Modbus Word to Float Conversion (Big-Endian Byte Swap)

When transferring floating-point values via Modbus TCP/IP, devices often split a 32-bit floating-point number into two 16-bit Modbus registers.

However, many industrial devices send the bytes in a different order than the receiving PLC expects. This is why we often need to apply a Big-Endian Byte Swap before converting the registers to a floating-point value.

This guide explains:

• Why byte swapping is needed

• What Big-Endian Byte Swap (BADC) means

• How to convert registers into a REAL / float

• Multiple real industrial examples

1. Why Byte Swapping is Needed in Modbus

The Modbus protocol was originally designed to transfer 16-bit register values.

A register contains:

16 bits = 2 bytes

But a floating-point number requires:

32 bits = 4 bytes

So a device must split the float across two registers.

Example float:

0.1251

IEEE-754 representation:

0x3E001ADB

Split into bytes:

A   B   C   D
3E  00  1A  DB

Stored in Modbus registers

Register 1 = A B = 3E00
Register 2 = C D = 1ADB

This layout is called:

ABCD

---

2. The Industry Problem

The Modbus specification does not define how 32-bit values should be arranged across registers. Different vendors, therefore, implemented different byte orders.

Common layouts used in industry:

Layout	Byte Order	Description
ABCD	Big Endian	Standard order
CDAB	Word Swap	Registers swapped
BADC	Byte Swap	Bytes swapped within registers
DCBA	Full Reverse	Completely reversed

Because of this, PLCs often must rearrange the bytes before converting to a float.

---

3. What Big-Endian Byte Swap Means

Big-Endian Byte Swap rearranges the bytes from:

A B C D

to:

B A D C

This means:

• swap bytes inside each register

• keep register order the same

Visual representation:

Original Registers

Register1      Register2
+----+----+    +----+----+
| A  | B  |    | C  | D  |
+----+----+    +----+----+

After Byte Swap

+----+----+----+----+
| B  | A  | D  | C  |
+----+----+----+----+

---

4. Example - Real Industrial Data

Modbus registers from a vibration sensor:

HighWord = 62
LowWord  = -9446

Convert to HEX:

62    = 0x003E
-9446 = 0xDB1A

Bytes:

00 3E | DB 1A
 A  B   C  D

Apply Byte Swap (BADC):

3E 00 | 1A DB
 B  A   D  C

Combine into 32-bit value:

0x3E001ADB

Convert to float:

0.1251

---

5. How PLCs Perform This Conversion

Typical steps used in PLC or C/C++ code:

Step 1 - Extract bytes

A = HighWord >> 8
B = HighWord & 0xFF
C = LowWord  >> 8
D = LowWord  & 0xFF

Step 2 - Rearrange bytes

B A D C

Step 3 - Combine into a 32-bit value

(B << 24) | (A << 16) | (D << 8) | C

Step 4 - Interpret as float

REAL value

6. Why Industry Uses Byte Swapping

Byte swapping is common in industrial automation because devices may have:

Different CPU architectures

Examples:

• ARM processors

• Intel processors

• DSP controllers

• Embedded microcontrollers

Each architecture may represent multi-byte values differently.

---

Legacy Modbus implementations

Many industrial devices were designed decades ago, when memory layout decisions differed between manufacturers.

Therefore:

Vendor A → ABCD
Vendor B → BADC
Vendor C → CDAB

PLCs must adapt to the device format.

---

Sensor manufacturers

Common devices requiring byte swapping:

• Vibration sensors

• Power meters

• Drives

• Temperature transmitters

• PLC-to-PLC communication

• SCADA systems

---

9. How Engineers Detect the Correct Byte Order

A common debugging trick:

1. Read the two registers

2. Try all four byte layouts

3. See which one produces a realistic value

Example:

Expected vibration value:

0 – 10 g

Possible interpretations:

ABCD → 3.2e-38
CDAB → -1.4e12
BADC → 0.125
DCBA → 6.3e-20

Correct format is clearly:

BADC

---

10. Summary

Key points:

• Modbus registers are 16-bit

• Floats require 32-bit

• Devices split floats across two registers

• Modbus does not define 32-bit ordering

• Different vendors use different layouts

• Big-Endian Byte Swap (BADC) rearranges bytes from:

A B C D → B A D C

This ensures the PLC reconstructs the correct IEEE-754 floating-point value.

---

11. Practical Tip

Whenever a Modbus float looks incorrect:

1. Check register order
2. Check byte order
3. Try ABCD / BADC / CDAB / DCBA

In many industrial sensors, the correct format is:

BADC (Big-Endian Byte Swap)

You can verify the conversion using this free tool: https://g2384.github.io/collection/Hex_Calc_IEEE754_conversion.html