As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Multimeter accuracy is written as a percentage of the reading plus a fixed number of counts, so the answer to which meter is most accurate depends entirely on what you are measuring and how much of the range you are using. The highest published figures belong to bench instruments: the Fluke 8808A bench multimeter states a basic direct voltage accuracy of 0.015 percent, which on a ten volt reading is one and a half millivolts. Beyond that level the error that decides your result is usually not the meter at all.
Accuracy is a percentage of the reading not of the display
The standard form of the specification has two terms. The first scales with what you are measuring, written as a percent of reading. The second does not, written as a number of counts or digits of the least significant place. Both have to be added, and the second term is what destroys a measurement taken near the bottom of a range.
Take the published 0.015 percent and work it across a decade.
| Reading on a direct voltage range | 0.015 percent of reading | As a share of a ten volt range |
|---|---|---|
| 10.000 V | 1.5 mV | 0.015 percent |
| 1.0000 V | 0.15 mV | 0.0015 percent |
| 0.1000 V | 0.015 mV | 0.00015 percent |
Read as a share of full scale the proportional term gets better and better as the reading falls, which is exactly why it cannot be the whole specification. The counts term is the floor underneath it, and the value of that term for this instrument is not published in the listing text. Not stated in the listing bullets, so the arithmetic above is the proportional half of the answer only. The practical consequence is the oldest rule in metrology: select the range that puts the reading in its top third, because a tenth of full scale multiplies the fixed error by ten relative to what you are reading.
Counts and digits describe resolution not accuracy
Counts tell you how many distinct values the display can show. The Fluke 177 True-RMS multimeter states 6000 counts with a digital update four times a second and a thirty three segment analogue bar graph updating forty times a second. Six thousand counts on a 600.0 volt range means steps of 0.1 volt, and on a 6.000 volt range steps of one millivolt. That is the grid the answer has to land on. It says nothing whatsoever about whether the answer is in the right place on the grid.
A handheld clamp can show a much finer grid and still be a field instrument rather than a reference. The Neoteck clamp multimeter states a 9999 count colour screen with a real time analogue bar, alongside a CAT III 1000 V and CAT IV 600 V safety rating and true RMS with a variable frequency drive mode. Those are the specifications that matter for its job, which is measuring up to 1000 A around a conductor in a panel. Clamping introduces its own errors, position in the jaw and magnetic coupling among them, which no number of display counts touches.
The digit notation works the same way. A 6.5 digit meter shows six full digits plus a leading digit limited in range, so the display can resolve about two million steps. The Hantek HDM3065X bench multimeter states 6.5 digit reading resolution, up to 30,000 readings per second, two sets of input terminals switchable from the front panel, and a graph and histogram display for recording without paper. Every one of those is a capability statement. A direct voltage accuracy figure for it is not stated in the listing bullets, which is the single specification you would want before using it as a reference. The resolution and the speed are published, the uncertainty is not, and those are different questions. The difference is set out further in what resolution means on a digital meter.
What a bench meter buys you over a handheld
A bench instrument starts with advantages that have nothing to do with the chip doing the conversion. It runs from a wall supply, so there is no battery voltage sagging under load and no low power design compromise in the front end. It sits still at a stable temperature, which matters because every analogue reference drifts with temperature and specifications are quoted at a reference temperature with a separate coefficient per degree outside it. It has room for a better voltage reference and better thermal layout around it. And it usually offers four wire resistance measurement, which the 8808A lists as a two by four terminal technique, removing the lead resistance that otherwise swamps any reading below a few ohms.
What it gives up is portability and category rating. A bench meter belongs on a bench with the circuit brought to it. A handheld belongs in a panel where the circuit cannot be moved, and its design budget goes into surviving what a panel can do to it. Choosing between them is a question about where the measurement happens, covered in how to choose a digital multimeter, not a question about which number is smaller.
The error sources that sit outside the specification
A published accuracy figure assumes reference conditions and a calibrated instrument. The errors that usually dominate a real reading are not in it at all.
- Loading. A meter with ten megohms of input impedance draws current from the circuit. On a high impedance node that current shifts the voltage you were trying to read. Low impedance modes deliberately load the circuit to kill induced ghost voltage, which fixes one problem by creating the other on purpose.
- Thermal voltages. Dissimilar metals at a connection generate microvolts that vary as your hand warms the terminal. Below a millivolt this is the limiting error on any bench, which is why precision measurements are made after a settling period and often reversed in polarity and averaged.
- Waveform. An averaging meter calibrated for a sine wave reads low on a chopped or distorted waveform. True RMS conversion is the fix, and it has its own crest factor limit and bandwidth. See what true RMS actually does.
- Temperature. Specifications are given at a reference temperature with a coefficient per degree beyond a stated band. A meter in an unheated plant room in winter is outside the conditions the headline figure describes.
- Calibration interval. Accuracy figures are quoted against an interval, commonly ninety days or one year. A longer interval means a looser number for the same hardware, so two figures are only comparable when their intervals match.
Reading an accuracy specification before trusting it
Four questions settle whether a published figure means what you want it to mean. Which function and range is it quoted for, since a basic figure is almost always direct voltage and everything else is looser. What is the counts term, because without it the figure is half a specification. What interval and what temperature band does it assume. And is the instrument in calibration, since none of this survives a meter that has never been checked against anything.
For most work the honest conclusion is that a well chosen mid range handheld on the correct range, with short clean leads and the right function, delivers a result closer to the truth than a higher resolution instrument used carelessly. Resolution is bought with money, accuracy is bought with technique, and the two are separate purchases. If you are comparing specific handhelds rather than classes of instrument, the figures worth lining up are listed in the Elejoy multimeter notes, where the same two part formula applies.
Frequently Asked Questions
What does basic accuracy mean on a multimeter?
It is the accuracy of the single best function and range, almost always direct voltage, under stated reference conditions and within a stated interval since calibration. Alternating voltage, current and resistance are always looser, often by a factor of ten or more, so a headline figure describes one corner of the instrument.
Is a 6.5 digit meter more accurate than a 4.5 digit one?
It is more precise in the sense of showing finer steps, which is not the same thing. Resolution is how small a change the display can represent, accuracy is how close the reading is to the true value. A meter can resolve a microvolt and still be out by a millivolt.
Why does my reading change when I touch nothing?
Thermal voltages at the connections, noise pickup on the leads, and the last digit of a high resolution meter following real variation in the source. At the microvolt level the junction between a copper lead and a brass binding post behaves like a small thermocouple, which is why precision work uses matched materials and short settling times.
Does a higher count meter need calibration more often?
It needs it more visibly. A drift of a few parts per million is invisible on a 6000 count display and obvious on a 6.5 digit one. Calibration intervals are set by the manufacturer and by what the measurement is used for, not by the display.
Can lead resistance ruin an accurate measurement?
On voltage, almost never, because the meter draws very little current. On resistance it is the dominant error at low values, which is why four wire measurement exists: two leads carry the current and two separate leads sense the voltage directly at the part.