Selecting a High-Performance DMM

DMM Terminology

An instrument data sheet gives a general description of performance, highlighting instrument resolution in number of digits, available functions and possibly maximum speed in readings per second. A more important figure of merit is the specification sheet that describes instrument capabilities for every function. It usually details the accuracy and resolution of each measurement function and includes how temperature, time, reading rate and other conditions will affect the performance of each type of measurement. In addition, the specification sheet often includes information on how the instrument will affect the device under test. Understanding the conditions for a particular application setup and environment helps determine if a particular instrument has the capabilities to make the required measurement.

Accuracy describes how close a reading is to the standard or true value and can be described as a level of uncertainty. For instance, when measuring volts, if an instrument displays 10.0000V, how close is that measurement to the actual volt standard that is kept by the NIST (National Institute of Standards and Technology)? The manufacturer that calibrates the instrument before delivery will calibrate volts to a traceable standard, which means that it is traceable back to an organization like NIST. Standards exist for volts, amps, ohms, and time or frequency.

Resolution is the value of the least significant digit of a reading, or how small of a change in the signal an instrument can detect. The number of digits displayed is one way to describe resolution. For instance, a 6-1/2 digit DMM on the 1V range will read 1.000000, where the resolution is the last digit, or 1 mV (microvolt).

Precision is the variation in multiple readings of a particular measurement. If the inputs to a voltmeter are shorted out, it should read 0V. Precision is the measure of variation in the measurement from reading to reading. Noise is the random source of error in an instrument that causes variations in readings.

The speed of measurement, or reading rate, is the time it takes to make an individual measurement and report a result. This includes the time it takes to measure the device under test and any additional internal measurement it takes per reading. Alternatively, it also means how many readings per second an instrument can perform and still meet the published specifications. Speeds will differ if readings are stored directly to system memory or over a network connection to a controlling PC.

Reading rate is also affected by how the instrument makes a measurement — either single ended or differential. A single-ended measurement involves making two measurements: first the high input, then the low, each at the number of cycles to achieve accuracy, then subtracting the two. An instrument that makes differential measurements needs only one measurement to create a reading.

Accuracy is typically given as a ±percent or ppm (parts per million) of a reading and ±percent or ppm of range where 1 ppm is equivalent to 0.0001 percent. The percent reading term is an accuracy gain error and the percent range term is an accuracy-offset term. These specifications are applicable for a given temperature window, usually 23°C ±5°C. Outside of this temperature window, an accuracy derating factor is applied to the original spec, usually specified as ±ppm (or percent) reading and ±ppm (or percent) range per degree C.

Another factor affecting accuracy specifications is the reading time or the aperture. Best readings are taken when reading time is an integer multiple of the power line cycle (PLC), 50 or 60 Hz, and an average number of readings are taken.

Accuracy specifications can be given for specific time periods. There is usually one spec for the recommended calibration cycle that is one year most of the time. There may also be a shorter time period, such as 90 days from last calibration. In some cases, an even shorter period, 24 hours or less, is based on an internal self calibration function by the instrument itself. In general, the shorter the time is between calibrations, the tighter the spec will be. Most manufacturers offer a calibration service. It is important that the calibration is traceable to one of the standards for each measurement function. Some units can be easily calibrated by end users.

In addition to accuracy, consider precision and resolution specs. Resolution is determined by the number of digits available from a measurement and the instrument range. For example, with a 7½-digit instrument on the 1A range, the resolution would be 100 nA. Precision is a specification of how much noise will be part of the measurement. Noise affects the ability to use the smaller value digits in the reading. Too much noise causes readings to fluctuate with a constant input which shows up on a display as flicker in the smaller value digits of the reading.

In the specifications, the precision or noise would be given as rms noise. Assuming the instrument noise is Gaussian, which is usually the case, then the peak-to-peak noise would be 6.6 times the rms noise. This gives an indication of how much the last digit or digits will vary from measurement to measurement. Precision is a function of reading speed, where 1 PLC typically gives the best reading (lowest noise) and the noise goes up with decreased measurement time. Filtering or averaging the readings will improve the noise, thereby improving the precision of a reading.

Other noise specifications such as CMRR and NMRR can also make an impact on the measurement. CMRR, or common mode rejection ratio, is the ability of the instrument to reject a signal common to both input test leads with respect to ground. NMRR, or normal mode rejection ratio, is the ability of the measurement to reject interference across its input terminals. Both of these specifications are given in decibels at a given frequency.

Some features that can distinguish a DMM’s capability might not be found on the spec page. For example, how does the instrument handle an over-range input? Does it give a maximum reading, or does it report an over-range? How does the instrument handle a broken or disconnected lead? Does it give a reading or report an error condition?

Types of DMMs

When considering high-accuracy measurement instruments, there are choices of multi-purpose DMMs or single function instruments. A distinguishing feature for voltage and resistance measurements is the value of the input impedance. For current measurements, important considerations are the voltage burden the instrument adds to the circuit under test and whether the instrument provides a voltage source for added versatility in some measurements.

