Modicon 875-002 analog input roll over to 9999 at 3.999 mA

I have several Modicon 984s using 875-002 4-20mA input cards. The sensor is a pressure transducer. When the pressure is at 0 the input word from the card tends to roll over backwards to 9999 when the input current drops just below 4.0 mA, ie 3.999 mA reads 9999. This of course creates a bogus pressure reading. To prevent this Bubba calibrates the transducer with a slight zero offset, but that means that it is no longer accurate.

Has any one else ever encountered a similar issue and how did you resolve it?

Shameless bump.

Wow... you wouldn't think they'd allow the reading to roll over like that precisely to avoid the type of situation you are describing.

I've never used any schneider products, so this is a dumb idea for a work around: You could keep a running average of the last 20 or so 'valid' readings and asume what end of the spectrum they are from when you reach the roll-around point.

The assumption should be correct unless the system can rapidly depressurize or reach maximum pressure in less time than it takes to get your 20 valid readings and avg. them.

That or you set/unset bits when certain thresholds are hit, so you know what end of the spectrum you are on if you see readings that are in the rollover range.

Well, enough of my stupid ideas. If the 984 is still supported, you could always give engineering support a call and see what they recommend.

edit: this is all assuming that 9999 is also the maximum reading you can get. I have no idea how the analog inputs are scaled on this PLC.

Can you add logic to work around it? In effect, all negative numbers=0.
If your input reading is higher than 3.999 mA than zero it.
IF IN > 4000 THEN IN=0.
You could use 5000 or even 9000.

Gosh.

I dont use Modicon, but I often use 0-20 mA where the sensor is really 4-20 mA.
In S5 I would get "channel error" if the input signal dropped to 3.99 mA or less. Not as catastrophic as roll-over to 9999, but irritating nonetheless.
So I changed to 0-20 mA and scaled accordingly in the program. And I made my own alarm for "< 4 mA" which was much less sensitive.

Good to see your still getting on with the old 984 series!...but as for the roll over..i used the same cards and never had that issue..what version of exe are you using?..the only work around i can thing of is that the max value of those cards is 4095...if its above that then force a zero to the register...i will pull up the old program i had when we used the 984 series and see if we put in a routine for this kind of thing..but its not ringing a bell now...
LOL..Dont forget i still have 4 of those processors kicking around!\

Greetings Darren, yep, they are still kicking.

I'm not sure of the version - how would I determine that?
The current program uses the SENS instruction to detect if the first bit of the input word is set, which would be the case if the input word value was 9XXX. Its a kludge and I don't like it, but then I didn't write it, I just have to maintain it. I was hoping for a neat and tidy solution, some trick I didn't know about. It just seems ludicrous to me to design an input card that rolls over backwards for anything below 4 mA. The channel alarm shouldn't be a nuissance that kicks in at 3.9999 mA, a 2.5mA threshold would give you the same information without the headache.

Is the 9999 an overflow/underflow indication? That is, does it show up on under-and-over the 4-20 ma range? You might have to add an error check for that and set the value to zero when you see the 9999.

I can't remember what equipment I used that had that ...feature... It wasn't Modicon.

Alaric...went over some old prog's and nothing in there to prevent that..the only thing i can see is i was using 984E..if my devices put out 3ma i read a zero...
now the sens tells me you are sending info over a network..??

I looked at some old programs also. In some I had no mods to input, but not using 875-002. However one program with 875-002 card configured I had massaged input with logic. If input greater than 4095 then I forced to zero. So it looks like I was seeing same as original poster. However so long ago I can't actually remember the situtaion...but I did write the code so it must of been happening! Can't find any old modicon manuals for the various analog input cards...but it does indeed appear that underrange needs to be accomadated with some logic at least with a 875-002.

From my old Modicon days 0-4095 ..... 9999 = out of range. You are going to have to set an off set in the instrument for the processor to react correctly. Bump your instrument down in range a little to "fit" the PLC's range. You got to match up the 4-20 mA ranges on both ends. 0-100 % is what you want 0-4095 on the Modicaon scale. Sometimes you have to cheat.

darrenj said:

Alaric...went over some old prog's and nothing in there to prevent that..the only thing i can see is i was using 984E..if my devices put out 3ma i read a zero...
now the sens tells me you are sending info over a network..??

Click to expand...

forget it..just re read your post..

One other thought. That may two reg long as well. Man it been too many years. If the first reg is say 40001 look in 40002 and see what value is there. You may have to set up so it writes the value to 2 reg or 32 bits. Just a thought. the 99999 may be in the first and the true value in the next ... worth looking at.
I may have an old manual some where. i will dig

You were right the first time 0-4095 with 9999 indicating a range error, which occurs at 3.99mA. Offsetting the transducer is what we do now, but that causes other problems, thats what I'm trying to get away from. I can and do use logic to trap the over-range, I was just hoping that there was a trick so that it wasn't so sensitive to range errors when pressure was very near zero.

Ok .... dug out Old Modicon Book ... If you are doing a Math function on the value and a 9999 appears and overflow condtion has occured. To be sure check the middle node of the register if it is on you have an over flow condition. In this case the the math was not performed. The math answer was to big for one register to hold thus an overflow you will have to use two consective registers for the answer.

I hope this helps.
Dog