Industries / Industrial equipment and services / IoT and industrial automation / Process instrumentation
Signals and wiring
This work decides how an instrument's reading, and news of its own faults, reach the control system, and how that wiring is kept from igniting gas where the air can explode.
4
Milestones achieved, from 1913 to 2015.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1913
Sparks from a signal bell
On 14 October 1913 an explosion at the Universal Colliery in Senghenydd, Wales, killed 440 people, by the count of the Northern Mine Research Society. The mine signalled with electric bells on bare wires fed by six to nine dry cells of about 1.5 volts each, and sparks formed every time the wires were touched together to ring them. The inquiry under R. A. S. Redmayne, Chief Inspector of Mines, found that the only apparent means of ignition were those sparks or falling rocks, recalled that the Bedwas Colliery explosion of March 1912 had been traced to an electric bell's sparks, and regretted "that the safer plan of excluding sparks altogether was not adopted." From 1919 the UK Mines Department tested and certified bells and relays, and in 1926 it opened a testing station in Sheffield for signalling equipment, according to a 1933 Ericsson bulletin. Limiting a circuit's energy so that its sparks cannot ignite gas became the method called intrinsic safety, written into British Standard BS 1259 in 1945.
LessonA low voltage does not prevent ignition; only a tested limit on the energy a circuit can release does.
Credit
R. A. S. Redmayne (inquiry); the UK Mines Department
1994
A current that means failure
A transmitter sends its reading as a current between 4 and 20 milliamps (mA), so a control system cannot tell a failed instrument from a low reading unless failure has a current of its own. NAMUR, an association of automation users in the process industries, issued its recommendation NE 43 on 18 January 1994: currents from 3.8 to 20.5 mA carry a measurement, and a current below 3.6 mA or above 21 mA means the transmitter has failed. NAMUR revised it in 1996, 2003 and 2021. ISA, which keeps the US 4–20 mA standard, still describes the use of the bands below 4 mA and above 20 mA as "proprietary additions" with "no general call to standardize" them.
Credit
NAMUR
2002
One safety check per segment
A fieldbus links many instruments on one cable. Proving such a cable intrinsically safe by the usual method meant adding up the energy every device and the cable could store, calculations that Ian Verhappen, writing in Control Global, called "very cumbersome." Instrument makers asked Germany's national metrology institute, the Physikalisch-Technische Bundesanstalt (PTB), for a simpler model, and its tests, in report PTB-W-53e, found that with a single power supply and cables within set limits, the devices' and the cable's stored energy could be neglected. The resulting Fieldbus Intrinsically Safe Concept (FISCO) became an IEC technical specification, IEC TS 60079-27, in 2002 and a full IEC standard in 2005, so an engineer checks a segment against a short list of limits in place of the full calculation.
Credit
PTB; the IEC
2015
One package per device
Every instrument model needs a device description, the file that tells control and maintenance software how to set the instrument up, read its diagnostics and calibrate it, and makers had been writing these in two separate forms: EDDL, a text-based description language, and FDT/DTM, small software drivers. FDI (Field Device Integration) combines the two in one package per device. It was developed jointly by FieldComm Group, the FDT Group, PROFIBUS & PROFINET International and the OPC Foundation, with ABB, Emerson, Endress+Hauser, Honeywell, Schneider, Siemens and Yokogawa, and the IEC published it as IEC 62769 in 2015. FieldComm Group tests and registers FDI devices and host software in its own lab.
Credit
FieldComm Group, FDT Group, PROFIBUS & PROFINET International and OPC Foundation
Watch for
Same data from every maker
The Process Automation Device Information Model (PA-DIM) gives each instrument's readings, identity and status the same names whatever its maker or network, so that one program can read them all. It is co-owned by FieldComm Group, ISA100 WCI, NAMUR, ODVA, PROFIBUS & PROFINET International, VDMA and ZVEI, working with the OPC Foundation. Its April 2026 brochure says the groups "are working together to establish a conformance and certification process" and that the OPC UA form "can be validated with the OPC Foundation CTT tool already."
Achieved whena certification program is running and an outside party reads PA-DIM data from instruments of several makers over at least two networks with one generic program and publishes what worked.
