Show floating point numbers in the Historian HMI
- Last UpdatedJun 25, 2024
The Historian HMI uses the IEEE 32-bit floating point format to show real numbers in an application. The IEEE floating point format is only an approximation of an actual real number. Unless the real number is an even power of two, it cannot be represented exactly using the IEEE 32-bit floating point format. The precision of an IEEE 32-bit floating point number is approximately eight decimal places.
When you want to show a real number in an Historian application, make sure the number does not exceed eight digits. The following floating-point number formats show valid real numbers within an Historian application:
#
#,###
#.##
0#
###.##
#.#######
###.#####
######.##
ABCDEF
###.####
Any floating-point numbers with more than eight digits are subject to rounding errors.
If you do not include a decimal in the format of the text, then the number is displayed with decimals, per the real format decimal precision configured in WindowViewer’s Advanced Format properties.
Note: If you add "#" to the left of a decimal, or if there is no decimal, do not limit the number of digits displayed.
Example 1
A Historian application should show the real number 2.3. But, the number 2.3 is not an even power of two and cannot be precisely represented by the IEEE 32-bit floating point format beyond 8 decimal digits.
To ensure the value 2.3 is shown from the application as the ASCII characters 2.3, the number must not exceed eight digits. If the number exceeds the eight digit maximum, the resulting number may be shown as 2.29999999 or 2.30000001 instead.
Example 2
When two real tags values are compared, the difference of two real tag values should be greater than FLT_EPSILON (value 1.19209290E-07F, in decimal 0.0000001192092896). However is the number exceeds 8 digits the resulting value may be incorrect. To correct for this multiply the value by 1000 or a larger multiple of 10. By doing this the value is be greater than 1e-7. Perform the necessary comparison operations and then divide by 1000, or the number you multiplied by.