Inclinometer
For an introduction to the liquid-level inclinometer and a description of the apparatus and early work on the instrument, refer to Inclinometer Performance. This documentation records day-to-day work on the instrument from February 2008 onwards. Section headings are provided for easier navigation, while results are maintained in strict chronological order. On February 21, 2008, the fundamental resolution of the A2065 circuit was measured by substituting a dummy sensor made of resistors for the liquid-level sensor. The resolution obtained from the dummy sensor was found to be no better than that from the liquid sensor. Various circuit modifications were attempted, including increasing R7 and R8 by a factor of ten to improve decoupling on VCOM, which had no effect. Increases in the number of samples and periods of the sinusoid recorded by the Inclinometer Instrument were shown to have a significant impact on resolution, which improved by more than a factor of two when the number of periods was increased by a factor of four. A resolution of 60 V/V corresponds to 0.15 mrad (0.008°). The default settings for the Inclinometer Instrument were changed to analysis_harmonic = 16, daq_delay_ticks = 12, daq_num_samples = 6836.
On the same day, data acquisition began from two A2065Bs, one with the dummy sensor and the other with a liquid sensor, both mounted on a granite beam. A thermometer was added for hourly measurements. The superior 6836-sample, 16-period data acquisition settings were utilized for both A2065Bs. Results were documented in a spreadsheet. Observations indicated clear correlations with temperature, revealing that dummy sensor measurements were less stable than those from the liquid-level sensor, suggesting a problem with the dummy sensor's Inclinometer Head (A2065).
By March 12, 2008, a new dummy sensor was constructed using 0.1% wire-wound 1-kΩ resistors for X and 5% 10-kΩ resistors for Y. A 16-bit ADC was employed with an LWDAQ Driver (A2037E) for precise sample intervals. Analysis_harmonic was set to 11, daq_delay_ticks to 132, and daq_num_samples to 557. The sensor was subjected to heating and cooling, revealing changes of up to 1 mrad/°C, although consistent inclination graphs with temperature were not achieved. Overnight tests showed the effect of a slow 0.5°C drop, with the liquid sensor standard deviation recorded at 60 rad in both directions, while the dummy sensor exhibited poor performance.
Further investigations highlighted that the center-electrode signal was corrupted by cross-talk, evident from the amplitude changes when cables were added or touched. Cyclic fluctuations were observed with a period of 10 minutes and peak-to-peak amplitude of 1.1 mrad. Various modifications to the circuit and data acquisition settings were made to investigate their impact on the amplitude of the sinusoidal signals, but no significant effects were noted.
On March 14, 2008, the standard deviation of the liquid sensor reading decreased to around 50 rad after several hours, while cyclic errors persisted in the test circuit. Decoupling efforts and alterations to resistors were conducted, resulting in notable reductions in the amplitude of cross-talk signals. The understanding of performance improvements with a sensor attached to the A2065 was clarified, revealing the impact of impedance on cross-talk.
In subsequent tests, two fully assembled Inclinometers were placed on a granite table to check for cyclic errors. Observations indicated a decrease in waveform amplitudes over time, and stable measurements were recorded for Sensor 2, which had been previously connected to the data acquisition system. The effect of temperature was noted, with a significant correlation between temperature changes and inclination jumps.
By April 11, 2008, the derivative of temperature with time was added to plots, correlating inclination changes with temperature variations rather than absolute values. A Hall probe (A1301 from Allegro) was integrated into the system, providing measurements of the Earth's magnetic field. The sensitivity of the Hall probe was confirmed through systematic rotations, and voltage outputs were calibrated against known magnetic field strengths.
The comprehensive data collected through these experiments contributes significantly to the understanding of the liquid-level inclinometer's performance, sensor stability, and the effects of temperature and cross-talk on measurements. This ongoing research aims to refine the instrument's accuracy and reliability in various applications.For an introduction to our liquid-level inclinometer, and a description of our apparatus and early work on the instrument, see Inclinometer Performance. Here we record our day-to-day work on the instrument from February 2008 onwards. We provide some section headings just to make it easier for us to navigate the page. But the results remain in str ict chronological order. [21-FEB-08] We resolve to measure the fundamental resolution of the A2065 circuit by substituting a dummy sensor for the liquid level sensor. We make the dummy sensor with resistors, as shown below. We find that the resolution we obtained from the dummy sensor was no better than that we obtained with the liquid sensor.
