Rules to be respected to connect a sensor to multiple PLCs/reader

[bgauv]

Hello everybody,

I work in the field of industrial computing and sensors. However, I am not an automation specialist. My expertise is more on the side of the Industrial Internet of Things.

I often need to connect sensor readers to sensors that are already connected to PLCs. Since these PLCs are used by my customers to control their processes, it is not an option for my sensor readers to cause interference with the signals read by the PLCs.

Could you please give me the rules to respect according to different scenarios (4-20mA, voltage signal, pulse counters, etc.) to be able to install my sensor readers correctly. For example, someone has already told me that there is an impedance issue to be respected.

Thank you in advance

Bert

 

[Steve Bailey]

Yours is not the way I'd approach the task of gathering data from an existing system. The potential for breaking something that is working properly is large. In addition, there is the old rule of thumb, "you touch it, you own it". In other words, the first time the customer has a problem with something you've tapped into, they'll blame the problem on you.
Why wire the sensors into your system? Why not purchase a license for an OPC server and read the sensor data that's already in the PLC?

Ohm's law and the spec sheets for your sensor readers and the devices you're trying tap into are the tools you need to figure out whether or not you can get away with it. Be sure to leave yourself a generous margin of safety.

 

[califflash]

Not trying to put down your job but I agree with Steve. This is definitely not the way to go on industrial processes. If the data already exists in the PLC it’s a trivial matter to get an OPC server to push this data up to a database or make it available through a variety of interfaces like restful or MQTT.

As was mentioned, you run the risk of breaking a process connecting this stuff. Also keep in mind that if you get this hooked up and working, then the device fails in a year, will the process go down due to that? Reliability is in my top 5 must have requirements of anything I put in the field.

 

[cardosocea]

+1 for Steve's opinion and solution.

However, if you must, there are devices in the market called splitters that would do what you want in a safe manner.
Not advertising for them, this is the first link in Google for Signal Splitter Isolator.

https://www.amelec-uk.com/products/Process-4%2d20mA-Signal-Splitter-Isolator-%2d-ADG231.html

This will solve your problem of reading existing sensors without affecting the system. You do, however, need to take into consideration what side is powering the loop when wiring something like this, but should be trivial.

Lastly, you also need to understand what signal it is and what does it impact... chances are that SIL rated loops may not have this type of device in them as it may drop their rating. The owner of the system you want to monitor should be able to tell you this unless it's an OEM machine.

Also, having two places where the sensor must be scaled is not very good... if it's read from the PLC it's already in Engineering units and much simpler to deal with.

 

[bgauv]

Thank you all for your advice and guidance.
So I will connect via Ethernet network to PLCs rather than by wire.

Thanks

Bert

 

[parky]

Sometimes because of OEM locks or lack of source code, or communications ports for the machine it may not be easy to connect to the existing system.
If that is the case then splitters may be the answer or even double up on the sensors & wire them separately, however, if connection is possible to existing sensors then you should be aware of potentials (power & signals) between the equipment and your system. In most cases where external potentials that are not isolated within the control panel then external interfaces should be wired in different colours and mounted in a separate enclosure with a label stating that isolation of equipment within this enclosure is elsewhere preferably with the location.

 

[danw]

I missed this being away for the weekend.

Amplifying Parky's warning about potential potentials and in case that you find you're road blocked from using digital networking connections and you have to do the connnection-to-the-sensors, here's some pointers:

One starts with the general rule that electronic measurements are voltage measurements.

• DC current is measured by running it through a known resistance and measuring the voltage drop (IR drop in Ohms law lingo). That’s what happens at the analog input connected to a 4-20mA signal.

• Varying resistances are measured by pumping a known current through the resistance and measuring the voltage drop.

• Hence, It is not possible to piggyback onto either an RTD or thermistor temperature sensor due to the nature of the excitation current needed to get a usable measurement.

When dealing with high level DC voltage signals, like 0-1V, 1-5 V, 0-5V, 0-10V, it can be feasible to piggy back by paralleling across with an AI with a 2nd reasonably high impedance device so as to not load source. Most voltage inputs are 1 Mohm so that’s not really an issue.

However, what is a known potential problem (that crops up here on this forum all the time) is the 2nd monitoring system not having the same ground potential as the base monitoring system which creates ground loop issues which creates a ground loop. A ground loop of mild severity adds an artificial offset to the measured value. A moderate severity ground loop drives the measured value off-scale. A severe ground loop burns out analog inputs.

Isolated analog inputs eliminate ground loops, but at great cost. I'm guessing IoT AI's will not be (relatively expensive) isolated analog inputs because all the IoT emphasis is on the other end - the glories of "all that data on your cell phone". Can't have expensive hardware scaring off the prospects.

4-20mA signals can be tapped into
• By measuring the voltage (in parallel) across the base system AI’s dropping/shunt resistor
• By adding an additional dropping/shunt resistor to the loop and measuring the voltage drop.
• Adding a (relatively expensive) splitter module, one 4-20 in, two 4-20’s out. Splitters almost always off isolation from input to output, and usually to the power supply as well.

The first two techniques involve the risk of ground loops. Same thing applies as with the voltage monitoring - isolation is a solution but is expensive.

All that said, I think you’re best off getting data from existing systems via OPC.

One of the problems I've run into is getting reliable scaling and engineering units for any signal you piggyback onto. It's amazing how different reality can be from what someone 'thinks' it is. If the data points are already tagged, then OPC UA compliant devices can transfer ID data (eng. units, span) along with the data values themselves.