A few weeks ago, I was sitting in the local Applebee's after a long and demanding day at the conclusion of a week-long outage. The barmaid made a "bitter-beer face" when I requested an O'Doul's. I was expecting to be back in the plant before the night was out. Between bites of dinner, the phone rang. It was the plant, of course. A variable-speed drive for a crucial 500 hp cooling tower fan had been replaced. Everything checked out, but the "setpoint" sent via 4-20 mA from the control house wasn't modulating. The drive was an older model when it was first installed more than a dozen years ago, and none of us were proficient enough in German to read the manual. It had made a trip to the vendor to repair some mysterious problem, where they pretty much just exchanged all the circuit boards. So why wouldn't the identical drive take a current signal like the old one did? It must be the DCS, right?
Current loops are in the physical layer of many forms of communication, from land-line telephones to Gigabit Ethernet. The type we still use for simple analog signals, 4-20 mA, has been around for decades, and was established by ISA's SP-50 standards committee sometime after the middle of the last century. It was preceded by 10-50 mA, which the early electronic transmitters needed to power force-balance transducers. But in both instances, current was desired for electronic signal transmission because it could be relied on to be precisely the same on both ends of an undamaged pair of wires, whether that pair was a few feet long or a few miles. A precision 250 Ω or 100 Ω resistor on the receiving end creates 1-5 V to represent 0-100% of span for analog or digital controllers. The analog I/O of virtually all modern devices uses such a "dropping resistor" for converting current signals to voltage.
Because of its pervasiveness, devices abound for converting all manner of analog measurements and signals to 4-20 mA. One of the most common varieties is for temperature measurements. Thermocouples develop a millivolt signal in proportion to temperature, and this small voltage can be read directly at a distance, provided matching thermocouple alloy wire is used. The extremely low signal is also very vulnerable to noise. Running the wrong extension wire can result in undetected temperature errors, which can cause machines and processes to run off-specification or unsafely. Today, devices to convert the minuscule thermocouple signal to the more robust 4-20 mA have become very affordable and small enough to fit in the termination head of the sensor. These little pucks include the necessary linearization for thermocouple signals and cold-junction compensation, ensuring the signal is corrected for the ambient temperature at the converter's terminals. There are also DIN-rail versions that provide additional features, such as NAMUR upscale burnout (forces the signal to full scale when the thermocouple fails) and broken-wire detection.
Similar accommodations can be made for resistance temperature devices (RTDs), which are preferred for accuracy and stability (thermocouples are known to drift with age) when the measured temperatures are below 1500 °F and the ruggedness and fast response of a thermocouple isn't critical. Acromag makes a line of head-mountable hockey puck converters that are configured via USB. Others, like products from E+H, Rosemount, and Moore Industries, are configured using HART, whose configurators are familiar tools for many veteran instrument specialists.
What about our troubled VFD? The circuit in question also used a current-to-current converter that's often needed for isolating separately powered systems. Tests eventually revealed that the converter was fine; it was just trying to drive current through very high impedance. The "refurbished" card in the VFD had been returned with all its jumpers in the default position—the wrong position for a current signal.
Jumper-configurable boards are becoming rare, but we work in a time with a befuddling assortment of legacy and novel communications. We now can specify VFDs to accept digital signals via Profibus, DeviceNet or Ethernet. Converters, too, can be obtained to interface directly to a variety of the same digital buses, or even wireless. Will the digital protocols become simpler to troubleshoot than our 4-20 loop? Hopefully someone's working on a converter "app" that will spare our successors a "conversion experience" like mine.