We tried a few changes to the circuit. We increased R7 and R8 by a factor of ten to improve decoupling on VCOM (see schematic ). This had no effect. We increased the number of samples and periods of the sinusoid we recorded with the Inclinometer Instrument. The increases had a strong effect. We took ten samples under each of a variety of conditions and determined the resolution of the instrument X and Y in each case.
Our data is in the Dummy Sensor sheet of Results. xls, here. We see that our resolution with both the dummy and liquid sensors improves by more than a factor of two when we increase the number of periods by a factor of four. A resolution of 60 V/V corresponds to 0. 15 mrad (0. 008 °). But note that the resolution with our dummy sensor is no better than that with the liquid sensor. We change the default Inclinometer Instrument settings to analysis_harmonic = 16, daq_delay_ticks = 12, daq_num_samples = 6836.
[21-FEB-08] We start acquiring from two A2065Bs. One has the dummy sensor attached, the other a liquid sensor. Both are on our granite beam. We add a thermometer as well, and record measurements ever hour. For the A2065Bs we use the superior 6836-sample, 16-period data acquisition settings. Our Acquisifier script is here. Our results are in the spreadsheet called Dummy Sensor, here. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature, Using 8-Bit ADC. The Xdm and Ydm graphs give X and Y recorded from the dummy sensor. The Xlq and Ylq graphs give X and Y recorded from the liquid-level sensor. [25-FEB-08] We see clear correlations with temperature. The dummy sensor measurements are less stable than the liquid level measurements. We suspect a problem with the dummy sensor`s Inclinometer Head (A2065). [12-MAR-08] We make a new dummy sensor out of 0. 1% wire-wound 1-k © resistors for X and 5% 10-k © resistors for Y. We switch over to the 16-bit ADC, using a LWDAQ Driver (A2037E) with firmware version 13, in which we can obtain precise sample intervals with the driver`s sixteen-bit converter. We use analysis_harmonic = 11, daq_delay_ticks = 132, daq_num_samples = 557. We heat and cool the senor with a heat gun and freezer spray, and see changes of up to 1 mrad/ °C, but we cannot obtain consistent graphs of apparant inclination with temperature.
We resolve to let the dummy and liquid sensors run over-night, to see the effect of the slow 0. 5 °C over-night drop. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature, Using 16-Bit ADC. The Xdm and Ydm graphs give X and Y recorded from the dummy sensor. The Xlq and Ylq graphs give X and Y recorded from the liquid-level sensor. [13-MAR-08] The standard deviation of the liquid sensor measurement over-night, and with a 0. 5- °C drop, is 60 rad in both directions. The dummy sensor performs poorly. Its measurements drift together, and we suspect they are correlated with temperature in the same way they were in our longer-term test with the 8-bit ADC. We resolve to try a new Inclinometer Head (A2065). Our center-electrode signal is being corrupted by cross-talk. With no sensor and no flex cable, the center-electrode signal has amplitude 41 mV. This amplitude increases when we add a cable, and increases further when we touch the cable. This cross-talk affects X and Y equally. If the cross-talk amplitude changes, X and Y will together, just as they did in the graph above. We set up our new circuit with no sensor and no cable, and watch it for an hour. We obtain the following extraordinary plot. Figure: Cyclic Fluctuations with No Sensor, Using 16-Bit ADC. Graphs Xdm and Ydm for no sensor, Xlq and Ylq for liquid sensor. We acquire Xdm and Ydm first. We reverse the order of the data acquisition, so that we read the liquid sensor first and then the test circuit: no effect.
We obtain the same graph from the test circuit. The shared cyclic fluctuation in X and Y has period 10 minutes and peak-to-peak amplitude 1. 1 mrad. We make a more detailed recording, as shown below. Figure: Amplitude Fluctuations Recorded from Test Circuit with No Sensor. Graphs X+ and Y+ are the input amplitudes for the X and Y directions, and XC and YC are the center electrode amplitudes. The cyclic variation in center-electrode amplitude is obvious. Its period is 10 minutes and its peak-to-peak amplitude is 4 mV. When we divide 4 mV by the X+ amplitude of 8 V, we obtain a 500 V/V change in X, which corresponds to a 1.
2 mrad change in apparant inclination. We start changing things to see if they affect the amplitude of this sinusoid. We increase A2065_settling_delay from 0. 05 s to 0. 5 s: no effect. We notice that we are using a 15-ft CAT-5 patch cable for the test circuit, but a 15-ft LWDAQ branch cable for the liquid sensor circuit. So we switched the patch cable for a branch cable: no effect. Go back to previous test circuit, on in a box with lid off, with no sensor plugged in: period of the cyclic variation in XC and YC increases to 30 minutes, but amplitude remains 4 mV.