Claimed by
the PA-DIM working group, chaired by Frank Fengler of ABB, April 2026
Open
01
Fault currents, measured
A bench test of how named transmitters really signal their own failures. Wire each transmitter in a 24 V loop, cause a set of faults (sensor wire cut, sensor shorted, supply voltage lowered, reading driven past its range), and log the current it sends, how fast it gets there, and the status it reports over HART, the digital signal on the same wires, against NE 43, NAMUR's NE 107 status categories and the maker's own manual.
The briefCredit · unclaimed02
Certified values as open data
An open, machine-readable table of the safety values printed on the certificates of intrinsically safe instruments and of the barriers and isolators that limit a loop's energy: the most voltage, current and power each may give out or receive, and the capacitance and inductance it holds or allows. An AI model reads the values from public certificates and a person confirms every one. Add a checker that runs the comparisons the standard IEC 60079-25 asks for over a list of loops: the source's output must not exceed what the field device is certified to accept, and the capacitance and inductance the source allows must cover the device's plus the cable's.
The briefCredit · unclaimed
Open 01 · the brief
Fault currents, measured
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A bench test of how named transmitters really signal their own failures. Wire each transmitter in a 24 V loop, cause a set of faults (sensor wire cut, sensor shorted, supply voltage lowered, reading driven past its range), and log the current it sends, how fast it gets there, and the status it reports over HART, the digital signal on the same wires, against NE 43, NAMUR's NE 107 status categories and the maker's own manual.
Why it is still open
NE 43 is a recommendation that makers declare they follow, and ISA calls the use of currents outside 4–20 mA "proprietary additions" that nobody has asked to standardize. Makers' alarm levels differ: an instrumentation guide lists 3.75 mA, 3.6 mA or less, 21.75 mA and 23 mA in use. Searches of the Crossref and OpenAlex indexes of published papers found no test that compares named transmitters' fault behaviour.
Built with what exists
used HART temperature and pressure transmitters from several makers; a 24 V bench supply and a 250 ohm precision resistor; a logging data acquisition unit such as a LabJack T7; relays to cut or short the sensor leads on cue; a hand pump kept below 2 bar for over-range tests; a USB HART modem with PACTware, a free device configuration program, to read status; Python to run the sequence and score the results.
First step
Buy three used HART temperature transmitters from different makers, wire each to a 24 V supply and a 250 ohm resistor, and log the current when one RTD lead is pulled.
Done looks like
At least 20 transmitter models from at least 8 makers, each put through at least 6 fault types three times, with the current reached, the time taken and the status reported for every run, set against NE 43 and the maker's manual, published with the raw logs.
Open 02 · the brief
Certified values as open data
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open, machine-readable table of the safety values printed on the certificates of intrinsically safe instruments and of the barriers and isolators that limit a loop's energy: the most voltage, current and power each may give out or receive, and the capacitance and inductance it holds or allows. An AI model reads the values from public certificates and a person confirms every one. Add a checker that runs the comparisons the standard IEC 60079-25 asks for over a list of loops: the source's output must not exceed what the field device is certified to accept, and the capacitance and inductance the source allows must cover the device's plus the cable's.
Why it is still open
Loop calculators exist, but each value goes in by hand. AutomationForum's free online calculator asks for each source value "from its certificate / datasheet" and tells users to confirm the result against the maker's control drawing, and MTL's guide says the document confirming a whole system's safety "is then produced using the information from the individual apparatus certificates." Searches found no public table of certified values that a program can read.
Built with what exists
Python; makers' public certificates and control drawings in PDF; an AI model that reads long documents and tables (see the Note "Seven more flagged" for a model finding problems in long documents that people had missed); the worked examples in MTL's guide as test cases for the checker. The output is an aid to checking and never stands in for the system document a plant must hold.
First step
Download 30 public certificates for common transmitters and barriers, have a model read their values, and count its errors against a hand reading.
Done looks like
Values for at least 300 certified devices, with a random sample of 50 checked by hand against the certificates and at least 99% of values correct; a checker that agrees with the guide's worked examples and finds every one of at least 40 errors that an engineer plants in 200 test loops while holding the answer key; table and code published.