Non-Sampling Current Measurement for Microampere Ranges

Abstract

The purpose of this thesis was to study an alternative method for measuring the average current consumption of a low power embedded system. Traditionally, a shunt resistor is used to periodically sample the momentary current draw, and these samples are then averaged to find the average consumption. The sampling method works well when consumption is constant, however, if the consumption changes quickly, a high sampling frequency is needed to capture all the changes accurately. In this alternative method, current is measured by charging up a capacitor, powering the Device Under Testing from the capacitor and measuring how long it takes to discharge the capacitor by a certain voltage. This eliminates the problem of sampling frequency. According to the formula of capacitance, a capacitor relates a voltage to an amount of charge stored in the capacitor. If a capacitor is discharged by a known voltage, the number of chargers that were released can be calculated, given that the capacitance is known. Then, by measuring the time it takes to release the charges, a current value can be calculated by dividing the number of charges by the measured time, which results in the average current consumption during the measured time. In theory, the consumption during the measurement period should be captured perfectly. A measurement device using this style of measurement was built. The device was designed to measure an average current of 50 μA for 40 seconds to obtain a good average from a varying consumption profile. The accuracy and precision of the device was then measured by connecting resistors as the load and calculating what the ideal consumption reading for the given resistance would be. Then the current consumption of the resistor was measured and compared to the ideal number to find the accuracy. The precision was measured by repeating the measurement with multiple devices and calculating the standard deviation. The accuracy of the device remained within 3 % when measuring in a range of 10 μA to 500 μA. The standard deviation was generally within ± 1 %. The device employed electrolytic capacitors for the main measurement capacitor, which added challenges to the timing of the measurements. Electrolytic capacitors exhibit a behaviour where they don’t fully charge immediately after being connected to power but take minutes or even hours to reach their maximum charge. This phenomenon is called dielectric absorption. The same applies for discharging, when discharged quickly, they don’t release all their charge and the capacitance appears smaller, which increases the current reading. Especially for smaller currents, around 1 μA, the measurement result was quite sensitive to how long the device was powered on before starting the measurement. The result changed around 30% between starting the measurement immediately after being powered, and having the capacitor powered for 10 hours before measuring. Although, when measuring very small currents, less capacitance is needed. If the device can be designed for the smaller currents, it may be possible to utilize ceramic or plastic capacitors instead. These capacitor types exhibit less dielectric absorption, which should help the accuracy of the measurements significantly.