Low accuracy at low voltage

[charlesG123]

Hello,
I have a 1734-AENT POINT I/O with 1734-IE2V in it and I have noticed that the accuracy of the value is low at low voltage. For example when I send 874mV, the PLC reads anywhere between 840 and 910 and it keeps changing even tho the 874mV I'm sending is stable. that is almost a 5% change in the value. If i send a higher voltage like for example 1.75V, the plc reads 1745 to 1755 which is about 1% precision which is much better. All of my IE2V does the same, I was wondering if this was a normal behavior or if something was wrong with my POINT I/O?

Thanks.

 

[parky]

Check your earthing, had this once where the cable (screened) the screen was connected both ends & also an earth bonding.
Removed one end of the screen & it improved the value swing,, it's definitely worth checking

 

[charlesG123]

Check your earthing, had this once where the cable (screened) the screen was connected both ends & also an earth bonding.
Removed one end of the screen & it improved the value swing,, it's definitely worth checking

I'll check that out thanks for the suggestion!

 

[danw]

Some form of noise. Some AI's have a setting to synch the sampling with the phase angle of the AC supply line so that the noise level is the same at each sampling. The synch setting is typically a choice of 50Hz or 60Hz.

 

[charlesG123]

Some form of noise. Some AI's have a setting to synch the sampling with the phase angle of the AC supply line so that the noise level is the same at each sampling. The synch setting is typically a choice of 50Hz or 60Hz.

Is it the Notch Filter? It is set on 60Hz, also what does AI stand for?

 

[mylespetro]

Is it the Notch Filter? It is set on 60Hz, also what does AI stand for?

I assume he means Analog Input

 

[plvlce]

also what does AI stand for?

Analog Input

 

[Cheeseface]

That module's resolution is 15 bits + sign over -10V to 10V. That means you get 65536 counts for 20 volts. This corresponds to roughly 3 counts per mV.

I don't think the fluctuation you're seeing is unusual for that module, the range 840-910 corresponds to about 21mV or 0.1% of the input range of the device.

All this is to say you should consider a different input device, although the 1734-IT2C only has an input range of +/- 70mV

 

[I_Automation]

When the voltage is very low (frequently under 1.5V) it may not be enough to overcome any corrosion or loose connection at terminals and the OP readings stated are exactly what happens.

The benefit to using mA vs voltage analog signals is the mA signal is always over 12VDC, and up to 24V.
Voltage systems can be setup as 1-5 or 2-10 to keep a minimum voltage applied.
0-60mV systems require very good connections at terminals and a straight run with minimal additional connections on terminal strips - or signal conditioners right at the device to use 10V or 20mA.

I have standardized on 4-20mA signals for all PLC analog IO and signal conditioners as needed, which in the long run is less items to stock and spec, easier CAD drawing, and a supply of PR Electronics 4114 Universal Transmitters has almost every possible need covered.

 

[charlesG123]

When the voltage is very low (frequently under 1.5V) it may not be enough to overcome any corrosion or loose connection at terminals and the OP readings stated are exactly what happens.

The benefit to using mA vs voltage analog signals is the mA signal is always over 12VDC, and up to 24V.
Voltage systems can be setup as 1-5 or 2-10 to keep a minimum voltage applied.
0-60mV systems require very good connections at terminals and a straight run with minimal additional connections on terminal strips - or signal conditioners right at the device to use 10V or 20mA.

I have standardized on 4-20mA signals for all PLC analog IO and signal conditioners as needed, which in the long run is less items to stock and spec, easier CAD drawing, and a supply of PR Electronics 4114 Universal Transmitters has almost every possible need covered.

Thanks, not being able to overcome the corrosion or loose connection makes sense. 4-20mA would definitely be the best solution if we ever get the opportunity to update our setup. Thanks for the insight!

 

[robertmee]

Hello,
I have a 1734-AENT POINT I/O with 1734-IE2V in it and I have noticed that the accuracy of the value is low at low voltage. For example when I send 874mV, the PLC reads anywhere between 840 and 910 and it keeps changing even tho the 874mV I'm sending is stable. that is almost a 5% change in the value. If i send a higher voltage like for example 1.75V, the plc reads 1745 to 1755 which is about 1% precision which is much better. All of my IE2V does the same, I was wondering if this was a normal behavior or if something was wrong with my POINT I/O?

Thanks.

As @Cheeseface stated, the resolution of any signal A/D is over the measurement range. The way you are calculating resolution isn't really applicable. Think of it in terms of weigh scale. It may have 2000lb capacity, with .1% resolution (2lbs). If you put a 2000 lb weight on it and it reads 1998 lbs, the scale is doing it's job, and you have 99.9% accuracy. If you put a 20lb weight on it, and it reads 18lb, the scale is STILL doing it's job, but now your accuracy 90%, but the resolution of the scale never changed. The resolution doesn't change as you narrow the field of measurement. But as you witnessed, the accuracy certainly does.

 

[drbitboy]

I think the original question was why range of noise is the equivalent of 70mV at voltages near zero but only 10mV at voltages far from zero.

@I_Automation's answer sounds interesting, but I don't understand the mechanism: higher voltages somehow lessen the noise induced by a poor connection?

 

[I_Automation]

I think the original question was why range of noise is the equivalent of 70mV at voltages near zero but only 10mV at voltages far from zero.

@I_Automation's answer sounds interesting, but I don't understand the mechanism: higher voltages somehow lessen the noise induced by a poor connection?

It's not really noise.

It's the inability of a low voltage to conduct through a slightly bad connection.

For a 0.1 to 0.6 volt signal it doesn't take much corrosion or a noticeably loose screw terminal to prohibit the low voltage. A little layer of corrosion not visible or a terminal tight enough to hold the wire in but not torqued to a few inch-pounds to make a good connection for current flow can easily cause a problem.


Unless the wire is very corroded a voltage over 1.5V almost always flows correctly, depending on the severity of bad connections the threshold may be as low as 0.3V or up to 1.4V. EDIT: Plus it's not a consistent blockage where a 1.2V signal will come through repeatedly at say 0.7, so that fluctuation is what is being called "noise"

There is a similar issue with DC motors run at lower voltages - cutting out or slower than should be when under 10V, but from 10.1 to full 90V runs at the speed it should. Had one customer with this that wanted the motor control replaced - I cleaned all the exposed wires with a wire brush, used Anti-Ox and torqued down all terminals from the motor control to the motor.

 

[drbitboy]

Thanks @I_Automation!

So you are saying the resistance of the terminal changes on a short timescale, when the voltage signal is under some threshold, typically 1.5V or somewhere around there, and that resistance is steady when the voltage is over that threshold?

Yes, when I wrote noise I meant temporal variations in the final visible value of whatever memory bits (INT, REAL, etc.) in the PLC are used to model the measurement; not "real" noise i.e. from the transducer or process (shot noise, instrument noise, process noise, etc.).

 

[I_Automation]

Thanks @I_Automation!

So you are saying the resistance of the terminal changes on a short timescale, when the voltage signal is under some threshold, typically 1.5V or somewhere around there, and that resistance is steady when the voltage is over that threshold?

Instead of resistance think of it more like the voltage drop across a diode that once flowing good drops from 1.4 to 0.001 volts as long as the threshold voltage is exceeded. Or the gate of a transistor.