Instrumentation amplifier extends DSO
To determine the specifications of a solar-generating plant, I needed to accurately measure the load current a product consumed. The product switched several internal devices on and off during an interval of several seconds. An ammeter showed that the current transitions occurred too quickly for visual logging, and my managers had requested an oscilloscope photo of the current waveform's peaks. I rolled out our company's cart-mounted DSO (digital-storage oscilloscope), inserted a low-value resistor in series with the product's positive-power-supply input, and attempted to make a differential-voltage measurement (Channel A minus Channel B) across the current-sampling resistor.
Unfortunately, RF noise from a local FM-broadcast station swamped the small-load-induced fluctuations in the voltage developed across the sampling resistor, and increasing its resistance introduced an unwanted voltage drop on the product's power-supply rail. Finally, the 12V supply rail introduced a voltage offset that limited the oscilloscope's ability to accurately resolve the small differential signal that I was attempting to measure. I disconnected the oscilloscope's ac ground to "float" the scope with respect to the sampling resistor, but the RF noise visible on the trace increased significantly. I briefly considered using an older analog (nonstorage) scope, but the DSO's storage feature would allow me to capture and print the waveforms required for my report.
In frustration, I scoured the workbench for stray parts and assembled a circuit that solved the problem. By chance, the parts collection included an instrumentation amplifier, IC1, which does an excellent job of extracting small signals from high-frequency background noise. The amplifier's inherently slow response attenuates RF noise but doesn't affect amplification of lower frequency signals. Adding RC lowpass filters to the amplifier's inputs and output further attenuates lower frequency noise induced by nearby switched-mode power supplies and digital logic or microprocessors.
Normally, I avoid using noise-emitting dc/dc converters as power supplies for analog circuits. However, in this case, IC2, a dc/dc converter, provided an expedient and technically sound approach (Figure 1). In general, dc/dc converters produce more noise as their load currents increase, but, in this circuit, the sole load comprises the instrumentation amplifier that draws only a few milliamperes. Adding a few filtering components provided additional noise suppression.
Under normal operation, the current that the product draws fluctuates from approximately 300 to 800 mA. To minimize the voltage drop induced in the power-supply loop, I used a 5×20-mm, 10A, 250V fuse, F1, as a current-sampling resistor. Voltage drop across the fuse is approximately 1 mV per 100 mA of current, and operating the fuse at a small fraction of its nominal rating avoids introducing nonlinearities in the measurement.
With a 475Ω gain-setting resistor, R2, the instrumentation amplifier, an Analog Devices AD620, provides a gain of 105V/V and delivers an output of approximately 1V, which corresponds to 1A of current flowing through the shunt. Capacitors C12 and C13 provide low-impedance paths for high-frequency noise.