Having built a small, portable 10.000V reference using the Analog Devices AD587 reference chip, now is a good time to evaluate its performance with a bunch of multimeters.

However, I have a feeling that the AD587 with its 10.000V ±5mV @25ºC is going to be the better performer than some of the multimeters being tested. In today’s performance shootout, ranked in order of their pedigree we have:

- Fluke 789 Processmeter
- Fluke 12 Multimeter
- Extech EX330
- Pro-Tester 03-150K

Evaluating each of the meters on their face value is unfair and unscientific. So, datasheets for all but the Pro-Tester multimeter were found and are detailed in the table below. Only details for the minimum measuring range that can be used to measure 10V are given. For example, the Fluke 789 has four DC voltage ranges, 4.000V, 40.00V, 400.0V and 1000V, therefore specifications for the 40.00V range given.

Multimeter | Range (V dc) | Voltage Resolution (V) | Voltage Accuracy |
---|---|---|---|

Fluke 789 Processmeter | 40.00 | 0.01 | ±0.1% + 1 count |

Fluke 12 | 40.00 | 0.01 | ±0.9% + 1 count |

Extech EX330 | 40.00 | 0.01 | ±1.0% + 2 counts |

Pro-Tester 03-150K | 20.00 | 0.01 | unknown |

So the first problem is apparent, none of the above meters have enough resolution at the required range to properly evaluate the AD587’s performance of 10.000V ±5mV. Each of the above multimeters will only display the voltage to a maximum of 2 decimal places, that is, 10.00V.

Resolution in the required range is something important to consider if you need to precisely measure a value of voltage or current or resistance. For general hobby electronics the reduction in resolution is unlikely to be a critical issue. However, if you were measuring a signal from a precise scientific instrument or sensor, this reduction in resolution may be significant.

Anyway, we’ll make do with what we’ve got and see how each of the meters perform. Lets assume that the AD587 is operating perfectly within specification and is outputting exactly 10.000V. Lets use the AD587 as a voltage standard to test the meters’ accuracy, that is, how close each of the meters read to 10.000V.

Before we can evaluate the results, we need to understand the potential measurement error of each meter.

Excluding the Pro-Tester meter, each of the meters detail their accuracy in terms of a percentage error and “counts”. Calculating the accuracy of a meter’s measurement is straightforward. The percentage value in the specification tells us that the meter will measure within ±y% of the actual value. So for example, the Fluke 789 has a voltage accuracy of 0.1%. At 10V, 0.1% x 10V = 0.01V. This value establishes the boundary of error in the measurement, that is 10V -0.01V to 10V + 0.01V, which equates to 9.99V to 10.01V.

The “counts” value tells us how many least significant digits the measurement will be incorrect by. For instance, when the Fluke 789 is operating in the 40.00V range, the least significant figure that will be displayed is the bold figure in “40.0**0**V”. At this range, 1 count equates to 0.01V. For the Extech EX330, 2 counts is 0.02V.

Once the percentage error and counts are known, they can be combined to determine the total error of the measurement. When calculating the high end, the percentage error and counts are added together. For the low end, the percentage error and counts are subtracted.

Each of the accuracies are shown worked out in the following table.

Meter | Voltage Accuracy | Error % at 10V | Counts Error | Low End
Error |
High End Error |
---|---|---|---|---|---|

Fluke 789 | ±0.1% + 1 count | ±0.01V | ±0.01V | 10V – 0.01V – 0.01V = 9.98V | 10V + 0.01V + 0.01V = 10.02V |

Fluke 12 | ±0.9% + 1 count | ±0.09V | ±0.01V | 10V – 0.09V – 0.01V = 9.90V | 10V + 0.09V + 0.01V = 10.10V |

Extech EX330 | ±1.0% + 2 counts | ±0.1V | ±0.02V | 10V – 0.1V – 0.02V = 9.88V | 10V + 0.1V + 0.02V = 10.12V |

So, how did the meters perform? As can be seen, both Flukes and the Extech measured well within their accuracy specification. From these results, it is fair to say that the AD587 would be operating very close to its 10.000V specification.

Meter | Voltage Accuracy | Acceptable Results at 10V | Measured Voltage | Pass? |
---|---|---|---|---|

Fluke 789 | ±0.1% + 1 count | 9.98V to 10.02V | 10.00V | PASS |

Fluke 12 | ±0.9% + 1 count | 9.90V to 10.10V | 10.02V | PASS |

Extech EX330 | ±1.0% + 2 counts | 9.88V to 10.12V | 9.99V | PASS |

Pro-Tester 03-150K | unknown | unknown | 10.25V | – |

However, the Pro-tester seems a bit off the mark. Assuming that the AD587 is outputting 10.00V, then the Pro-Tester’s measurement has an error of 10.25V/10.000V x 100% = 2.5%.

Interestingly, when these results were gathered, the Pro-Tester was initially indicating that it had a low battery. A low battery did significantly impact its performance with the meter measuring a value of 11.42V, an error of 14.2%! Once changed to a fresh battery, the results improved to 10.25V.

So, if you ever need to precisely measure a value, you need to understand the resolution and accuracy limitations of your measuring instrument. If you are measuring a sensitive output, then you need to ensure your measuring instrument is suited for the task. Once a suitable instrument is chosen, calculating the accuracy of the instrument is very straightforward too.

Also, if the low battery indicator is showing, change the battery! A low battery compromises the meter’s ability to measure accurately. This could have significant consequences if your are measuring higher voltage levels, such as mains 240VAC, to prove de-energised before you commence work.