We switch back to the test circuit with the 10-minute period and switch to daq_num_samples = 219, daq_delay_ticks = 434, and analysis_harmonic = 14: no effect. Try daq_num_samples = 439, daq_delay_ticks = 106, and analysis_harmonic = 7: no effect. We modify the firmware so that the sinusoidal frequency runs independently from the circuit`s ring oscillator.
Up until now, we synchronised the 32. 768-kHz reference clock to the local ring oscillator and used the synchronised clock to generate the sinusoids. We re-programmed the test circuit: no effect. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature. Xdm and Ydm are from test circuit. Xlq and Ylq are from the liquid-level sensor. We shook the liquid sensor before we began taking data. The test circuit has no sensor or cable plugged in. [14-MAR-08] We obtain the graph above. The standard deviation of the liquid sensor reading drops to around 50 rad after the first few hours.
The cyclic error in the test circuit remains strong and steady. Decouple VCOM with a 10- F capacitor (see schematic ): no effect. Decouple ±15V power supplies: no effect. Replace R26 with a solder lump: XC amplitude drops from 41 mV to 0 mV and cycle drops from 4 mV to 0 mV. Replace R26 with 10 k ©: XC amplitude drops from 41 mV to 0. 5 mV and cycle drops from 4 mV to <0. 1 mV. On the oscilloscope, we see other asynchronous clock frequencies superimposed upon the fundamental 1.
17 kHz. We believe that it is the interaction of these clock frequencies with the fundamental and the clock on the driver that give rise to the cyclic changes in cross-talk amplitude. But we are not certain. With no sensor attached, the voltage on CTR is cross-talk induced upon the PCB traces, and loaded by R26.
When we decrease R26 from 1 M © to 10 k ©, we load the cross-talk voltage source and attenuate it by a factor of almost 100. The source impedance of the cross-talk is around 1 M ©. When we plug a sensor into the board, the sensor`s impedance loads the cross-talk. The sensors`s AC impedance is much lower than its DC impedance. With a voltmeter we measure 2 M © between adjacent pins. But with a sensor plugged in and R26 = 1 M ©, the XC amplitude is 2. 296 V. With R26 = 10 k ©, the XC amplitude is 2. 161 V. The impedance between CTR and VCOM presented by the sensor at 1 kHz is around 600 ©. We now understand why the A2065 performs better with a sensor plugged in than with no sensor. We put two fully-assembled Inclinometers on our granite table and check for cyclic errors. We obtain the graphs below from the newly-connected Inclinometer. We see the amplitude of the waveforms decreasing by about 0. 5% over the first hour. Figure: Start-Up Changes in Waveform Amplitudes. This is for an Inclinometer with tilt sensor, connected to the LWDAQ just before we began recording. We see no sign of cyclic changes in the waveform amplitudes. Figure: X and Y Inclination over Weekend for Two Sensors. Sensor 1 had been sitting sideways in a box until we placed it on the beam and connected it to our data acquisition system.
Sensor 2 had been sitting on the beam, connnected to the data acquisition system for weeks. But we disconnected it and carried it around briefly before the experiment. [18-MAR-08] We suppose that Sensor 1 is settling, while Sensor 2 was settled prior to the start of our experiment. The measurements of Sensor 2 are stable to 70 rad rms. We will leave the instruments running for another week. Looking at the X_2 graph, we see that the effect of temperature is of order +100 rad/ °C. The effect upon the remaining graphs is too small to observe. [11-APR-08] We add the derivative of temperature with time to the plot, and see how the jumps in inclination are associated with jumps in temperature, not the absolute value of the temperature.
The hall probe is an A1301 from Allegro. We connect the hall probe to an Input-Output Head ( A2057 ), which supplied 5-V for the sensor and reads the sensor output. With a 5-V supply, the sensor output is nominally 2. 5 V at zero field, and varies by 2. 5 mV/Gauss. The Earth`s magnetic field is roughly 0. 5 Gauss (50 Tesla). We fixed the hall probe to a rotation stage, oriented it straight up, and rotated it in 30 ° steps to confirm its sensitivity to the Earth`s magnetic field.
