Data formats, Floating point data format (float) – Basler Electric DGC-2020HD Modbus Protocol User Manual
Page 14
8
9469300991 Rev A
General Information
DGC-2020HD Modbus
™ Protocol
CRC Hi error check
CRC Lo error check
Response
The response message echoes the Query message after the register has been altered.
Data Formats
DGC-2020HD systems support the following data types:
•
Data types mapped to 2 registers
o Signed Integer 32 (Int32)
o Unsigned Integer 32 (Uint32)
o Floating Point (Float)
o IP Address (IP Address)
o Strings maximum 4 characters long (String)
•
Data types mapped to 1 register
o Unsigned Integer 16 (Uint16)
o Unsigned Integer 8 (Uint8)
o Strings maximum 2 characters long (String)
•
Data types mapped to more than 2 registers
o Strings longer than 4 characters (String)
Floating Point Data Format (Float)
The Modbus floating point data format uses two consecutive holding registers to represent a data value.
The first register contains the low-order 16 bits of the following 32-bit format:
•
MSB is the sign bit for the floating-point value (0 = positive).
•
The next 8 bits are the exponent biased by 127 decimal.
•
The 23 LSBs comprise the normalized mantissa. The most-significant bit of the mantissa is
always assumed to be 1 and is not explicitly stored, yielding an effective precision of 24 bits.
The value of the floating-point number is obtained by multiplying the binary mantissa times two raised to
the power of the unbiased exponent. The assumed bit of the binary mantissa has the value of 1.0, with
the remaining 23 bits providing a fractional value. Table 4 shows the floating-point format.
Table 4. Floating Point Format
Sign
Exponent + 127
Mantissa
1 Bit
8 Bits
23 Bits
The floating-point format allows for values ranging from approximately 8.43X10^
-37
to 3.38X10^
38
. A
floating-point value of all zeroes is the value zero. A floating-point value of all ones (not a number)
signifies a value currently not applicable or disabled.
Example: The value 95,800 represented in floating-point format is hexadecimal 47BB1C00. This number
will read from two consecutive holding registers as follows:
Holding Register
Value
K (Hi Byte)
hex 1C
K (Lo Byte)
hex 00
K+1 (Hi Byte)
hex 47
K+1 (Lo Byte)
hex BB
The same byte alignments are required to write.