Understanding Multimeter Accuracy: What Does +/- (0.8% +1) Really Mean?

In digital measurement systems, the last digit in the least significant position is treated as the margin of error of the analog / digital converter. The measurement value displayed in increments, e.g. every 1 unit, shows the state of the comparator and does not correspond to the actual measured value, which may assume intermediate values between the comparator switching thresholds.
Regards.

The number of bits of the analog-to-digital converter is one weld and the number of digits in the display is another matter.
For a 3-digit meter, the reading is 0-999 and regardless of the quality of the converter, the result will be included in these 3 digits. e.g. 0-200V has an accuracy / resolution of 1V.

We also have displays 3.5 digits, 4 digits, etc. The more, the more accurate the indications.

Regards

Hello

lordpakernik wrote:

Accuracy: +/- (0.8% +1)

Well, maybe I will also try to write something. In the old analog devices there was a parameter known as the accuracy class and everyone (well, almost everyone) knew what it was and how to approach it. The measurement uncertainty was due to the inaccuracy of the analog measurement unit, because it was the only one.
In digital meters, the measurement uncertainty is a component of the uncertainty of the measured value (here 0.8%) as well as a component of the uncertainty of the measuring range (here 1 digit). In the meters (and finally it is a voltmeter) there is a reference voltage, i.e. a reference voltage to which the value of the input (measured) voltage is compared. It should be remembered that both components of the measurement uncertainty come from terms analog . The measurement uncertainty of the digital element is 0 If the digital module starts to be wrong it should be replaced
PS Unfortunately, it is not so rosy that the more numbers, the greater the accuracy. There are many meters where, by definition, the last digits are something like a "random number generator" - such constructions have the right to exist on the market because they are calculated for buyers who only count the number of digits on the display instead of reading the manual (if any)
greetings
pm001

Hello, the error you provided:
+ / - (0.8% +1)
means:
+/- (0.8% ww + 1c)
where ww-indicated value
c - value of the last digit on the display
for example:
for the given values: + / - (0.8% ww + 1c), measuring range 1.999V and measuring value 1V
the permissible error of the instrument is:
0.8% of the indicated value
that is: 0.8% ww * 1V = 0.008V = 8mV
plus:
the error of the last digit, i.e. 1c * 0.001V = 0.001V = 1mV
total error: +/- (8mV + 1mV) = +/- 9mV

Hmm ... however, I prefer the term measurement uncertainty rather than error.
Contrary to appearances, it is not synonymous.
greetings
pm001

formant wrote:

the permissible error of the instrument is:
0.8% of the indicated value
that is: 0.8% ww * 1V = 0.008V = 8mV
plus:
the error of the last digit, i.e. 1c * 0.001V = 0.001V = 1mV
total error: +/- (8mV + 1mV) = +/- 9mV

I think my friend got a little confused :?:

How will a colleague read values below 1V for an example range of 200V and 3 digits :?:

Regards

After all, he did not write that it is the 200V range. After all, it can range up to 2V

sm48 wrote:

After all, he did not write that it is the 200V range. After all, it can range up to 2V

That's right, but for the example 200V range, these considerations don't make sense.

There are two errors in digital meters.

For small ranges, the quantization error is important, and for large ranges, the error of the last digit / resolution is prevalent.

Regards

The manufacturer provides a measurement uncertainty table and there is the measurement range, its resolution and the contributed uncertainty to the measurement in percent and number.
The number relates to the resolution and the number is multiplied by the resolution.
Example:
range 100? resolution 0.01? uncertainty 3% + 3 digits
and now we have the measurement result 81? 81 + 3% = 83.43 and to this we add the digital error, i.e. 3 x 0.01 = 0.03 that is 83.43? + 0.03? = 83.46? 83.46 - 81 = 2.46 this is the measurement uncertainty.

Hello
In fact, everything is relative in nature here too.
The value of the last digit obviously depends on the measuring range and cannot be defined permanently. (It is true that the maximum range of the test voltage is 2V (at least this is what I remember), but since there are systems that match the measured signal, do not bother about how it happens inside the meter.) Regarding the first example, the mV value may or may not be correct as it looked like that we would be a bit comical to measure the current or resistance (I prefer not to mention others) and estimate the measurement uncertainty in volts.
The same is true of the second example, you cannot add a unitless number to the uncertainty of the measured value expressed in volts / amps / ohms in 5 and 10.
Maybe I shook the translator a little, but the idea is that in order to add something to the nominated value, this something must also be a nominated value in the same units and it looks like this
measurement result: 123 V / A / ?
extreme measurement uncertainty: +/- 2.23 V / A / ? ( because 1% was assumed = 1.23 V and 1c = 1V)
I pay attention to the term "extreme" because in nature it does not have to be this way that both components will be added, but this is the next lesson :D

greetings
pm001

Hello, colleague formant well described the accuracy of the measurement.
It is the percentage of the measured value + the number of digits in the last position (according to the meter specification, for different ranges and frequencies, these values may be different)

I enclose the instructions for the metrology exercise "Development of measurement results"
When measuring once, it is worth paying attention only to the type B uncertainty
Link inactive
KR moderator

In the attachment you have a description of what and how the given number represents, as well as calculation examples of uncertainty.

Take a look here http://www.chem.uw.edu.pl/people/AMyslinski/informator_08/Pracownie/fiz_i_radio/analiza.pdf there is an example on a few multimeters how to count uncertainty (from page 13)

exactly that's the point:
absolute measurement error = ? (% of indicated value + m * least significant digit)
otherwise
absolute measurement error = ? (% of the indicated value + m multiple resolution).