We know the approximate direction of north in our laboartory, and it corresponds to the maximum sensor output. The total change in output as we rotate the sensor is 2 mV. If the sensor sensitivity is 2. 5 mV/G, this gives the strength of the Earth`s magnetic field as 0. 4 G, which is about right. We will translate hall probe voltage into magnetic field by subtracting 2. 5413 and dividing the result by 25 V/T (T is for Tesla, 1 Tesla = 10, 000 Gauss). We attach a power supply to the electromagnet. We put the positive voltage on the lead that emerges at an inner corner of the core. We place the hall probe at the center of the air gap with the front face to the left. We increase the coil current form zero, and obtain the following plot. 🔗 External reference
On the same day, data acquisition began from two A2065Bs, one with the dummy sensor and the other with a liquid sensor, both mounted on a granite beam. A thermometer was added for hourly measurements. The superior 6836-sample, 16-period data acquisition settings were utilized for both A2065Bs. Results were documented in a spreadsheet. Observations indicated clear correlations with temperature, revealing that dummy sensor measurements were less stable than those from the liquid-level sensor, suggesting a problem with the dummy sensor's Inclinometer Head (A2065).
By March 12, 2008, a new dummy sensor was constructed using 0.1% wire-wound 1-kΩ resistors for X and 5% 10-kΩ resistors for Y. A 16-bit ADC was employed with an LWDAQ Driver (A2037E) for precise sample intervals. Analysis_harmonic was set to 11, daq_delay_ticks to 132, and daq_num_samples to 557. The sensor was subjected to heating and cooling, revealing changes of up to 1 mrad/°C, although consistent inclination graphs with temperature were not achieved. Overnight tests showed the effect of a slow 0.5°C drop, with the liquid sensor standard deviation recorded at 60 rad in both directions, while the dummy sensor exhibited poor performance.
Further investigations highlighted that the center-electrode signal was corrupted by cross-talk, evident from the amplitude changes when cables were added or touched. Cyclic fluctuations were observed with a period of 10 minutes and peak-to-peak amplitude of 1.1 mrad. Various modifications to the circuit and data acquisition settings were made to investigate their impact on the amplitude of the sinusoidal signals, but no significant effects were noted.
On March 14, 2008, the standard deviation of the liquid sensor reading decreased to around 50 rad after several hours, while cyclic errors persisted in the test circuit. Decoupling efforts and alterations to resistors were conducted, resulting in notable reductions in the amplitude of cross-talk signals. The understanding of performance improvements with a sensor attached to the A2065 was clarified, revealing the impact of impedance on cross-talk.
In subsequent tests, two fully assembled Inclinometers were placed on a granite table to check for cyclic errors. Observations indicated a decrease in waveform amplitudes over time, and stable measurements were recorded for Sensor 2, which had been previously connected to the data acquisition system. The effect of temperature was noted, with a significant correlation between temperature changes and inclination jumps.
By April 11, 2008, the derivative of temperature with time was added to plots, correlating inclination changes with temperature variations rather than absolute values. A Hall probe (A1301 from Allegro) was integrated into the system, providing measurements of the Earth's magnetic field. The sensitivity of the Hall probe was confirmed through systematic rotations, and voltage outputs were calibrated against known magnetic field strengths.
The comprehensive data collected through these experiments contributes significantly to the understanding of the liquid-level inclinometer's performance, sensor stability, and the effects of temperature and cross-talk on measurements. This ongoing research aims to refine the instrument's accuracy and reliability in various applications.For an introduction to our liquid-level inclinometer, and a description of our apparatus and early work on the instrument, see Inclinometer Performance. Here we record our day-to-day work on the instrument from February 2008 onwards. We provide some section headings just to make it easier for us to navigate the page. But the results remain in str ict chronological order. [21-FEB-08] We resolve to measure the fundamental resolution of the A2065 circuit by substituting a dummy sensor for the liquid level sensor. We make the dummy sensor with resistors, as shown below. We find that the resolution we obtained from the dummy sensor was no better than that we obtained with the liquid sensor.
We tried a few changes to the circuit. We increased R7 and R8 by a factor of ten to improve decoupling on VCOM (see schematic ). This had no effect. We increased the number of samples and periods of the sinusoid we recorded with the Inclinometer Instrument. The increases had a strong effect. We took ten samples under each of a variety of conditions and determined the resolution of the instrument X and Y in each case.
Our data is in the Dummy Sensor sheet of Results. xls, here. We see that our resolution with both the dummy and liquid sensors improves by more than a factor of two when we increase the number of periods by a factor of four. A resolution of 60 V/V corresponds to 0. 15 mrad (0. 008 °). But note that the resolution with our dummy sensor is no better than that with the liquid sensor. We change the default Inclinometer Instrument settings to analysis_harmonic = 16, daq_delay_ticks = 12, daq_num_samples = 6836.
