ET online
November 2007

# Let's be accurate about accuracy!

02 November 2007

Kerry Burdett - Technical support adviser

There’s no doubt that test equipment needs to be accurate, as inaccurate test results can lead to potentially dangerous electrical systems being certified as safe. The way that instrument accuracy is specified, however, is not always as straightforward as it might seem.

If you were offered two digital meters, one with an accuracy of ±2.5% and ±5 digits, the other with an accuracy of ±5% and ±2 digits, which would you expect to give more accurate results? Most people would say the first one, but they would be wrong. In normal use, the second instrument is the more accurate.

Clearly, instrument accuracy merits some explanation. Let’s start by making it clear that accuracy is specified in a different way for analogue and digital instruments. With analogue instruments, accuracy is usually quoted as a percentage of full-scale deflection (FSD). For example, if an instrument has an accuracy of ±2% FSD on the 100 V range, any reading taken on that range can be in error by up to ±2 volts.

That’s not much of a concern for readings toward the high end of the scale where a measuring, say, a 70 V supply might produce a reading between 68 V and 72 V. It’s much more of a concern, however, if the same range is used with low voltages, where measuring, for example, a 4 V supply could produce a reading anywhere from 2 V to 6 V!

The message is simple - with analogue instruments, to ensure reasonable accuracy, always take measurements on the range that gives the greatest deflection.

With digital instruments, accuracies are usually quoted as a percentage of the measured value, and a number of digits. For example, an instrument might be quoted as having an accuracy of ±2% and ±3 digits. If such an instrument is connected to 100 V source it could, as a result of the percentage accuracy alone, read anywhere from 98 V to 102 V.

However, we also have to consider the ±3 digit error which means that the final figure in the result can be out by 3. Hence, connected to an accurate 100 V source, this instrument might read anywhere from 95 V to 105 V.

In practice, things are usually much better than this, as the digit error always refers to the last digit in the display. Most instruments these days have displays with at least four digits and, when a fourdigit instrument is used to measure 100 V the result will be shown to one decimal place. A digit error of ±3 is, therefore, equivalent to just ±0.3 V rather than the ±3 V in our example above, which assumed that were using an instrument with only a three-digit display.

Nevertheless, the use of two factors to specify the accuracy of a digital instrument can lead to confusion. The graph shows a comparison between two loop testers – tester X, with an accuracy of ±2.5% and ±5 digits, and a Megger instrument with an accuracy of ±5% and ±2 digits.

Although tester X looks more accurate it’s easy to see from the graph that the Megger instrument is always more accurate for readings below 0.8 ohms, which, for loop testing, is where the majority of results should lie.

While the quoted accuracy of a meter is a very useful guide to its performance, it would be unreasonable to expect any meter to retain this accuracy indefinitely without checking or attention. For this reason, Megger strongly recommends that instruments are regularly recalibrated, typically at yearly intervals.

But how can accuracy be ensured between calibrations? To make this easy, Megger offers specialised test boxes, the CB101 for 5 kV insulation testers and the VCM100 for use with oil test sets.

Accuracy in test equipment is vital, but fortunately it’s not hard to achieve provided that, when buying a new instrument, you are clear about what the accuracy specifications mean and that, after you’ve bought your instruments, you check them regularly while they’re in service.