High-End DMMs Simplify Measurements

Go beyond the obvious specs and features when you buy a precision DMM.

by Jon Titus, Senior Technical Editor 

Jon TitusWhen engineers start to evaluate a precision digital multimeter (DMM) they first think about resolution and accuracy. "They may start with requirements for a high and low measurement limit and a tolerance," said Chuck Cimino, a marketing director at Keithley. "Then they figure out how many digits they need without going into a lot of the DMM's detailed specs."

During the selection process engineers might forget about important specifications and characteristics such as measurement uncertainties, nonlinearities, calibration needs, and de-rating for non-lab environments. "Customers often overlook two accuracy or linearity specifications," said Hilton Hammond, product manager for Fluke Precision Measurements. "First comes percent of range, which defines a DMM's overall noise floor or the sensitivity of the ADC's input circuits. Second comes percentage of measurement, or percentage of reading. Try to make a 10-millivolt measurement with a one-volt or 10-volt range, for example, and you discover the unknown signal got buried under the DMM's noise floor and readings have becomes invalid."

10ECN-Titus CS DMMs Art BAccording to Keithley's Cimino, engineers also must consider a system error budget. "Say you plan to make a resistance measurement and you have a given measurement tolerance. A measurement system may include interface cables, probes, fixtures and handling equipment, all of which can add to measurement uncertainty. I recommend designers create a tolerance or error budget so they can quantify the uncertainty margin they have. Often they realize they don’t have as big a margin for the DMM as they thought they had. You cannot look at just the DMM and what you want to measure and forget about everything in between."

Keep in mind that manufacturers specify their DMMs under ideal conditions with a low-impedance calibrator and shielded cables. "Those 'clean' conditions don't exist most of the time," noted Cimino. "You will have electrical noise from power lines, motors, relays and other devices and those extraneous signals can decrease measurement accuracy. Suppose you need a certain test rate that a chosen DMM can meet. But on a production line you discover you must average 10 readings to get a stable measurement, which means production throughput could slow ten fold."

Travis White, the DMMs and Power Supplies Product Manager at National Instruments stressed that NI goes to great lengths to provide detailed specifications for DMM modules and cards for different operating conditions, aperture times and digital resolutions. That information helps engineers with their error or tolerance budget. "If a test system includes any sort of switch components, perhaps for many thermocouples or sensors, you must account for the decreased accuracy caused by small offset voltages from the switches and connectors. Choosing a good DMM is only about half the battle, you need to think about cables, shielding, contacts and other parts of the measurement apparatus to understand how to minimize how they affect DMM readings."

10ECN-Titus CS DMMs Art CDMM suppliers may offer accessories, such as switch matrices and low-EMF signal-switching relays designed specifically for certain types of high-accuracy measurements. A connection block for thermocouple, for example, could include a DMM-compatible temperature sensor for cold-junction compensation. "We specify the accuracy of our DMMs based on measurements made over an integer number of power line cycles, which reduces parasitic noise," explained Keithley's Cimino. "Many engineers don't have the luxury of long measurement periods and they forget about power-line noise. So they buy a 7½-digit DMM, measure over a tenth of a power-line cycle and wonder why they get noisy readings."

Bryan Boswell works at Agilent Technologies as a senior test engineer and part of his work involves using one of the company's 33410A DMMs to measures the voltage drop across a current shunt as it warms up. From that information he can calculate settling times and time constants for a shunt. For these tests, though, he uses the DMM as a data logger. The built-in data logger takes repetitive sample rate over a set time or for a fixed number of samples. Then he can read the samples into an Excel spreadsheet through a network connection with the DMM. "I ran tests and produced graphs in less time than it would have taken to find the programming manual for my old DMM," noted Boswell. Users invoke the data-logger functions through a 33410A front-panel control or by way of a Web interface on a PC.

DMMs also show up in unusual applications that take advantage of one or more special capabilities that Chris Kelly, a design engineer at Agilent, explained. "A customer needed to measure the junction temperature of one diode in a long string of LEDs. An LED poorly or improperly bonded to its substrate cannot get rid of heat and it can fail and ruin an LED illumination string during manufacturing tests or later in a product." The measurement sequence involved driving the LED string at its maximum current and then switching the current off for 200 ?sec. During that short window, an instrument must make a forward-bias measurement across an individual LED so engineers can determine its junction temperature.

Kelly recommended an Agilent 33411A DMM because it offers a 50 Ksamples/sec acquisition rate and it could withstand the common-mode signals present "far up" the LEDs from ground potential. The 33411A offers a 140 dB common-mode rejection ratio up to a maximum of 500V, which 10ECN-Titus CS DMMs Art Dsufficed for the 6½-digit accuracy the customer needed. "You might not think of a bench DMM as an instrument that can take 50,000 readings a second with a 140 dB common-mode rejection ratio [CMRR]," said Kelly Although older DMMs displayed only basic electrical measurements, newer instruments take advantage of multi-line displays and internal computer power to show users peak-to-peak voltage, RMS voltage, frequency, standard deviation and so on. "Often that second line of measurements provides additional information that points engineers to the source of a problem," said Agilent's Boswell. "The display gives you diagnostic information you might not get otherwise so a user might say, 'Wait a second, the standard deviation for this is way off what it’s supposed to be.'" Users can select the "extra" information on the display depending on the main DMM function they choose.

Fluke takes its LCD a step farther and uses a dot matrix to display information and plot trends in real time. "Our 8845A and 8846A DMMs present statistics information such as minimum and maximum values, averages, standard deviation, and so, on one display so users don't have to flip back and forth between functions," explained Hammond. "Suppose you need to characterize an analog circuit. In addition to statistical information, you can see a measurement trend. That's important when you want to see right away how temperature, for example, affects an electrical characteristic over a given time. You used to have to find an interface cable for a PC, load the needed driver software and write code to obtain that type of trend information. Now you can get it directly from the DMM display in real time."

Travis White of National Instruments explained that even for PC-based DMMs, developers no longer need to expend much time to load drivers, configure hardware and write data-analysis code. The PC and instrument-control software now handle those tasks. And PC-based DMMs provide a display as large as your PC's display. Thus, engineers can use a few mouse clicks with NI's Signal Express software--which comes with the company's DMMs--to set up and operate an NI DMM. "No one wants to waste a week trying to get drivers to work and then have to write low-level code to use a USB or PXI DMM product."

For further reading 

"Understanding specifications for precision multimeters," Fluke.

"Removing Thermal Offset Errors from Contact, Switch, and Relay Resistance Measurements Tutorial," National Instruments.

"Part I: How To Reduce Errors When Switching Low-Voltage Signals," National Instruments.