Introduction

Temp Sensor Circuit Implementation Details

Fundamentally, the DS1925 uses p-n junctions of a bipolar transistors (BJT) to measure temperature. These BJT p-n junctions are electrically equivalent to a circuit element known as a diode. The electrical characteristics of the BJT diodes have predictable voltage-temperature dependencies that are described with mathematical equations enabling them to be used as temperature sensors. The DS1925 uses two BJT diodes, with voltages Vbe1   and Vbe2  , as shown in Figure 1.

A constant current through each diode in Figure 1 produces the temperature dependence of the base-emitter voltage (Vbe  ) described by Equation 1. Since the bias current sources are different, the results are two distinct base-emitter voltages. The ∆Vbe   is the difference of the base-emitter voltage measurements from the two excitation currents.

∆Vbe= K*Tq* ln⁡(Ic1Ic2)

Eq. (1)

∆Vbe=Vbe1-Vbe2

Eq. (2)

Equation 1 is for an ideal transistor, so a non-ideality factor (η  ) is needed for real world usage.  Applying this adjustment, Equation 3 is the fundamental BJT base-emitter diode voltage equation.

∆Vbe= K*T*ηq* ln⁡(m)

Eq. (3)

m= Ic1Ic2    (constant and known by design)

K   = Boltzmann's constant

T   = circuit operating temperature in Kelvin

q   = constant charge on an electron

With Equation 3, from a known ∆Vbe   the temperature T   of the DS1925 sensor, and therefore the DS1925 operating condition, can be precisely determined.

Advanced Single-Point Calibration

Unlike traditional calibration techniques requiring multiple temperature points and precise references, this advanced single-point calibration methodology does not require either one. This section presents the principles and methodology used to calibrate the DS1925’s temperature sensor, which does not require a forced IC temperature or a reference temperature sensor to measure the actual IC temperature. Instead, it relies on applying a high-precision external voltage to the device under test (DUT) to determine its temperature followed by adjusting, and therefore calibrating, it’s Vbe   to an ideal characteristic². Once performed at temperature T  , this calibration allows Vbe   to be precisely known across all operating temperatures. There are two steps to the calibration process:

1. Through application of a precision voltage by test equipment, determine the actual temperature (T  ) of the DS1925 IC.
2. Adjust and calibrate Vbe   such that it’s value at T   is equal to ideal.

Figure 2 shows the DS1925’s internal analog-to-digital converter (ADC) and the various inputs. Equation 3 has two unknowns: temperature and ∆Vbe . We can configure the DS1925 ADC to select Vbe  1 & 2 as its references and apply a precision external voltage (Vext  ).

The resultant ADC code from applying an external voltage Vext   is:

ADC Code= ∆VbeVext+∆Vbe

Eq. (4)

Since the ADC code was read from the device and the externally applied voltage Vext 

is known, we can solve for ∆Vbe   in Equation 4.

Next, with ∆Vbe   known, solving for temperature in Equation 3 gives:

T= ∆Vbe*qK*η* ln⁡(m)

Eq. (5)

This temperature value T   determined from Equation 5 is the precise operating temperature of the DS1925 at this test step and used for the next step in the calibration process.

With T   known, a precision ∆Vbe   is applied externally (∆Vext  ) and the DS1925 ADC is next configured to measure internal Vbe   according to Equation 6.

ADC Code= ∆VextVbe+∆Vext

Eq. (6)

However, relative to an ideal BJT diode characteristic that is required for operation of the temperature sensor circuit, due to wafer process variations this Vbe   has a temperature error and needs to be calibrated.  Shown in Figure 3 is the natural voltage (Vbe)  vs. temperature characteristic of a BJT diode under a constant current condition.  There are three Vbe   curves plotted: ideal, Vbe   with positive error, and Vbe   with negative error.  During this test step the uncalibrated Vbe   is obtained from equation 6 and an adjustment is performed to effectively move (i.e. calibrate) the measured Vbe  , positively or negatively as needed, to the ideal curve.  As noted previously, once Vbe   is calibrated to ideal at T  , it operates with the ideal characteristic across all temperatures.

Figure 3: Vbe   vs. Temperature

Verification

Calibrated units were swept in a temperature chamber across the -40C to +85C range. Figure 4 captures the data demonstrating a tight distribution of temperature measurement performance well within the +/-0.5C datasheet limits.

Figure 4: Calibrated DS1925 Temperature Accuracy

Conclusion

By using the known diode equations and applying precise external voltages during the calibration process, each of the unknown variables in the temperature equation can be solved to a) determine the DS1925 operating temperature at the calibration test step, and b) calibrate the diode Vbe   characteristics at this same test step to ideal. Once calibrated at this single test point, Vbe   provides a precise +/-0.5C accurate measurement of DS1925 operating temperature across the entire -40C to +85C range.

1. https://en.wikipedia.org/wiki/Silicon_bandgap_temperature_sensor
2. This adjustment of the Vbe   characteristic to ideal is a well-established electrical engineering principle within the domain of IC-based band-gap temperature sensor design.  Within the industry it is known as adjusting to the “Magic Voltage”.