[21-FEB-08] We start acquiring from two A2065Bs. One has the dummy sensor attached, the other a liquid sensor. Both are on our granite beam. We add a thermometer as well, and record measurements ever hour. For the A2065Bs we use the superior 6836-sample, 16-period data acquisition settings. Our Acquisifier script is here. Our results are in the spreadsheet called Dummy Sensor, here. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature, Using 8-Bit ADC. The Xdm and Ydm graphs give X and Y recorded from the dummy sensor. The Xlq and Ylq graphs give X and Y recorded from the liquid-level sensor. [25-FEB-08] We see clear correlations with temperature. The dummy sensor measurements are less stable than the liquid level measurements. We suspect a problem with the dummy sensor`s Inclinometer Head (A2065). [12-MAR-08] We make a new dummy sensor out of 0. 1% wire-wound 1-k © resistors for X and 5% 10-k © resistors for Y. We switch over to the 16-bit ADC, using a LWDAQ Driver (A2037E) with firmware version 13, in which we can obtain precise sample intervals with the driver`s sixteen-bit converter. We use analysis_harmonic = 11, daq_delay_ticks = 132, daq_num_samples = 557. We heat and cool the senor with a heat gun and freezer spray, and see changes of up to 1 mrad/ °C, but we cannot obtain consistent graphs of apparant inclination with temperature.
We resolve to let the dummy and liquid sensors run over-night, to see the effect of the slow 0. 5 °C over-night drop. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature, Using 16-Bit ADC. The Xdm and Ydm graphs give X and Y recorded from the dummy sensor. The Xlq and Ylq graphs give X and Y recorded from the liquid-level sensor. [13-MAR-08] The standard deviation of the liquid sensor measurement over-night, and with a 0. 5- °C drop, is 60 rad in both directions. The dummy sensor performs poorly. Its measurements drift together, and we suspect they are correlated with temperature in the same way they were in our longer-term test with the 8-bit ADC. We resolve to try a new Inclinometer Head (A2065). Our center-electrode signal is being corrupted by cross-talk. With no sensor and no flex cable, the center-electrode signal has amplitude 41 mV. This amplitude increases when we add a cable, and increases further when we touch the cable. This cross-talk affects X and Y equally. If the cross-talk amplitude changes, X and Y will together, just as they did in the graph above. We set up our new circuit with no sensor and no cable, and watch it for an hour. We obtain the following extraordinary plot. Figure: Cyclic Fluctuations with No Sensor, Using 16-Bit ADC. Graphs Xdm and Ydm for no sensor, Xlq and Ylq for liquid sensor. We acquire Xdm and Ydm first. We reverse the order of the data acquisition, so that we read the liquid sensor first and then the test circuit: no effect.
We obtain the same graph from the test circuit. The shared cyclic fluctuation in X and Y has period 10 minutes and peak-to-peak amplitude 1. 1 mrad. We make a more detailed recording, as shown below. Figure: Amplitude Fluctuations Recorded from Test Circuit with No Sensor. Graphs X+ and Y+ are the input amplitudes for the X and Y directions, and XC and YC are the center electrode amplitudes. The cyclic variation in center-electrode amplitude is obvious. Its period is 10 minutes and its peak-to-peak amplitude is 4 mV. When we divide 4 mV by the X+ amplitude of 8 V, we obtain a 500 V/V change in X, which corresponds to a 1.
2 mrad change in apparant inclination. We start changing things to see if they affect the amplitude of this sinusoid. We increase A2065_settling_delay from 0. 05 s to 0. 5 s: no effect. We notice that we are using a 15-ft CAT-5 patch cable for the test circuit, but a 15-ft LWDAQ branch cable for the liquid sensor circuit. So we switched the patch cable for a branch cable: no effect. Go back to previous test circuit, on in a box with lid off, with no sensor plugged in: period of the cyclic variation in XC and YC increases to 30 minutes, but amplitude remains 4 mV.
We switch back to the test circuit with the 10-minute period and switch to daq_num_samples = 219, daq_delay_ticks = 434, and analysis_harmonic = 14: no effect. Try daq_num_samples = 439, daq_delay_ticks = 106, and analysis_harmonic = 7: no effect. We modify the firmware so that the sinusoidal frequency runs independently from the circuit`s ring oscillator.
