M's, N's, & F's in AB plc prog

Good day, all. I have a devicenet system (slc 5/05) to which I've connected a Seimens flow meter. Despite what the (seimens) dnet interface manual says, the upper/lower halves of the flow rate data words as presented to the network require swapping (no big deal, but the pdf is incorrect). My solution, which I have used previously in exactly the same way without incedent, is to move the two "M" words as they appear from the dnet, into two contiguous N registers (eg: M1:1.0 into N7:1 and M1:1.1 into N7:0) then copy those swapped words into one F register, and all is well. The pdf shows the representation of the value as output by the flow meter as: bits 0-22: mantissa; 23-30: exponent; bit 31: sign, but I know this to be end-over-end. Is it possible that what I'm doing can cause the N word to exceed its legal range? The plc has fallen over (minor error) a couple of times in the few days since I added the unit. The problem is intermittent and the guy who looked at it missed the error number before resetting it. Maybe I should just keep S:5/0 reset, or limit the values for the N words?

i don't see anything wrong with what you do. i had the same problem (on different plc) and just re-arranged bytes, interpreted result as float and end of story.

Trap it like this (see pic):

You can have a trap like that in several different places with a different "B" file bit for each place so that you can tell more specifically what math caused the fault.

Just make sure one of those places is the very last rung, so you unlatch S:5/0 before causing a major fault.

Paul

Thanks, panic, for reassuring me that I'm not going mad, and Okie for confirmation of the S5 bit.