1794-IE8 RAW COUNTS FLUCTUATING

[dugmick 0]

Good morning all - 

I have a Flex I/O 1794-IE8 that the raw counts are varying by about 30 with no change in input.  It is used as a 4-20 mA input, reading the output from a signal conditioner from a LVDT.  This is the only input used in this module.  The stackup for the Flex system is a 5434 CPU (with Ethernet and DeviceNet cards), 2 - 1794-IB16 input modules, 1 - 1794-OB16 output module, 1 - 1794-OW8 relay module and the 1794-IE8 analog input at the end of the line.  

I have checked the wiring to ensure it was as specified in the installation instructions.  The LVDT and signal conditioner was changed to see if the problem moved, but it didn't.  Input 1 is used on the IE8, other inputs were also tried, with no change in results.  I tried connecting the output of the signal conditioner to a spare MicroLogix 1400 and a 1762-IF4 analog card in a further attempt to isolate the problem.  The MicroLogix raw count reading was rock steady, while the IE8 was still fluctuating.  I have seen some references online to other instances of issues similar to this, is it a known issue with these modules?  It appears that the IE8 is the problem, but I hate to pay to replace it with another if this is a known issue with the IE8 cards.

As another observation, the raw counts coming from the MicroLogix were substantially different from the Flex, like 10258 Flex vs 14541 Micro.  I had considered reading the conditioner with the Micro and messaging the count to the Flex.  The Flex is communicating with a drive via DeviceNet, and I would rather not tackle exchanging the Flex at this time.  Thus the thought of using the Micro to send the Flex a stable count value.  What would cause the raw counts to be so much different?

Thanks for reading...

[PLCMentor.com 60]

My first inclination is that the small fluctuations you are seeing are due to noise.  That is a very small variation considering that your full scale is 30840 counts.  A change of 30 is less than .1%.  You hooked it to a 1762-IF4 which has around 25000 counts for the same range.  It also has filtering available which may be eliminating any noise.  Cant remember if the flex IE8 has such.  Also did you have the same cable run for both connections?  If it is noise you may not see it if you are trying the micro on a test stand.  Finally the raw data ranges of the two cards are very different so it doesnt surprise me that you have two different values.  The flex input of 10258 translates to 9.3 ma and the raw of 14541 on the IF4 also correlates to 9.3 ma.

[dugmick 0]

56 minutes ago, PLCMentor.com said:

My first inclination is that the small fluctuations you are seeing are due to noise.  That is a very small variation considering that your full scale is 30840 counts.  A change of 30 is less than .1%.  You hooked it to a 1762-IF4 which has around 25000 counts for the same range.  It also has filtering available which may be eliminating any noise.  Cant remember if the flex IE8 has such.  Also did you have the same cable run for both connections?  If it is noise you may not see it if you are trying the micro on a test stand.  Finally the raw data ranges of the two cards are very different so it doesnt surprise me that you have two different values.  The flex input of 10258 translates to 9.3 ma and the raw of 14541 on the IF4 also correlates to 9.3 ma.

Thanks for the reply.  The cable run is the same for both conditions.  The first test I moved the wire from the Flex to the Micro, second test I connected the wire to both inputs in series, same result both times.  The problem the fluctuations is causing is that we are trying to measure to 1/100 of a millimeter, and the variation is absorbing most of the part tolerance.  The full scale count difference explains perfectly the difference in readings.  

[panic mode 246]

what LVDT and signal conditioner are used? what is the measuring range? 

[kaiser_will 65]

Is this a new installation or is this a mature application with a new problem?  

If this is a new installation, do you have access to a power analyzer (such as a Dranetz)?  A power analyzer is a critical tool when assessing the condition of your ground signals.

What is the LVDT units range, probably something 0-32,766 counts over the length.  What is the analog input module scaled for range?  I am trying to flush out if you are experiencing accumulative error from noise or scaling.

[dugmick 0]

7 hours ago, panic mode said:

what LVDT and signal conditioner are used? what is the measuring range? 

Thanks for the reply.  I was mistaken on the device, it is not a LVDT but a linear position transducer, a linear potentiometer of sorts.  The signal conditioner is a Novatechnik TRS50A502, and the signal conditioner is also a Novatechnik MUP 100-4.  I am not really sure what you mean about the measuring range, the transducer has about a 2.5" travel.

[dugmick 0]

2 hours ago, kaiser_will said:

Is this a new installation or is this a mature application with a new problem?  

If this is a new installation, do you have access to a power analyzer (such as a Dranetz)?  A power analyzer is a critical tool when assessing the condition of your ground signals.

What is the LVDT units range, probably something 0-32,766 counts over the length.  What is the analog input module scaled for range?  I am trying to flush out if you are experiencing accumulative error from noise or scaling.

Thanks for the reply.  This is a mature application, but I am not sure if it a new problem or an old problem that someone decided just now needed addressed.  I do not know if there is a power analyzer here, I haven't been in this job long enough to know what kinds of tools are available.  As far as the LVDT, is is actually a linear position transducer (linear potentiometer), 0-5k ohms, feeding into a signal conditioner to provide a 4-20 mA signal to the PLC input.  The PLC input is scaled 0-30840.

[panic mode 246]

"about" 2.5 inch travel?

soooo.... you did not even read the datasheet? 

 

 

 

 

[kaiser_will 65]

So the signal source is a linear pot, 0-5K ohms that is converted into a 4-20mA signal into the PLC.  

Look at every piece of the pie and find out their accuracy and resolution.  Accurate position feedback/measurement devices should never have more than 10% error of the desired measurement accuracy. 

For example, if the process accuracy is to be 0.001" (1 mil), an effective measurement device would have no more than 10% or 0.0001" (0.1 mil) total error.

Where I am going with this is...if you are made aware now that a mature process is observed to have significantly more accumulative error than your process spec, this is a good time to re-think the measurement device.

The source of the inaccuracy, in addition to the difference between raw counts measured in the Flex rack and the Micro PLC, could be due to bit resolution.  If you have a 16-bit resolution analog source and run it into a 12-bit analog input, for instance, you will not have a one-to-one measurement.