Kevin Bisking
National Instruments
Austin, Tex.

 Simplified autobalancing circuit 1-MHz constant current source and dc bias A simplified block diagram of the FlexDMM shows how software is used to measure the L and C of a DUT. The instrument stimulates a DUT using a square wave generated by a constant current source. An analog/digital converter reads the resulting voltage across the DUT. A digital signal processor performs a Fast Fourier Transform on the sample data to determine the magnitude of the first and third harmonics as measured across the DUT. From this information the DMM can calculate the L or C of the DUT. Simplified block diagram, FlexDMM A simplified block diagram of the FlexDMM shows how software is used to measure the L and C of a DUT. The instrument stimulates a DUT using a square wave generated by a constant current source. An analog/digital converter reads the resulting voltage across the DUT. A digital signal processor performs a Fast Fourier Transform on the sample data to determine the magnitude of the first and third harmonics as measured across the DUT. From this information the DMM can calculate the L or C of the DUT. Real and imaginary components of DUT impedance A vector diagram helps visualize how software in the FlexDMM measures the L or C of a DUT. In the case of a capacitor, the vector representation of its impedance appears as shown at the fundamental and third harmonic frequencies. The real component is the same at both frequencies. The fundamental and harmonic frequencies are of course known. So software can use the amplitude of the two frequencies measured across the DUT to deduce the imaginary part of the device's impedance. From this the instrument can calculate the DUT capacitance. An inductor DUT would be measured in an analogous way, but its impedance vector would be in the opposite quadrant.

Design engineers have long measured inductance and capacitance using the same basic idea: A sine-wave source sends a signal to a precision bridge. The bridge produces an output proportional to the L or C in one of its legs.

Measurements based on impedance bridge circuits lend themselves to the kind of testing that takes place in component characterization because they can provide programmable test frequencies and excitation voltages or currents. But many applications don't need the level of complexity and the costs that a measurement bridge incurs. The hardware for this approach can be expensive because it employs several stages of complex analog building blocks that are phase sensitive.

These were among the underlying motivations for development of an innovative, patent-pending technique for precision measurement of inductance and capacitance. Built around an easy-touse digital multimeter, it handles these measurements at a fraction of the cost and size of traditional stand-alone instrumentation.

The advantages of the new approach can perhaps be best understood in the context of how ordinary LCR meters measure capacitance and inductance. The usual technique for components that work up to about 40 MHz or so is to employ what is called an autobalancing bridge. The technique draws its name from its ability to automatically null out drift and other imbalances in the measurement circuit that would otherwise cause a nonzero reading on the meter.

An autobalance circuit is usually modeled as an operational amplifier with a feedback element, which gets a stimulating signal from a precision signal source. The stimulating signal is usually a sinusoid of about 1 MHz. The device under test (DUT) is inserted between the stimulating source and the inverting input of the operational amplifier.

There is a resistor R in the feedback loop of the op amp. This resistor value is known precisely. With the DUT so connected, its output terminal (i.e., the terminal connected to the op amp inverting input) is at a virtual ground, from the definition of how op amps operate. This relationship lets the DUT impedance be expressed as a function of the op amp output voltage, feedback resistor R, and the voltage across the DUT. The L or C of the DUT can then be calculated from its measured impedance.

The point to note is that the feedback resistor R must be a precision device. This makes it relatively expensive. There is also some elaborate circuitry to compensate for inherent voltage drop across the source resistance and DUT. This adds to the cost of the approach as well.

In practice, LCR meters often replace the op amp with a zero phase detector and a modulator. This scheme lets autobalancing bridges work to about 100 MHz.

MEASUREMENTS IN SOFTWARE
A measurement technique based on software takes a dramatically different approach than ordinary LCR meters. First, it stimulates the DUT with a current square wave instead of a voltage sine wave. The reason is that square waves can be thought of as a superposition of numerous harmonics. It is possible to measure the reactance of a DUT by gauging how it affects the harmonics of a square wave that stimulates it.

Specifically, a capacitor or inductor stimulated with a square wave will redistribute the energy of the harmonics impinging on it. This redistribution can be interpreted to yield DUT capacitance or inductance.

This approach first digitizes the voltage waveform read across the DUT. A digitization rate of 1.8 Msamples/sec provides enough points to calculate a 256-bin Fast Fourier Transform (FFT). The FFT operation takes place in a digital signal processor (DSP). An FFT, of course, describes a time-domain waveform in terms of the amplitude and phase of the frequencies that comprise it.

In a nutshell, FFT software compares the amplitude at the first and third harmonics of the current square wave generator to those passing through the DUT. The ratio of the amplitudes between the input and output signals is proportional to the L or C of the DUT.

The implementation of this technique is in the FlexDMM from National Instruments. The size of the sample window determines the cost and accuracy of the measurement system, so the bin size of the FFT (i.e., the FFT resolution) is a trade-off.

Using this approach the FlexDMM can detect C and L values down to 0.01 pF and 0.01µH with ranges extending to 10,000 F and 5 H. This level of sensitivity and basic accuracy to within 0.25% makes the technique suitable for applications that include production test, component screening, cable capacitance testing, wafer level go/no-go, and automotive sensors.

The software-based technique lets design engineers quickly make precision capacitance and inductance measurements with the same instrument used for traditional DMM functions and waveform measurements. It has the advantage of costing a fraction of what's normal for traditional standalone instrumentation.

Finally, the use of software in such measurements heralds a trend. Future digital multimeters will employ even more novel software techniques on the road toward becoming universal measurement systems. Eventually engineers will make a wide variety of common test measurements with a single, low-cost device.

MAKE CONTACT:

National Instruments Corp., www.NI.com