Modbus word order for floating point

[Ken Moore 23]

Okay, I have a Control Logix PLC with a Prosoft serial modbus card passing data to a DCS, works great. However...(isn't there always a however), we are moving to a new DCS, and I need to know the word order that the floating point is sent over by. I contacted Prosoft, and they said AB controlled the word order. Does anyone know if AB complies with the IEEE standard? Least significant word first? thanks, Ken

[ratcliffe_ic 2]

I have some information here that the A-B floating point format complies with IEEE-754.

[Ken Moore 23]

Can you share, what document?

[ratcliffe_ic 2]

Document attached How floating point numbers are represented in binary.doc

[b_carlton 61]

I believe he's asking how a real is transmitted via Modbus and the Prosoft card, not how it's stored in memory.

[DanW 41]

If the connection is Modbus/TCP over ethernet, then Modbus TCP can handle multiple masters, or rather, ethernet through a switch can handle sequential requests from more than one Modbus master. So a test using a Modbus Master like Modscan23 or Simply Modbus could be used to poll a pair of registers in the CLX, grab a floating point value. and see whether the value comes through correctly as a "swapped" or not-swapped value. It's hard to believe that any moderm DCS would not have the ability to switch from little to big endian or vice versa, though.

[Ken Moore 23]

My 15 year old DCS can handle either case, just set a bit and the words are swapped. The new system that management has chosen, is not as good as my 15 year old system as far as capability goes. However the new system looks good and they know how to sell to the people buying vs. the people using. So...I'm stuck with it. I'm using Modbus RTU via RS485.

[DanW 41]

I didn't know IEEE had a endianness standard, but more on that later. Apparently CLX uses big endian format, according to an app note, Using Logix5000 Controllers as Masters or Slaves on Modbus (page 4) http://literature.ro...ap129_-en-p.pdf With regard to IEEE preferences, according to Wikipedia, IEEE 754, Floating Point standard does not define endianess: "This means that floating point data written on one machine may not be readable on another, and this is the case even if both use IEEE 754 floating point arithmetic since the endianness of the memory representation is not part of the IEEE specification." And a quick scan through the formatting mentioned here http://www.tacc.utex...g1/pg3formt.htm shows references to both IEEE little endian and IEEE big endian. Here's the lay-out of the two common ones, Big Endian and Little Endian byte swapped. I found a note from early 2000's that says Modicon and Wonderware use Little Endian Byte swapped as their standard format. But it matters not, if you know what you're starting with and what you need on the other end, right?

[Peter Nachtwey 15]

Prosoft should have provided you with this answer instead of passing the buck to AB. There is no excuse for the answer you got.