By John Rezabek, Contributing Editor
In a welcome, end-user inspired solution for distributed I/O, Emerson Process Management has introduced a unique way to get signals into its DeltaV DCS—Electronic Marshalling. Unlike traditional remote I/O solutions, Electronic Marshalling collapses knife edge disconnects and I/O conditioning to a small module that publishes the signal to an "I/O cloud" accessible by controllers. Called Characterization Modules (CHARMs), they are modular on a point-by-point basis. Instead of adding I/O with four-, eight-, 16- or 32-point modules of a given species, you snap an individual CHARM onto its bus rail for each successive point you need.
If you've ever brought in a multi-pair cable and discovered that 20% of the signals were four-wire, while the rest were two-wire, you now have a new and elegant solution. Likewise for discretes that originate from both low- and high-voltage circuits, the flexibility is there to accommodate diversity as well as oversights, errors or wrong guesses from detailed design.
I expect some of those same engineers cringe when I lump the CHARM solution in with remote I/O and multiplexers, and, to a degree, that reflex is justified. Their analyses show that CHARMs are not ordinary muxes—they have a predicted mean time to fail (MTTF) that equals or even exceeds that predicted for a traditional solution using redundant I/O cards.
CHARMs also arguably reduce the mean time to repair (MTTR) because one addresses single-point faults by replacing a single CHARM. CHARMs are designed to preserve single-channel integrity, so only one loop is affected by a fault, whether it's in the CHARM, in the field wiring, or in the field device itself.
So why would I lump CHARMs in with their lesser brethren, the multiplexers?
Just as with remote I/O, one uses redundant, twisted-pair cables to bring the mux into the system, as well as redundant I/O processors in-house to transfer data to the host. But, this redundancy can be compromised by sharing the same routing, conduit and cable tray. If the data coming in from remote I/O or mux is "nice-to-know," "indicate only," then worries about a common-mode failure are less stressful. But, if you're doing real process control through the mux, the effort to design and install geographically separate paths might be worth it.
The other issue is field-sourced power. CHARM's I/O runs on 24 VDC, usually impractical to run hundreds of feet into the field for higher-current appliances. A typical factory-certified, pre-manufactured JB with 48 low-voltage CHARM capacity requires a 5-amp, 24-VDC power supply. One hundred feet of 10 AWG copper has a nominal installed resistance of 0.1 Ohm, so even this larger-gauge wire could drop 2.5 volts of DC power if run 500 feet, making the comfortable N+1 redundant "bulk DC" solutions we use for conventional point-to-point and fieldbus less applicable.
There are industrial-strength power supplies that are good for Division 2/Zone 2, and CHARM I/O has some in-built diagnostics designed to alert users when one component of a redundant pair is failing, but they require AC line power. You can feed such redundant pairs from opposite sides of the AC bus, so that only a total plant power outage would crash the remote I/O, but many process plants want their DCS I/O to ride through brown-outs, and remain serviceable, while the plant crashes during total power outages. If this is a requirement, one has to figure how to get UPS power to the field JBs, and such circuits have distance limitations, too. A 20-amp circuit breaker protecting a UPS circuit may cease to protect against shorts when the circuits are so long a fault looks like a load instead of a short circuit. You can install UPS distribution panels in the plant, but this involves additional significant design effort and lifecycle management.
Once we've addressed our concerns about field-sourced power, the dream of a rack room uncluttered by massive marshalling panels and scatter wire can become the "new normal."