Loss of Floating point resolution in MSG

I'm messaging a Floating point from a ControlLogix to a Micrologix 1400, and for some reason the full float value from the CL doesn't come make it to the destination file.

The ControlLogix real tag has a value of 1.000035e-08, but when its writing the float on the Micro, it shows up as 1.0e-8 . Its not a huge deal, but I was surprised to see this. I suspect its something with the way Rockwell handles their messaging statements, or ?

Anyone else ever notice this ? Is there a work-around to get that point over to my Micro ?

All A-B controllers use the same Floating Point representation; 32-bit IEEE 754 Single-Precision Floating Point. The data is sent as four bytes so I'm confident it's really getting from the ControlLogix to the MicroLogix.

RSLogix 500 and RSLogix 5000 take different approaches to displaying floating point data.

RSLogix 5000 attempts to show the "exact value", while I think that RSLogix 500 has a number of zeros after which it won't display digits.

The value is exactly the same, and will have the same result if you use it in a math operation or show it in an HMI.

Ken, thanks. I figured you'd know the answer. I noticed that even when I pointed my HMI to the CL the data was represented the same way. Thanks again.

I'm not sure if this proves my point or disproves it: FactoryTalk View ME 6.0 shows that value from both the SLC and Logix (emulators) as 1.0000000000.

I guess that FTView has yet another way of displaying very small fractions.


If the value 1.00000035 is copied from a FLOAT into two INTs or a DINT, it shows as 0x3F800003 in both the SLC and the Logix, verifying that the two editors are just displaying the same value differently.

Those values out to the 5th and 6th decimal place are meaningless.

Here's several accuracy statements for Agilent Pirani style vacuum sensors.







The accuracy statements are in the whole percentages of reading (2.5% to 50%, 30%, 15%), which means the 10's and 100's of thousandths values in the 5th and 6th decimal places number is pure noise.

A 2 digit truncated mantissa value is all the sensors are capable of, so the value the Micrologix is using is just fine.

Accuracy wasn't my primary concern, it's the fact that the msg'd value didn't appear to contain exactly the same data. I've never seen this, so I asked. I have multiple display points, Wonderware and a PV600 and panel mounted meters, etc. I'd prefer that they all display as to close to the same as possible.

What happens if you msg the other way? From the MLX to the CLX

Haven't tried that. I'll be out at the client site again Tuesday, and was curious to see this myself. I'll let you know.