DMMs provide a broad range of measurement types at a trade off of the measurement circuit’s affect on the DUT. Specialized instruments are optimized for a particular function and provide more ideal measurement circuits to ensure the highest accuracy. Also, specialized instruments typically have the lowest noise readings because the design was optimized for a particular type of measurement. They also provide specialized input connectors and low-noise cabling optimized for the specific measurement.

Applications

Applications for high accuracy DMMs are numerous. For instance, high-accuracy DMMs can be used as internal calibration standards for calibrating lab equipment and for verifying the accuracy of internally used instruments. They can also be used to spot-check instruments during calibration cycles.

One application requiring precision voltage circuits is measuring temperature. Temperature sensing devices convert temperature to voltage, typically with a very small temperature-to-voltage coefficient. High-accuracy DMMs can make these precision voltage measurements and detect the small voltage changes of sensors responding to temperature changes. In some cases, the transfer function is very non-linear, and the instrument will use a look-up table to convert the measured voltage to a temperature.

Another application involves heated Zener diodes used as calibration standards in calibration labs. Heated Zeners are rated at 10V and periodically need to be checked for accuracy. Heated Zeners are calibrated to within 1 ppm of uncertainty, or less than 10 uV. The Zener also needs to be tested to study noise and drift characteristics. To do these studies, the Zener is tested against a calibration standard called a Josephson-Junction array (JJ). In a cryogenic environment, the JJ array provides an output voltage in precise, stable 175 uV steps.

Typically, using a JJ for calibration involves measuring the precise voltage of the JJ and the difference between the JJ and the Zener. Noise on the last digits of the reading can cause uncertainty in transfer. To do this calibration, two high-accuracy DMMs are required. The first DMM needs to measure the JJ on the 10V range, and measure to less than 10 uV. This reading will be 10.000000V or 7½ digits. The second DMM will measure the difference between the JJ and the Zener. A nanovoltmeter would be a good choice because it allows for precise calibration as well as study of the drift and noise in the Zener.

In the test, the DMM connects across the JJ. The accuracy of the JJ is known to 175 uV steps. It is set up to output near 10V. The DMM measures the JJ to 7½ digits at a fixed temperature and determines which step the JJ is on. This determines the precise voltage. The Zener is the unknown and needs to be calibrated to within 1 ppm or 10 uV. The nanovoltmeter connects between the positive terminal of the JJ and the positive terminal of the Zener, and it measures the difference between the JJ and the heated Zener. On the 100 mV range, the nanovoltmeter can easily measure the Zener to within 10 uV, the last digit representing 10 nV. By knowing what step the JJ is on and the difference between the JJ and the Zener, the Zener’s value can be measured and the device can be characterized.

Another example involves thermistors. These devices are used as temperature sensors and have a well defined temperature coefficient. Characterizing or testing thermistors in manufacturing involves measuring resistance at specific temperatures. When measuring resistance with a DMM, a test current is applied to the DUT and the voltage is measured. Resistance is calculated using Vdut/Itest. During each resistance measurement, it is important to keep the test current as small as possible to limit the DUT power and the resulting self heating to get an accurate measurement.

Using the largest ohms range while still getting the desired resolution of measurement limits the self-heating of the thermistors during test. For thermistors in the 10 KO range, manufacturers try to measure the value of the resistance down to single ohm precision. With a 7½-digit DMM, a 10 KO thermistor can be tested on the 1 MO range and still get 1O precision on the last digit. The test current will be less than 1 µA, and the resulting power in the DUT less than 10 nW. For required test throughput, it is important to understand the required time for the DMM to make this precision measurement. Shorter measurement time in general increases noise. The quality of the DMM as well as the test setup determines how well and how fast this measurement can be made.

Making measurements on semiconductor devices with small geometries requires equipment that can measure small values and have minimum impact on the DUT. Voltage, current, resistance and capacitance are all device parameters that continue to decrease as the miniaturization process continues. The sensitivity of today’s devices requires a near ideal measurement environment to attain precise, accurate results. High-accuracy DMMs are the right instrument for these types of measurements. General purpose 7½- and 8½-digit DMMs are suitable when many different types of measurements are required. For measurements of a specific quantity there are nanovoltmeters, picoammeters and electrometers specifically designed for one type of measurement where the circuitry has the minimum impact on the DUT. Nanotechnology applications are similar to semiconductor applications, and they require precise measurement on very small devices and structures. This calls for instruments that have minimal affect on the DUT as well as high ohms ranges and very low voltage and current ranges.

Conclusion

With so many types of DMMs available on the market, from low-cost, general-purpose DMMs to very specialized function measurement instruments, understanding the particular measurement requirements helps in selecting the right piece of equipment. It is not enough to look for an instrument with the most available display digits. Understanding the requirements for measurement accuracy, precision and speed, and the expected test setup are required in order to properly select a test instrument. With these requirements in hand, careful review of a candidate instrument’s data sheet specifications will lead to the correct choice of an instrument for the particular application. In addition to the measurement capabilities, other instrument features, such as open lead detect, over-range detect and differential measurements, can make the decision between competing instrument choices of equal measurement capability.

Bibliography: Low Level Measurements Handbook, Keithley Instruments.

Kevin Cawley is a Senior Staff Engineer at Keithley Instruments in Cleveland, OH. He can be reached at (440) 498-2807, or via email at kcawley@keithley.com.