Understanding Accuracy Specifications for Digital Multimeters and Test Equipment

"Accuracy" actually means "inaccuracy", or uncertainty in metrology terms.

This is the worst-case allowable error to the true value being measured,

• across a defined percentage of the total number of units manufactured. 

These specifications are stated to be valid over a set period of time, typically one year after the date of purchase. 

Calibration at regular intervals documents the actual performance of your individual tester, which can dramatically improve the accuracy performance.

 

Accuracy is specified in a percentage of error: % of Reading, or % of Range

• Often with an added "floor", or a "+ or - number of counts"  : [ .5% +- 5 counts ]

What does this mean? 

Example: 12V DC [.5%] of Reading: 

Measured Value is specified to be between (12V + .5%) and  (12V - .5%)

 

To calculate the specification to a particular reading: 

.5% of 12V = 12V x .005 = .06V

The accuracy (uncertainty) of the meter reading will be within 11.94V to 12.06V: 

• 12V  - .06V = 11.94V 
• 12V + .06V = 12.06V

​The displayed measured value is specified to be within 11.94V to 12.06V of the actual real voltage

​

​Look for a footnote on how long (time span specification) the meter accuracies are specified for.

• For everyday meters and electrical testers, this is typically a "1 year specification"
  • High quality meters typically hold their specifications for much longer.
    • ​I have a 30 year old meter that is still reading within specification.

For regulated industries such as Pharmaceuticals,  this time span is the maximum time allowed by regulators between calibration certifications of just about every device used on the production line.  

• Auditors check the dates printed on the calibration stickers attached to test equipment, sometimes the first thing they check. 
  • Plants have been shut down when critical equipment and devices are found to be beyond their calibration due dates.
  • ​More information on Calibration will be available soon at this Link. 

 

​​Each Measurement Function typically has different accuracy specifications.

• This is because different internal meter electronic circuits are used for each type of measurement.

DC V is usually the best (lowest %, highest accuracy) specification, and highlighted as the "banner" specification.

• Make sure to check the accuracy specification for the measurement function you are using.  
• Why? Inside the meter, most function measurement types are converted into DV volts before being run through the A/D converter to the display. The specification for the measurement you are taking could be significantly worse than the banner DC volts accuracy. 

 

 

• This is referred to as the "Test Uncertainty Ratio", or TUR . 

For example, if your test procedure calls for a measured value of  "12V DC plus or minus 5%" , 

• the Accuracy of the meter being used for the measurement should ideally be .5%  or better
• for the measurement function and range used for this measurement. 

​​Please note: A 10:1 TUR is not always possible. Minimum TURs are specified as 4:1 in some industries.

• A TUR of 10:1 or higher helps you have more confidence in the tester's measurement readings over a longer period of time.

 

% of Reading verses % of Range:

•  1% of 1000V (plus or minus) = 1000V x .01 =   +/- 10 V. ​

A reading anywhere between: 

• 120V - 10 V = 110V
• 120V + 10V = 130V
• 120V to 130V is within specification

is within specification on a true value of 120 V ,

• which is actually  +/- 8.3% of reading.

• 118.8V to 121.2V is within specification.

 

Number of Counts [x% + # of Display Counts]  in Specification

• can significantly affect measurements taken on the low end of the range.
  • ​why it is best to take readings in the mid to high end of the range 

There may also be a "floor", or least significant digit unstable. This is a multiplier of the minimum resolution, or sensitivity, on each measurement range: (see Ranging and Resolution)

Example: Accuracy spec = +/-(1 % of reading +2 counts), using a 6000 count meter:

Printed specifications indicate the  worst-case allowable error to the true value being measured,

across a defined percentage of the total number of units manufactured. 

The percentage of total number of units manufactured is listed in the specifications as the "Sigma" or standard deviation factor used by the manufacturer.

