This IEEE 754 float converter splits a float into sign, exponent and mantissa: 20.5 is 41A40000 as a float32, and 0.1 is stored as 0.100000001490116, because it has no exact binary form.
- Hex (big-endian)
–- Value stored
- –
- Difference from what you typed
- –
- Sign · exponent · mantissa (bits)
–
- Sign
- –
- Exponent field
- –
- Mantissa field
- –
- Kind
- Normal numberSubnormal (very small)ZeroInfinityNot a number (NaN)
Enter a decimal number, or 8 hex digits for float32 and 16 for float64.
How it works
How is a number stored as an IEEE 754 float?
A float32 has 1 sign bit, 8 exponent bits (bias 127) and 23 mantissa bits; a float64 has 1, 11 (bias 1023) and 52. A normal value is (−1)sign × 1.mantissa × 2exponent − bias.
Most decimals cannot be stored exactly: 0.1 as a float32 is 0.100000001490116. The difference row shows how far the stored value is from what you typed.
An exponent of all zeros means zero or a subnormal value; all ones means infinity (mantissa zero) or NaN.
Glossary Modbus register
Questions
Questions about the IEEE 754 float converter
Why is 0.1 stored as 0.100000001?
Binary fractions can only add up halves, quarters, eighths and so on; 0.1 has no exact binary form, so the nearest float32 is stored instead.
My device sends floats in two Modbus registers. Which order?
Try the byte orders in the Modbus register decoder: ABCD is big-endian; CDAB, BADC and DCBA are the common swaps.
More tools

