W. Stephen Woodward
University of North Carolina, Venable Hall, CB3300, Chapel Hill, NC 27599-3300; (919) 966-1358; fax (919) 962-1254;
e-mail: woodward@unc.edu
Chuck Wojslaw
Xicor Inc., 1511 Buckeye Dr., Milpitas, CA 95035; e-mail: cwojslaw@xicor.com
ELECTRONIC DESIGN • May 29, 2000から引用
A mong temperature transducers,
the platinum resistance temperature detector (PRTD) is generally
accepted as the “gold standard.” PRTDs
are available with interchangeability accuracies as tight as 0.1°C over operational ranges extending from −200°C
(colder than liquid nitrogen) to
+850°C (hotter than liquid aluminum). As a result, PRTDs are ubiquitous in aviation, environmental, industrial, and scientific instrumentation.
The PRTD temperature response consists of resistance variations on the order of only tenths of ohms/°C, so strict
attention must be paid to the effects of
transducer lead-wire resistance. Either
the signal-conditioning circuitry must
be packaged so that it’s close to the sensor (making the lead wires short and
their resistance insignificant), or
Kelvin-sensing arrangements will be
needed to explicitly sense and cancel
out wiring resistance. The excitation
current must be kept at less than 1 mA
or excessive I2R PRTD power dissipation will cause unacceptable self-heating measurement errors.
The combination of low excitation
currents and small resistance changes
means that the PRTD signal will typically be on the order of only tens of microvolts/°C. As a result, stable highgain dc amplification in the signal
chain is required. Also, although the
PRTD temperature coefficient is ex
tremely stable and reproducible, it's
only “reasonably” invariant with temperature. Thus, the PRTD’s response is
significantly nonlinear and deviates
from a linear approximation by tens of
degrees over large temperature spans.
The accurate measurement of temperature over wide ranges therefore depends on provisions for linearization
of the PRTD signal. These design considerations, which are incorporated in
the circuit in the figure, result in a precision thermometer with output span
of −1 V to +3.5 V, corresponding to a
temperature range of −100 to +350°C.
The maximum error over this span can
be adjusted to ±0.02°C at 0°C and
±0.05°C elsewhere. Current excitation
(approximately 250 µA) for the PRTD
is sourced by the 2.5-V voltage reference VR1 via R1. A 256-tap digitally
controlled potentiometer (DCP1)
provides automated adjustment of the
thermometer scale factor and span.
Amplifier A1 is used to scale up the
~100-µV/°C raw PRTD temperature
signal to 0.01 V/°C. The DCP2 network implements a high-resolution
zero adjustment in which each increment in DCP2’s setting will result in a
200-µV shift in A1’s output—a 0.02°C
zero adjustment. The symmetry of the
R6-R9 network surrounding DCP2
causes zero adjustment to have no effect on A1’s gain and, therefore, no effect on the thermometer’s span/scale
factor. Likewise, performing span adjustments via trimming of the VR1 reference prevents interaction betwee
DCP1 and the zero calibration point
established by DCP2.
Positive feedback provided by R2 linearizes the thermometer’s response
curve by providing the Thevenin equivalent of a negative resistance (−2064
Ω) in parallel with R1. This ploy introduces a positive slope (roughly
+0.016%/°C), which effectively cancels the tendency of the PRTD temperature coefficient to decline with increasing temperature. The result is better
than a factor of 100 improvement in
linearity over the raw PRTD response.
The illustrated circuit implements a
signal-conditioned precision temperature sensor that’s compatible (thanks
to DCP1 and DCP2) with full automation of the calibration process, low in
total power draw, and low in cost. Substituting digitally controlled potentiometers in place of manually adjusted trimmers can dramatically
improve reliability. At the same time, it
allows packaging of the signal conditioning circuitry that’s compatible
with rugged encapsulation appropriate for mounting close to the PRTD itself. This avoids problems related to
lead length and resistance.
