A signal calibrator is selected by the physical quantity it sources or simulates, its accuracy class, channel count, channel-to-chassis isolation, and how closely it mirrors the real sensor output [S3]. A bench instrument that simulates a 350 Ω strain-gage full bridge to 0.025 class behaves nothing like a portable 8-channel DC voltage/current source for ECU inspection [S1][S2].
The category spans 56 products across 31 manufacturers in the industrial vendor index, ranging from passive resistor-network reference units to multifunction process calibrators covering -270 °C to 2,320 °C and -1 bar to 1,000 bar [S3]. The right tool sits at the intersection of signal class, accuracy, and the field-versus-lab deployment envelope.
Physical Quantity and Sensor Emulation
The first gate is the physical quantity the calibrator must source or measure: DC voltage, DC current, resistance, strain-gage mV/V, charge in picocoulombs, or a process variable like pressure and temperature translated to mA [S1][S2][S4]. A passive K3607 calibration unit is built around a high-precision resistor network in star configuration to simulate 350 Ω strain-gage full-bridge transducers, delivering 30 calibration signals between 0.05 mV/V and 10 mV/V [S2]. That is a fundamentally different instrument from a U8794 VIR generator unit, which sources DC voltage (-0.1000 V to +5.3000 V at 0.1 mV resolution), DC current (four ranges from ±50 µA to ±5 mA), and resistance (10 Ω to 1 MΩ, 6-digit setting) for ECU sensor simulation rather than bridge emulation [S1].
For vibration and piezoelectric chains, charge-mode calibration applies: Q (pC) equals E (mV) when the reference capacitor is exactly 1000 pF, and the charge must be expressed in units of pC pk to keep the transfer function linear [S4]. Selecting the wrong physical-quantity gate — for instance, a voltage-only source for a charge-mode accelerometer calibration — produces meaningless data regardless of the instrument's stated accuracy.
Accuracy Class and Resolution
Accuracy class governs whether a calibrator is a primary reference or a workhorse. The K3607 strain-gage unit sits in accuracy class 0.025, achieved through extremely low self-capacitance and self-inductance of the inbuilt precision resistors [S2]. Multifunction process calibrators in the same vendor index, such as the MSC series, hold better than 0.05 % accuracy for each type of input and output, with voltage handling up to 24 V [S3].
Resolution must be checked against the set value, not just the range. The U8794 specifies 0.1 mV on its ±5.3 V DC voltage range, 0.01 µA on its ±1 mA current range, 0.001 µA on its ±50 µA range, and 6-digit setting on resistance from 10 Ω to 1 MΩ [S1]. For loop calibrators reading 4-20 mA, sub-µA resolution is the threshold that separates trimming a transmitter zero from guessing it; for a loop calibrator used in mA simulation the same resolution rule applies. Always pair the stated accuracy class with the range end-value, because a 0.025 class instrument at 0.05 mV/V full scale behaves very differently from the same class at 10 mV/V full scale.
Channel Count, Isolation, and Carrier Frequency
Channel architecture decides whether one calibrator can replace several. The U8794 packs 8 isolated output channels on a single 25-pin D-sub connector, with isolation between each channel and between channel and chassis, and a maximum rated voltage to earth of 25 V [S1]. This topology suits parallel simulation of multiple ECU sensor channels in a single test sequence. A single-channel K3607, by contrast, drives one bridge at a time but supports carrier frequencies up to 5 kHz for transducers supplied with DC voltage, and can be operated in six-wire circuit to balance ohmic and capacitive effects of long measurement cables [S2].
Where multiple signal classes coexist on the same plant, a signal conditioner upstream and a multi-channel calibrator downstream are typical pairings. Channel-to-chassis isolation at 25 V is adequate for low-voltage process electronics; projects that demand higher common-mode rejection, especially in substations or field loops with long cable runs, should re-check the isolation voltage against the loop's working voltage before specifying.
Reference-Value vs Multifunction Calibrators
The vendor index cleanly separates reference-value calibrators from multifunction process calibrators [S3]. A reference-value unit like the K3607 is a passive resistor network whose job is to substitute a known electrical input for an unknown mechanical quantity; it does not generate, it simulates [S2]. A multifunction process calibrator like the ADT 222 sources, simulates, and measures pressure (-1 bar to 700 bar), temperature (-270 °C to 2,320 °C), and electrical signals in one chassis, with smartphone-style interface and data logging for planned and completed calibrations [S3].
The selection decision is workload-driven. A calibration lab that certifies strain-gage load cells buys K3607-class hardware for the bridge; a process instrumentation team that handles pressure transmitters, RTDs, and 4-20 mA loops on a single shift buys a multifunction unit. pressure calibrator and pressure transmitter workflows overlap with the ADT 761A and ADT 762W family (-0.95 bar to 1,000 bar with internal electric pump and precision pressure controller) rather than with a resistor-network reference [S3].
Portability, Power, and Field Use
Field use reshapes the spec list. The K3607 is described as mobile application for many measuring chains, with no power input because it is a passive network [S2]. The U8794 is a bench/instrument unit intended to combine with the Memory HiCorder MR8740T and is not portable in the handheld sense [S1]. Handheld process calibrators in the index target 0 °C to 50 °C operating envelope and 0 % to 90 % humidity, which matches typical field and laboratory conditions but excludes cold-weather or condensing-environment use without further protection [S3].
Power and enclosure determine where the instrument can sit on the loop. A 24 V maximum signal rating on the METRACAL CM series means it can be powered from the loop under test in some configurations; a signal isolator placed in series is sometimes required to break ground loops when the calibrator and the device under test share a common return. Specifying a calibrator without checking the loop supply architecture is a common procurement error.
Limitations, Failure Modes, and Mismatches
The dominant failure mode is range-end error: a calibrator with 0.1 mV resolution is precise at full scale but coarse at the low end where transmitter zero lives. Cable capacitance is the second trap: long measurement cables introduce ohmic and capacitive effects that only six-wire calibration circuits can balance, and a two-wire or four-wire calibrator will read correctly at the instrument terminals but wrongly at the sensor [S2]. Charge-mode calibration, where the transfer function Q (pC) = E (mV) holds only when the reference capacitor is exactly 1000 pF, drifts if the capacitor ages or if the calibrator is paired with a non-1000 pF input [S4].
Class-of-instrument mismatch also bites. A signal calibrator is not a signal repeater — the calibrator sources or measures; the repeater regenerates a live loop signal. Treating a calibrator as a permanent loop component or a repeater as a calibration source breaks traceability and, in metered custody transfer, breaks compliance. A reliable data-acquisition pairing, including logger-based verification, follows the spec gates covered in data logger selection, and isolator positioning relative to a temperature controller is detailed in signal isolator vs temperature controller.
Two trackable signals to watch: the convergence of multifunction calibrators with smartphone interfaces into documented, audit-ready workflows, and the gradual replacement of passive resistor-network reference units by software-defined bridge simulators in labs that have standardised on 6-wire measurements.