ExternalTemperatureSensorCalibrationfortheMAX16031/MAX16032SystemMonitor...
加入技术交流群
External Temperature Sensor Calibration for the MAX16031/MAX16032 System Monitors
Eric Schlaepfer
Abstract: The MAX16031/MAX16032 system monitors support external diode-connected transistors to sense remote temperatures. Diode-connected transistors exhibit temperature-dependent characteristics, which can be measured using a simple process. These characteristics are slightly different depending src="/data/attachment/portal/201007/ET40352201007260325231.gif">
where T is the temperature measured in Kelvins, n is the diode ideality factor, k is Boltzmann's constant (1.38e-23), q is the charge of an electron (1.6e-19), VBE is the base-emitter voltage, IS is the reverse saturation current, and I is the diode current.
A simple, yet accurate way to sense the temperature is to force two different currents through a diode and take the voltage difference between the base emitter voltages for each current. This cancels out the dependency src="/data/attachment/portal/201007/ET40352201007260325232.gif">
where ΔVBE is the difference in base-emitter voltages at both currents, IH is the higher forcing current, and IL is the lower forcing current. The other parameters remain the same from before. Solving for T and converting to degrees Celsius yields the following:
This equation shows that the temperature is directly proportional to ΔVBE with a ratio determined by physical constants, the diode ideality factor, and the ratio of the two sensing currents.
MAX16031/MAX16032 Temperature SensingThe MAX16031/MAX16032 measure the temperature of the external temperature diodes using the two-current technique discussed in the last section. After scaling and converting ΔVBE to a digital code, the device subtracts a fixed code representing the -273.15 degree offset. The resulting number is the temperature in degrees Celsius with a resolution of 0.5°.
The temperature-sensing circuitry requires calibration to compensate for different diode ideality factors. There are two calibration values required for each external temperature sensor in the MAX16031/MAX16032: a gain and an offset.
A gain value, stored in r19h[7:2] (sensor 1) and r4Fh[5:0] (sensor 2) controls the IH current (the IL current is fixed at 6µA). Table 1 shows the bit weights for each gain register bit. For example, the value 110000b corresponds with an IH of 84µA.
Table 1. Gain Register Bit Weights r19h[7:2], r4Fh[5:0] BitIH Value Added to 80µA (µA)Logic '1'Logic '0'0+0.250.01+0.250.02+0.50.03+1.00.040.0+2.05+4.00.0
The offset value, located in r1Bh[7:5] (sensor 1) and r4Dh[6:4] (sensor 2), gets added to temperature conversion result to compensate for offset errors. Table 2 shows the possible offset values.
Table 2. Offset Register Values r1Bh[7:5], r4Dh[6:4] ValueOffset (°C)100+8101+6110+4111+20000001-2010-4011-6
When using the MAX16031 evaluation kit (EV kit), the software provides convenient access to each calibration register through the Temperature Settings dialog shown in Figure 1. From the Current and Temperature tab (seen in Figure 2), click the blue External Temperature 1 or External Temperature 2 link to access the settings dialog for each temperature-monitoring channel.
Figure 1. Temperature Settings Dialog.
Figure 2. Current and Temperature Tab.
During the calibration procedure, as well as during normal operation, it is very helpful to enable the built-in digital temperature sensor filter. In the EV kit software, choose a center frequency from the Temp sense filter time constant dropdown list located in the Miscellaneous tab. This corresponds with register r5Bh[6:4], which is described in Table 3.
Table 3. Temperature Sensor Digital Filter r5Bh[6:4] ValueCutoff Frequency (Hz)000Filter Disabled0012.530105.0601110.110020.210140.511081111162
Calibration ProcedureTo calibrate the temperature-sensor circuit of the MAX16031/MAX16032, data needs to be taken at two different temperatures. The temperatures should be widely separated for best results. For simplicity, the first temperature may be +25°C. For better accuracy, the second temperature should be higher than room temperature rather than lower; a good value is +85°C. For the best accuracy, data should be taken at -40°C and +85°C. The following MAX16031 example takes calibration data at +25°C and +85°C, and uses a Fairchild 2N3904 transistor in the diode-connected configuration—the collector and base shorted together.
