AD590
The AD590 is basically a PTAT (proportional to absolute
temperature)
current regulator. That is, the output current is
equal to a scale factor times the temperature of the sensor in
degrees Kelvin. This scale factor is trimmed to 1 µA/K at the
factory, by adjusting the indicated temperature (that is, the
output current) to agree with the actual temperature. This is
done with 5 V across the device at a temperature within a few
degrees of 25°C (298.2K). The device is then packaged and
tested for accuracy over temperature.
5V +
+
AD590
–
R
100Ω
950Ω
–
–
+
Figure 7. One Temperature Trim
CALIBRATION ERROR
At final factory test, the difference between the indicated
temperature and the actual temperature is called the calibration
error. Since this is a scale factory error, its contribution to the
total error of the device is PTAT. For example, the effect of the
1°C specified maximum error of the AD590L varies from
0.73°C at –55°C to 1.42°C at 150°C. Figure 6 shows how an
exaggerated calibration error would vary from the ideal over
temperature.
ERROR VERSUS TEMPERATURE: WITH
CALIBRATION ERROR TRIMMED OUT
Each AD590 is tested for error over the temperature range with
the calibration error trimmed out. This specification could also
be called the “variance from PTAT,” because it is the maximum
difference between the actual current over temperature and a
PTAT multiplication of the actual current at 25°C. This error
consists of a slope error and some curvature, mostly at the
temperature extremes. Figure 8 shows a typical AD590K
temperature curve before and after calibration error trimming.
ABSOLUTE ERROR (°C)
ACTUAL
TRANSFER
FUNCTION
2
BEFORE
CALIBRATION
TRIM
CALIBRATION
ERROR
0
AFTER
CALIBRATION
TRIM
00533-C-008
I
OUT
(µA)
I
ACTUAL
CALIBRATION
ERROR
298.2
IDEAL
TRANSFER
FUNCTION
00533-C-006
–2
298.2
TEMPERATURE (°K)
–55
TEMPERATURE (°C)
00533-C-007
V
T
= 1mV/K
150
Figure 6. Calibration Error vs. Temperature
Figure 8. Effect to Scale Factor Trim on Accuracy
The calibration error is a primary contributor to maximum
total error in all AD590 grades. However, since it is a scale factor
error, it is particularly easy to trim. Figure 7 shows the most
elementary way of accomplishing this. To trim this circuit, the
temperature of the AD590 is measured by a reference
temperature sensor and R is trimmed so that V
T
= 1 mV/K at
that temperature. Note that when this error is trimmed out at
one temperature, its effect is zero over the entire temperature
range. In most applications there is a current-to-voltage
conversion resistor (or, as with a current input ADC, a
reference) that can be trimmed for scale factor adjustment.
ERROR VERSUS TEMPERATURE: NO USER TRIMS
Using the AD590 by simply measuring the current, the total
error is the variance from PTAT, described above, plus the effect
of the calibration error over temperature. For example, the
AD590L maximum total error varies from 2.33°C at –55°C to
3.02°C at 150°C. For simplicity, only the large figure is shown on
the specification page.
NONLINEARITY
Nonlinearity as it applies to the AD590 is the maximum
deviation of current over temperature from a best-fit straight
line. The nonlinearity of the AD590 over the −55°C to +150°C
range is superior to all conventional electrical temperature
sensors such as thermocouples, RTDs, and thermistors. Figure 9
shows the nonlinearity of the typical AD590K from Figure 8.
1
T(°C) = T(K) –273.2. Zero on the Kelvin scale is “absolute zero”; there is no
lower temperature.
Rev. C | Page 7 of 16
AD [ ANALOG DEVICES ]
