cal axis. This line is a shifted end-points line such that
the positive and negative nonlinearity errors with
respect to this line are equal. For this reason, the line is
called the best straight line (BSL). Maxim internally
trims the MAX4460/MAX4461/MAX4462 with respect to
this line (changing the gain slope to be as close as
possible to the slope of the ideal line and trimming the
offset such that OS gets as close to the origin as possi-
ble) to minimize all the errors. The total accuracy error
is still the summation of the gain error, nonlinearity, and
offset errors.
As an example, assume the following specification for
an instrumentation amplifier:
Gain = 10
GE = 0.15%
Offset (BSL) = 250V
NL = 0.05%
VDIF (input) = -100mV to +100mV
What is the maximum total error associated with the
GE, offset (BSL), and NL? With a differential input range
of -0.1V to +0.1V and a gain of 10, the output voltage
assumes a range of -1V to +1V, i.e., a total full-scale
range of 2V.
The individual errors are as follows:
GE = (0.15%) (10) (100mV) = 1.5mV
Offset (BSL) = (250V) (10) = 2.5mV
NL = (0.05%) (2V) = 1mV
Maximum Total Error = 1.5mV + 2.5mV + 1mV
= 5mV
So, the absolute value of the output voltage, consider-
ing the above errors, would be at worst case between
0.995V to 1.005V. Note that other important parameters
such as PSRR, CMRR, and noise also contribute to the
total error in instrumentation applications. They are not
considered here.
VIN
VOUT
ACTUAL CURVE
END-POINT LINE
BSL LINE
NL+
NL-
0
E
Z
B
S
A
NL+ = NL- = NL
NLBSL (%) = (NL / FULL-SCALE OUTPUT RANGE) X 100
OFFSET (BSL) = OSL
GAIN AND OFFSET WILL BE FACTORY-TRIMMED FOR BEST STRAIGHT LINE
Figure 6. To Minimize Nonlinearity Error, the MAX4460/MAX4461/
MAX4462 are Internally Trimmed to Adjust Gain and Offset for the
Best Straight Line so NL- = NL+