IO module replacements: Look at physical location, measurements, processing needs

I/O wiring can be among the most challenging of upgrades. When replacing input/output connections, what are key considerations? Brett Burger, senior product manager for embedded I/O with National Instruments, offered answers.

By Brett Burger October 19, 2011

Today’s I/O systems are a mix of single-board computers, signal conditioning circuitry, and rugged mechanicals. When it comes time to replace one, there are many factors to consider. In all of these discussions, flexibility is key. Flexibility extends the amount of time before an upgrade is needed. When buying or specifying I/O systems, consider:

1. Physical location. I/O systems are usually designed around criteria for specific deployments. Where do you want to mount your system?

  • Inside a panel or box, on a wall, or on a test-bench? Make sure to look at mounting options.
  • Indoor or outdoor? Weather or IP (ingress protection) ratings? It’s often easier and more cost-effective to use IP/NEMA rated enclosures for systems rather than to find “weatherproof” I/O modules.
  • Near industrial equipment? Electromagnetic radiation (EMI) from motors, lighting, and power supplies can degrade signal performance. It’s recommended to mount I/O modules closer to sensors to help with shorter signal wire runs, use shielded cables, and look for modules with isolation.
  • What power source is available? This isn’t weighed too heavily in the decision, but it’s more convenient to drive multiple pieces of equipment from one power source. Additionally, it’s important to check the input ranges of the measurement and control systems under observation.
  • How do you want to connect to your I/O? Many connectivity solutions are available. Some will mount directly to the module, others cable out to a terminal block, and some make it possible to easily create a custom cable for a system. Check the wire gage. I/O systems rarely need large gage wire since they don’t move a lot of current, but if cables are already in place, then it can be expensive to rewire a plant, floor, or machine.
  • What standards are important to you? These might include UL, CE, C-tick, RoHS, and NIST calibration? Do you need conformal coating?

These physical features are some of the first ones considered. Mechanical design can affect accuracy and could be the difference between a 1% or a 10% failure rate. Not all systems are designed for all environments. For example, the NI CompactRIO embedded I/O system was designed for use in rugged environments, is tested to 50g/5g operational shock/vibration, and maintains maximum accuracy specifications over a -40 C to 70 C temperature range.

2. I/O needs: There are many modular I/O systems (PACs/PLCs) on the market. It’s important to consider the entire catalog of I/O available to make a sound decision.

What types of measurements or applications do you need now and might you need in the future?

  • For industrial needs, make sure there is a good selection of “industrial-type” I/O such as 0-5 V, 0-10 V, 4-20 mA, 24 V DIO, and quad/tach/counter.
  • For “test” type measurements, look for I/O that can measure from a broad array of sensors such as thermocouple, RTD, strain gage, load/pressure/torque cells, and accelerometers.

What channel counts are covered by the modules and chassis?

  • For higher channel count systems, look for density (more channels per module) as those are going to be more cost-effective.
  • For system needs that have a lower channel count but a higher mix of channel types, look for systems that offer fewer channels/module, or “universal” modules for better granularity.

Though there is no such thing as a “future-proof” system, a large selection of I/O modules is advantageous for system growth. It’s a good practice to determine a backup plan in case I/O is needed that is not available off-the-shelf from a system vendor as custom circuitry is not always easy to integrate with off-the-shelf platforms. Some systems try to combat this problem by publishing development kits for custom module development. The NI CompactRIO family of hardware, for instance, has more than 50 I/O modules available as off-the-shelf products and a module development kit (MDK) for custom, third-party module creation. These custom modules fit in the same mechanicals and chassis as the off-the-shelf modules providing better integration for custom circuitry or connectivity.

3. Communication and data accessibility: The consumer “electronics and information” space has driven new technology around how data is disseminated.

  • Who is going to use this data? Are they sitting at a computer at their desk? Are they walking around with an iOS or Android device? Does the system run headless?
  • To what other devices does your I/O system need to communicate? OPC Server connectivity? Busses/protocols, such as serial, Modbus, Ethernet (TCP/IP), Foundation Fieldbus, DeviceNet, CAN, EtherCAT? More flexible systems will have multiple options.
  • Does there need to be local data storage? FTP site? Local disc?

The importance of communication protocols and data accessibility scales with the number of different systems you have in use. Much like I/O, multiple options for communication and data storage will make it a lot easier to integrate a new, replacement system with an existing legacy system. This issue can be as much a software challenge as it is a hardware challenge due to the growth of web technology and smart devices. For example, LabVIEW running on CompactRIO can communicate via MODBUS (DB9) to existing equipment and publish data to the web, which helps engineers design control and monitoring systems that increase capabilities without requiring an entire retrofit.

4. I/O data processing needs: Data processing can be important even if you only care about “dumb” I/O measurement transmission.

  • Is your I/O system going to be used for control? Many older systems are configured with central intelligence. Sensor data from the whole network is communicated back to a central processing unit. Setting up a new system with distributed intelligence can increase decision speed and decrease data communication bandwidth. Processing on PACs today often takes place in one of two locations: real-time host side of the system (standard floating point processor), and dedicated, programmable FPGA for low latency and high-speed closed-loop control rates (100 kHz).
  • If your I/O system is going to be used for simple data communication, do you have a need for “data preprocessing” to reduce data sent over the network? 
    • Calculate RMS, peak, and frequency values from a waveform locally rather than transmit full waveform data over the network. 
    • Communicate on value change only rather than broadcasting the same data multiple times. 
    • Oversample and average to improve accuracy on a dc signal. This can help to reduce data and signal noise. 
    • Scale data at the module location and transmit data in real engineering units rather than raw values. This will not reduce data traffic, but it will improve ease of use.

Processing capability extends beyond how many cores are on the processor or how many lookup tables (LUTs) are on the FPGA. The development environment for an I/O system can have a large impact on capabilities as well as the deployment time. Ladder logic was designed around digital/relay style logic and is not well suited to more advanced control algorithms such as model free adaptive control. Many standard text-based languages can program at a very low level, but as a result can increase development time. NI CompactRIO and LabVIEW were designed to help program simple or advanced control schemes using graphical code that can be deployed to processor based real-time targets or FPGAs.

As you can see, when it comes to replacing modular systems, there are several considerations well beyond the modules themselves.

– Brett Burger is senior product manager for embedded I/O, National Instruments. Edited by Mark T. Hoske, CFE Media, Control Engineering,

Also see Channels for:

System Integration

Machine Control