Multimeter Accuracy Calculator
Turn a multimeter's ±(% + digits) accuracy spec into an actual uncertainty range, and check whether your theoretical value falls inside it.
Reading the spec
Accuracy is usually listed as ±(0.8% + 3): 0.8% of the reading plus 3 counts of the last digit. One count (the resolution) is the range divided by the display counts, so the 20 kΩ range of a 2000-count meter resolves 10 Ω. Accuracy differs by function and range, so use the matching line in your meter's manual.
Worked example: 4.62 kΩ, ±(0.8% + 3), 20 kΩ range
- Resolution (1 digit): (20 kΩ)/(2000) = 10 Ω
- Uncertainty: 4.62 kΩ × 0.8% + 3 × 10 Ω = 37 Ω + 30 Ω = ±67 Ω
- So the true value lies between 4.553 kΩ and 4.687 kΩ. The marked 4.7 kΩ is outside, so the meter alone does not explain the difference.
Resolution (1 digit) by range and counts
| Range | 2000 counts | 4000 counts | 6000 counts |
|---|---|---|---|
| 200 Ω | 100 mΩ | 50 mΩ | 33.3 mΩ |
| 2 kΩ | 1 Ω | 500 mΩ | 333 mΩ |
| 20 kΩ | 10 Ω | 5 Ω | 3.33 Ω |
| 2 V | 1 mV | 500 µV | 333 µV |
| 20 V | 10 mV | 5 mV | 3.33 mV |
More counts mean a smaller step on the same range. Choose the smallest range above the reading to keep the digits part of the error small.
Using it in practice
Step by step
- Find the accuracy line for your range in the multimeter manual, for example ±(0.5% + 2) on 20 V DC.
- Enter the reading, the percent part, the digit count and the resolution of that range.
- Write the result as reading ± uncertainty, and use the interval when you compare it with theory.
Common mistakes
- Treating "digits" as a percentage. "+2 digits" means two counts of the last displayed digit. On a range that shows 0.01 V steps, that is ±0.02 V, regardless of the reading.
- Measuring on a range that is too high. The digit term is fixed, so a small reading on a large range has a much bigger relative uncertainty. Use the lowest range that does not overload.
- Using the DC spec for AC. AC ranges have their own, usually wider, accuracy and are only valid over a stated frequency band. Many cheaper meters are also not true-RMS, so non-sine waveforms read wrong.
Other error sources
The spec holds only within the stated temperature range, often 23 ± 5 °C, and within the calibration period. On current ranges the meter's shunt drops a burden voltage that lowers the current in the circuit itself. On voltage ranges the 10 MΩ input resistance loads high-impedance circuits. If your error is bigger than the interval this calculator gives, look at those effects before blaming the circuit.
Worked example: the same voltage on two ranges
A 2000-count meter with a DC voltage spec of ±(0.5% + 2) reads 1.85 V.
- 2 V range (one count is 0.001 V, display 1.850): 1.85 V × 0.5% + 2 × 0.001 V = 0.00925 + 0.002 = ±0.011 V.
- 20 V range (one count is 0.01 V, display 1.85): 0.00925 + 2 × 0.01 V = ±0.029 V.
Same meter, same voltage, almost three times the uncertainty, all of it from the digit term. That is why the lowest range that does not overload is worth the extra button press.
Deciding whether theory and measurement agree
Turn both sides into intervals before you compare. The measurement above is 1.839 V to 1.861 V on the 2 V range. If your calculation, including part tolerances, says 1.80 ± 0.05 V (1.75 V to 1.85 V), the two intervals overlap, so the difference is explained by the parts and the meter. If they do not overlap, something systematic is going on: meter loading, a wrong part value or an assumption in the theory. A percent error from the percent error calculator tells you how big the gap is, and this interval tells you whether the gap is meaningful.
Frequently asked questions
What does ±(0.5% + 2) mean on a multimeter?
The error is 0.5% of the reading plus 2 counts of the last displayed digit. On a 2000-count meter's 20 V range the last digit is 0.01 V, so a 12.00 V reading is uncertain by 0.06 V + 0.02 V = ±0.08 V.
Why does the range setting matter?
The digit part of the error is a count of the last digit, and its size depends on the range. Measuring on the lowest range that still fits the value gives the finest resolution and the smallest error.