For best results, the sensor being calibrated should be immersed in a nonconductive fluid in a temperature-controlled bath. The example uses Fluorinert FC-77, although other fluids such as mineral oil may also work. If such equipment is not available, a commercially-available toaster oven may suffice if the sensor is attached to a metal object with a large thermal mass—this increases temperature stability during calibration. Use an accurate thermocouple with good thermal coupling to the sensor in order to measure the temperature.
Use the following procedure to gather the calibration data:
Several simple calculations should be performed once the data is collected. For each temperature collected from the MAX16031, calculate the error value:
TERR = TMAX16031 - TMEASUREDEq. 4
Then, calculate the difference between each error value for a single gain value:
Δ = TERR_85 - TERR_25Eq. 5
Table 4 shows the example calibration data. The columns Gain Reg Code and Gain Value (µA) show the gain register setting as both a hex code and as the equivalent current source value. The TMAX16031 (°C) column shows the readings obtained from the MAX16031 temperature-conversion-result registers for each gain setting at both the low temperature (+25°C) and the high temperature (+85°C). The TERR (°C) column lists the error (difference) between the temperatures recorded by the MAX16031 and the actual, measured temperatures. The Δ (°C) column lists the differences between the error values recorded at the high temperature and the low temperature.
Table 4. Example Calibration Data Gain Reg CodeGain Value (µA)TMAX16031 (°C)TERR (°C)Δ (°C)+25+85+25+850x108017.575-7.5-10-2.50x1280.2517.575.5-7.5-9.5-20x1480.517.576-7.5-9-1.50x1680.751876.5-7-8.5-1.50x188118.576.5-6.5-8.5-20x1A81.2518.577-6.5-8-1.50x1C81.51977.5-6-7.5-1.50x1E81.7519.578-5.5-7-1.50x0822078.5-5-6.5-1.50x282.252079-5-6-10x482.520.579-4.5-6-1.50x682.752179.5-4-5.5-1.50x8832180-4-5-10xA83.2521.580.5-3.5-4.5-10xC83.52281-3-4-10xE83.7522.581.5-2.5-3.5-10x308422.582-2.5-3-0.50x3284.252382-2-3-10x3484.52382.5-2-2.5-0.50x3684.7523.583-1.5-2-0.50x38852483.5-1-1.5-0.50x3A85.2524.583.5-0.5-1.5-10x3C85.524.584-0.5-1-0.50x3E85.752584.50-0.5-0.50x208625.5850.50-0.50x2286.2525.5850.50-0.50x2486.52685.510.5-0.50x2686.7526.5861.51-0.50x288726.586.51.51.500x2A87.2527872200x2C87.527.5872.52-0.50x2E87.752887.532.5-0.5
The next task is to find the gain and offset calibration parameters. Examine the Δ (°C) column and find the cells with the values closest to zero. In this case, two rows for gain settings of 87 and 87.25 contain zero. This corresponds to a slope of zero, meaning there is negligible gain error. Next, examine the TERR (°C) values for these rows and choose the row containing error values that can be subtracted out using the offset register. In the example, the row with an offset error of 2 is chosen so that the offset register value of '001' can cancel it out.
The values obtained for the offset and gain registers may now be loaded into the EEPROM configuration registers of the MAX16031 monitor used in the application circuit. The same values can be used every time, resulting in reasonable accuracy.
To increase the accuracy, the gain parameter can be programmed in circuit for each unit at board test time. Place an accurate thermocouple near the device and measure the temperature. Adjust the gain register contents until the MAX16031 temperature matches the measured temperature.
Fluorinert is a trademark of the 3M Company.
SMBus is a trademark of Intel Corp.
-- END: DB HTML -->
评论