Every resistance measurement puts energy into the element being measured. In an ordinary installation nobody thinks about it. In an explosive atmosphere it is the entire certification question — and the way you answer it decides how much protection circuitry ends up between your sensor and your ADC.
What the standard is actually asking
Intrinsic safety, defined in IEC 60079-11, does not ask whether a circuit works correctly. It asks whether the circuit can release enough energy — as a spark or as surface heat — to ignite the surrounding atmosphere, both in normal operation and under a defined number of faults. Protection level Ex ia requires that safety hold with two faults applied; Ex ib with one.
The energy thresholds involved are small. Minimum ignition energies are of the order of a few hundred microjoules for methane and propane, and roughly an order of magnitude lower for hydrogen — in the region of twenty microjoules. Gas group IIC, which covers hydrogen and acetylene, is the difficult one, and it is also the one that turns up in refineries, battery rooms and electrolysis plant.
Two things follow. First, the margins are set by physics, not by convention. Second, the assessment covers stored energy too — the capacitance and inductance in the circuit, not just the current the source can deliver.
The conventional answer, and what it costs
The usual approach is to leave the measurement alone and wrap protection around it. A safety barrier or galvanic isolator limits voltage and current at the boundary. Series resistance limits fault current. Shunt elements clamp voltage. Each element is redundant so that it survives the fault count required by the protection level.
This works. It is well understood, widely certified, and it has three costs.
It degrades the measurement. Series resistance in the sense path is series resistance in a resistance measurement. Barriers add their own tolerance, temperature coefficient and leakage, and those errors sit directly on top of the parameter you are trying to resolve to milliohms.
It adds parts that need the same rating. In a high-temperature installation, every protection component now needs a temperature rating too. The barrier that was simple at 40 °C is a sourcing problem at 175 °C.
It moves the certification burden into the system. The measurement channel is only intrinsically safe as an assembly. Change a cable length, a capacitance, an inductance, and the assessment has to be revisited.
Taking the energy out of the measurement instead
There is a different way to arrive at a resistance value. Instead of forcing a sustained excitation current and measuring the resulting voltage, you can measure a time.
Resistance appears in the time constant of any circuit that combines it with a known reactance. Charge a known capacitance, discharge it through the unknown resistance, and time the transition. The result is a time interval that a time-to-digital converter can resolve very finely, and the energy involved is set by the capacitor rather than by a continuous current.
That is the important consequence. The energy stored on a capacitor is one half C times V squared. Ten picofarads at one volt is five picojoules. A nanofarad at one volt is five hundred nanojoules. The designer chooses the energy budget directly by choosing the reactance and the excitation voltage, and the measurement consumes that energy once per conversion rather than continuously.
Compare that with the ignition energies above. A measurement that injects energy in the nanojoule range sits three to four orders of magnitude below the minimum ignition energy of hydrogen — the hardest gas in the hardest group.
Why this is not a certification shortcut
It would be wrong to say that a low-energy measurement principle makes a system intrinsically safe. It does not, and any supplier who tells you otherwise is worth a second look.
Certification is assessed on the complete circuit, including everything that stores or supplies energy: the power rail, the cable capacitance and inductance, the connectors, the enclosure, and the surface temperature classification of the components. A device that limits its own injected energy removes one of the largest contributors to that assessment — it does not remove the assessment.
What it changes is the shape of the problem. Instead of designing a measurement and then protecting it, you start from a measurement that has very little to protect. The protection that remains is smaller, cheaper, and interferes less with the thing being measured.
Where the trade-offs actually are
Time-domain resistance measurement is not free.
Resolution comes from timing, so timing quality matters. The reference used for the time-to-digital conversion, and its stability over temperature, becomes a first-order contributor to measurement accuracy. This is a different set of design problems from a conventional excitation-and-sense front end, not an easier one.
Conversion rate and energy trade against each other. Lower energy per conversion generally means smaller signals and longer settling.
Cable and connector parasitics enter the measurement. In any high-accuracy resistance measurement they do, but the mechanism differs, and lead compensation has to be designed for the topology rather than borrowed from a four-wire ratiometric front end.
None of these are reasons to avoid the approach. They are the questions to ask when evaluating it.
If you are specifying a measurement channel for a classified area
Three things are worth settling before the architecture is chosen.
Which gas group and protection level. Ex ia IIC is a very different starting point from Ex ib IIB, and the difference shows up in every energy budget downstream.
What accuracy has to survive to what temperature. Accuracy at 25 °C is a different specification from accuracy across the operating envelope, and protection components are often the dominant error term at the top of the range.
What the fault set actually looks like in your installation, including cable runs. The standard defines the fault count; your installation defines the parasitics.
If you are working on resistance measurement in a classified or high-temperature area, tell us which standard you are qualifying against and what accuracy has to hold. We can tell you what a device does and does not do for you before you design it in.