The "sigma" factor represents the quality of the published specifications: 

• A higher sigma indicates a higher percentage of the total number of units manufactured are and will remain within the stated specification for the time period specified. 
• A higher sigma can indicate that your individual meter will measure with much better accuracy than the worst-case specifications in the datasheet.  
  • All electronic curcuits drift with time
  • Higher sigmas on specifications indicate the manufacturer as taken this into account, either by manufacturing products well within the stated specification, and/or using two components in the same curcuit that drift in opposite ways

​

Manufacturers use 1, 2 or 3 sigma specifications.

A tester with a  1 sigma specification can look better on paper than one with a  3 sigma specification , when the actual performance could be worse.

The "sigma" level of the specification indicates the quality of the specification, or how many of the units manufactured will meet this specification after the specified time period. Industrial meter products are typically one or two years, 30 days to 90 days for higher accuracy products.

A 1 sigma ( 1 standard deviation) specification means the manufacturer is stating that 68% of the units manufacturer will meet the accuracy specification for the specified period of time.

A 2 sigma (2 STDs)specification means the manufacturer is stating that 95.43% of the units manufactured will meet the accuracy specification for the specified period of time..

A 3 sigma  (3 STDs)specification means the manufacturer is stating that 99.73% of the units manufactured will meet the accuracy specification for the specified period of time..

 

Why is this significant?

All electronic circuits have components that change value, or drift , over time. A quality design, with conservative 3 sigma specifications takes this into account and builds corrections for this drift into the circuit. For example, two components with opposite drift characteristics are used instead of just one. As the first drifts in one direction, the second drifts in the opposite direction, canceling each others' changes  out to maintain a better measurement accuracy over a longer period of time.

 

Need to know more about comparing specifications? There is  more on  "specsmanship"  in my archives, including a short presentation. Let me know if you would like a copy by using my contact form.

Additionally, feel free to send me your questions, or instruments you wish to compare, using the contact form and I'll try to help. 

 

How can accuracy specifications affect you in everyday life? 

Accuracy Calculations Example 1: Spend $4 a gallon on 2 gallons you did not receive 

Given a gas pump with  an up to 200 gallons dispensing meter,  accuracy is 1% of full scale

• Accuracy = 1% of full scale = 1% of 200 = plus or minus 2 gallons
• Reading  on pump of  "100" gallons could be 102 or 98 gallons (2% off)
• Reading  on pump of  "10"  gallons could be 8 or 12 gallons (20% off)

• Accuracy = 1% = plus or minus 1% of value read
• Reading of  "100" gallons could be 101 or 99 gallons
• Reading of  "10"  gallons could be 10.1 or 9.9 gallons

 

Gas pumps have much higher accuracy,  to match the resolution of the pump.

• This example is just to make it easier to visualize the difference between specifications that look the same except for a footnote you can miss. 

Calibration sticker for each pump can usually be found on the side of the pump, typically checked and issued from your state's Division of Weights and Measures. 

Accuracy Calculations Example 2

Measuring 120V AC  on  a 3 1/2 Digit, 4000 Counts DMM:

• 400V Range Specification: [1.5% + 4 ] (% of reading + # of counts)

Step 1:  Accuracy % = [1.5% of reading]:   

Measured value = 120V AC

• 120 * [1.5%] = 120 * .015 = 1.8V  = +/- [1.8V]

Step 2: # of Counts factor = [4] counts of resolution:   

400.0V range has .1V resolution

•  “+4” counts =  4 x .1V = .4V  = +/- [.4V ]

Step 3:  Total worst-case deviations = [1.5% +4] = [1.8V + .4V ] =  +/- 2.2V

Worst case specified error limits are between

• (120V - 2.2V) and (120V + 2.2V)

Worst case specified measured value difference from the real world analog 120V value  is:

• between a lower limit of  117.8V AC and an upper limit of 122.2V AC

​

Absolute Accuracy on 120V is 1.8% 

• 2.2V is 1.8% of 120V