unstable PLC analog input

i have a GE fanuc 23 point versa max with 2 inbuilt analogue inputs,
the problem i am facing is an unstable reading, i am using a 4-20 mA danfoss MBS 3100 pressure transducer.

now i have tried all the shielding and grounding that is possble and all the possible combinations, i have also tried to connecting a capacitor and inductor to help me with the unstablity, but this was all in vain.
my values keep fluctuating between, 0,8,16 out of 32767 at atmospheric pressure. the fluctations are also present at higher pressure, they remain in the range of 0-24
now after 2 days of fiddling around with the PLC, i have found half a solution.
there are 2 analgue channels in this PLC, and 3 terminals for each channel(IN+,IN-, INJP)the IN+ is connected to INJP by a jumper when taking current input.
the fluctations stopped and the transducer was working perfectly when i connected the INJP1 to INJP2(second analogue channel(this channel was not in use)), sadly i cant do this as ill be connecting two transducers to both my analogue inputs,
so my question is what do i do, how do solve this?
and why did the fluctations stop when i connected the INJP1 terminal to the INJP2 via a jumper cable?
please help, deperately looking for any kind of help

Have you checked the shielding and cabling of the analog signal?

In reality though, I don't think varying by 16 to 24 points on a 16bit analog input is that bad. You can't really expect things to be that perfect. (at least, in my experience).

As far as the reason why it worked when you connected INJP1 to INJP2, I cannot say. Do you have a manual for the analog input module you are using? What does it say about the INJP1 and INJP2 terminals?

thnks for replying

http://support.ge-ip.com/support/re...tent/live/DOCUMENT/0/DO458/en_US/GFK1645J.pdf

heres a link to the manual, there are some diagrams in chapter 15 of the analogue input channel.
its a 12 bit input channel, and im using the inbuilt analogue channels with the processor i have no external module or card.
the fluctuations are not because of the shielding or grounding, i spent 2 whole days experimenting with them, its due to some internal thing, which i cannot comprehend due to lack of electronics background
my requirement is such that i require it to be as stable as possible.

That's a 12-bit analog conversion in that model PLC. The least significant bit in the analog word is bit 04. That's why you're seeing steps of 8 counts in the %AI word. A value of 8 is equal to 1 part in 4096, or 0.00000488 amps.

my requirement is such that i require it to be as stable as possible

Click to expand...

You're there.

One way I know to minimize this is to get rid of the LSB before you use the input. One way to do this is to shift the %AI one bit. of course then you have an 11 bit analog so plan accordingly.
Another way I have used is with a rolling average of the last 5 updates (or changes) but this puts your Input at a lag to the actual.
You need to decide the proper method based on the needs of your application. I cannot tell what is best for you.

...or just keep only the MSB

LOL, "keep the MSB". That has some real merit.

The pressure transmitter in this case is a 0.5% accuracy device - 1 part in 200. (That's all combined error)

A 12 bit A/D is ~ 1 part in 4,000

So the A/D is 20 times more resolute than the inherent accuracy from the transmitter.

He could throw away 4 LSB's (the 1/4000th, the 1/2000th, the 1/1000th, the 1/500th) and just use the remaining remaining 8 bits at 1/250 resolution. (Yeah, I know. I'm rounding the conversions)

Dan

This thread (Or should I say this project?) is a good lesson in the need to know your system requirements and select your hardware accordingly. When a good engineer from a different field jumps into the automation world there is a lot of up-front study that needs to be done. And do not listen to sales people, do the research.