True-RMS Handheld DMMs Get a Hands-On Test

The root-mean-square value, VRMS, is what we really want, because it is the only value that is independent of waveform. VRMS also represents a signal’s power capability and this is almost always the figure you need for calculations. And, when measuring totally random signals, such as noise, VRMS is even more significant because it represents one standard deviation using a Gaussian distribution.

RMS-sensing circuitry is more costly to design and calibrate than mean-sensing versions, so low-cost DMMs have been slow to provide true VRMS measurements. Most low-cost meters in use are still mean-sensing (see  “Scrap Mean-Sensing Meters”), but that situation is changing because all leading DMM vendors now offer handheld models that proclaim true-RMS performance. In fact, vendors, unable to contain their exuberance, award these new models a “true-RMS DMM” title (when if fact AC measurement constitutes only part of the instrument’s overall multifunction capability).

Adding true-RMS AC capability to a humble handheld is an industry achievement that should not pass unnoticed, so Test & Measurement Europe decided to mark the event with a hands-on test of several models. Using standards laboratory facilities (courtesy Wavetek Precision Measurements, Norwich, UK), we gave the true-RMS function on all models a similar and rigorous check. We used a range of voltage calibration test signals with pulsed outputs (see Figure 1) on the basis that, if the DMMs measured these waveforms satisfactorily, they would handle most waveforms in everyday use.

Figure 1. Test signals applied were 1 V RMS, with crest factors ranging from 5:1 to 10:1.

Broadly speaking, all the DMMs performed well although not always within their specifications. Most notably though, none of the models produced gross errors such as you would find using earlier mean-sensing models (see the Product Comparison Table). This situation held even for the most extreme case of a pulse input with a peak value 10 times above its RMS value and a mark:space ratio of 1:99. The DMMs still performed well, although this input was well beyond what vendors commit to in their specifications. The Product Comparison Table details the results of our measurements.

Know the Crest Factor
While vendors are swift to emblazon “true-RMS” on their instruments and supporting material, they have not given similar prominence to an equally important ACV relationship, called crest factor. While VRMS represents the power capability of an AC source, crest factor represents the degree of waveform distortion. A DMM’s crest factor specification is important because it tells you how much waveform distortion your DMM can accept before it produces bad readings.

The product specialists we spoke with had heard of crest factor, but those same specialists struggled to find it in their own specifications. All the detailed specifications include crest factor figures if you look carefully, but you’ll have to search in the small print. Crest factor is easy to understand because it is equal to the waveform peak divided by waveform RMS. Table 1  presents some values for crest factor.

While you would expect a true-RMS meter to read properly with a 1 V DC (when DC-coupled) or sine wave input, you can’t assume that same meter will read a pulse input without knowing the meter’s crest factor specification. In fact, the handheld DMMs we surveyed specify an input with a maximum crest factor of around 3:1.

It’s saturation within a DMM that sets the meter’s crest factor. This saturation can occur in the input circuitry of all meters, and in the squaring circuitry of meters that use analogue-computing RMS detectors. So, if the 3:1 specifications given apply at full-scale input, then you should expect the same meter to accept at least 6:1 crest factor inputs at half-scale, and so on. This relationship does hold to some extent but, remember, as you increase crest factor more and more (while holding pulse frequency constant), the pulse narrows and, therefore, contains more higher frequency components. Eventually, the DMM hits another design wall when those same high-frequency components fall outside the bandwidth of the DMM’s measurement circuits.

This situation poses a grey area of specification for AC measurement on any DMM, including top-end models. You have to assess yourself whether or not your DMM is likely to read correctly. For a 3:1 crest factor signal at 100 Hz it will; however, with a 3:1 crest factor signal at 20 kHz (if you can work out how to generate it), it certainly won’t.

The fundamental frequency of the pulse waveforms we applied was in the range 55 Hz to 220 Hz, which compares with the upper bandwidth limits of 20 kHz and 100 kHz on the DMMs. Using that range explains why we were able to successfully measure 10:1 crest factor inputs.

In practice, making high-confidence true-RMS measurements is a partial Catch-22 situation. First, in order to be sure you don’t saturate the DMM’s input, you need to know the crest factor of the input signal. But to know the crest factor, you need some idea of the true-RMS value, which is what you are trying to measure in the first place. One way out of this impasse is to view the signal on an oscilloscope so you can assess its RMS value and peakiness by observation. In practice, most engineers won’t bother with this step, and they’ll simply apply the DMM and hope for the best (in the same way we all did using mean-sensing DMMs). Our tests indicate that you’ll get away with it providing your RMS reading is well below full-scale. For example, while most handheld DMMs specify a 3:1 crest factor ratio (at full scale), we successfully measured a 10:1 crest factor signal at around 20% of full scale.

How Low Can You Go?
While suggesting you measure high crest factor signals at the low end of a DMM’s scale to avoid saturation, be careful that another DMM AC specification doesn’t catch you out on some models. Most DMMs — even top-end bench models — don’t measure true-RMS AC inputs down to zero on any scale. This inability is because the internal RMS detector uses a network of log, and anti-log amplifiers to handle the dynamic range you need when calculating true-RMS. You’ll have to study the specification’s small print again to reveal this limitation. Fluke’s 89 IV, for example, specifies accuracy for inputs only above 5% of scale. (Even top-end AC DMMs don’t measure below 1% of scale.) So, for example, if you short the input leads together, don’t be surprised if your DMM reads several digits of internal noise (up to 80 digits on the 89 IV). This zero error doesn’t add linearly to an input signal and becomes insignificant when you measure inputs above 5% of full-scale. The advantage of using the log amp analogue computing technique is the bandwidth it provides (up to 100 kHz on the 89 IV).

Tektronix’s Model TX3 doesn’t have this low-end limitation because it uses DSP-controlled signal sampling and a true-RMS algorithm to read true-RMS values. This technique does limit AC measurement bandwidth (20 kHz on the TX3), though, because sampling frequency needs to exceed maximum input frequency (which is not an easy feature to design into a handheld DMM if you want a measurement bandwidth up to 100 kHz).

True-RMS Results Round-Up
If you want to get picky, yes, some of the DMMs were outside specification, and vendors may like to suggest how this out-of-spec situation can happen. Overall, though, the message is clear. Handheld DMMs do a surprisingly good job at measuring true-RMS AC. For what these models cost (approximately FFr2500/DM750/£250) the price-performance ratio is exceptional.

Considering the performance we measured at 10:1 crest factor (well beyond what vendors specify) suggests you’ll encounter no problems measuring heavily distorted sine wave inputs. In measuring pulse trains, one important proviso is to be conscious of crest factor and use the meter away from the full-scale end of the range for crest factor inputs above 3:1.

Thanks go to Peter Crisp, Head of Metrology at Wavetek Precision Measurements, Norwich, UK, for his generous help in providing calibration facilities for conducting the tests.