Up until now, we synchronised the 32. 768-kHz reference clock to the local ring oscillator and used the synchronised clock to generate the sinusoids. We re-programmed the test circuit: no effect. Figure: Stability of Dummy Sensor and Liquid Sensor with Temperature. Xdm and Ydm are from test circuit. Xlq and Ylq are from the liquid-level sensor. We shook the liquid sensor before we began taking data. The test circuit has no sensor or cable plugged in. [14-MAR-08] We obtain the graph above. The standard deviation of the liquid sensor reading drops to around 50 rad after the first few hours.
The cyclic error in the test circuit remains strong and steady. Decouple VCOM with a 10- F capacitor (see schematic ): no effect. Decouple ±15V power supplies: no effect. Replace R26 with a solder lump: XC amplitude drops from 41 mV to 0 mV and cycle drops from 4 mV to 0 mV. Replace R26 with 10 k ©: XC amplitude drops from 41 mV to 0. 5 mV and cycle drops from 4 mV to <0. 1 mV. On the oscilloscope, we see other asynchronous clock frequencies superimposed upon the fundamental 1.
17 kHz. We believe that it is the interaction of these clock frequencies with the fundamental and the clock on the driver that give rise to the cyclic changes in cross-talk amplitude. But we are not certain. With no sensor attached, the voltage on CTR is cross-talk induced upon the PCB traces, and loaded by R26.
When we decrease R26 from 1 M © to 10 k ©, we load the cross-talk voltage source and attenuate it by a factor of almost 100. The source impedance of the cross-talk is around 1 M ©. When we plug a sensor into the board, the sensor`s impedance loads the cross-talk. The sensors`s AC impedance is much lower than its DC impedance. With a voltmeter we measure 2 M © between adjacent pins. But with a sensor plugged in and R26 = 1 M ©, the XC amplitude is 2. 296 V. With R26 = 10 k ©, the XC amplitude is 2. 161 V. The impedance between CTR and VCOM presented by the sensor at 1 kHz is around 600 ©. We now understand why the A2065 performs better with a sensor plugged in than with no sensor. We put two fully-assembled Inclinometers on our granite table and check for cyclic errors. We obtain the graphs below from the newly-connected Inclinometer. We see the amplitude of the waveforms decreasing by about 0. 5% over the first hour. Figure: Start-Up Changes in Waveform Amplitudes. This is for an Inclinometer with tilt sensor, connected to the LWDAQ just before we began recording. We see no sign of cyclic changes in the waveform amplitudes. Figure: X and Y Inclination over Weekend for Two Sensors. Sensor 1 had been sitting sideways in a box until we placed it on the beam and connected it to our data acquisition system.
Sensor 2 had been sitting on the beam, connnected to the data acquisition system for weeks. But we disconnected it and carried it around briefly before the experiment. [18-MAR-08] We suppose that Sensor 1 is settling, while Sensor 2 was settled prior to the start of our experiment. The measurements of Sensor 2 are stable to 70 rad rms. We will leave the instruments running for another week. Looking at the X_2 graph, we see that the effect of temperature is of order +100 rad/ °C. The effect upon the remaining graphs is too small to observe. [11-APR-08] We add the derivative of temperature with time to the plot, and see how the jumps in inclination are associated with jumps in temperature, not the absolute value of the temperature.
The hall probe is an A1301 from Allegro. We connect the hall probe to an Input-Output Head ( A2057 ), which supplied 5-V for the sensor and reads the sensor output. With a 5-V supply, the sensor output is nominally 2. 5 V at zero field, and varies by 2. 5 mV/Gauss. The Earth`s magnetic field is roughly 0. 5 Gauss (50 Tesla). We fixed the hall probe to a rotation stage, oriented it straight up, and rotated it in 30 ° steps to confirm its sensitivity to the Earth`s magnetic field.
We know the approximate direction of north in our laboartory, and it corresponds to the maximum sensor output. The total change in output as we rotate the sensor is 2 mV. If the sensor sensitivity is 2. 5 mV/G, this gives the strength of the Earth`s magnetic field as 0. 4 G, which is about right. We will translate hall probe voltage into magnetic field by subtracting 2. 5413 and dividing the result by 25 V/T (T is for Tesla, 1 Tesla = 10, 000 Gauss). We attach a power supply to the electromagnet. We put the positive voltage on the lead that emerges at an inner corner of the core. We place the hall probe at the center of the air gap with the front face to the left. We increase the coil current form zero, and obtain the following plot. 🔗 External reference