Long-term stability figures on current production decade resistance box lines span roughly 20 ppm/year on mid-range laboratory units up to 100 ppm/year on sub-1 ohm ranges, while high-resistance decade systems and dedicated resistance standards push below 5 ppm/year through foil or hermetic metal-film elements.
Four published data sheets (HARS, 1067, DB62, HRRS-5 kV, plus the GenRad 1433 / 1413 lineage still in distribution) cover the decision space from field service to primary-lab reference work, and they agree on which design choices actually move the ppm/year number.
What "ppm per year" means on a decade box
A decade resistance box is a switched assembly of fixed resistors, one resistor per dial position, that presents a settable total resistance from 1 mΩ to over 100 MΩ depending on configuration. The annual drift figure is the resistance change, normalised to the set value, after 12 months of powered-off or low-duty-cycle storage at controlled lab temperature, expressed as ppm of reading [S2][S4].
For reference, 20 ppm/year on a 10 kΩ dial step equals 0.2 Ω of allowable drift; on a 100 Ω step the same spec only allows 2 mΩ, which is why the same instrument carries a different stability figure on its low-ohm decades [S4]. The Time Electronics 1067 datasheet states "Better than 20 ppm/year (> 1 Ω), Better than 100 ppm/year (< 1 Ω)" precisely for that reason [S4][S5].
Mid-range laboratory benchmark: 20 ppm/year
The 20 ppm/year figure has become the de facto floor for bench-grade decade boxes. The IET Labs HARS series quotes "good stability typically less than 20 ppm/year" alongside ±(0.01% + 2 mΩ) accuracy and < 1 mΩ zero resistance per decade [S2]. The Time Electronics 1067 matches it: "Better than 20 ppm/year" at 0.01% accuracy across a 10 mΩ to 12 kΩ range with six-decade resolution [S4][S5]. The ESI / Tegam DB62, a six-decade DC and audio-frequency unit, is specified at ±(0.01% + 2 mΩ) accuracy with stability to ±20 ppm per year and temperature coefficient to ±5 ppm/°C [S7].
The Time Electronics 1067 precision decade resistance box, which specifies better than 20 ppm/year stability, uses multiple self-wiping silver alloy switch contacts and low thermal EMF terminals in a fully screened metal case. The HARS datasheet calls out "High-performance solid silver-alloy contacts" and "Non-inductive or low-inductance resistors" with a 5 ppm/°C temperature coefficient as the supporting numbers behind the 20 ppm/year headline [S2].
High-stability tier: 5 ppm/year and below
When the application is calibration of a Megger, sourcing into a teraohm bridge, or transferring a resistance ratio between standards, the resistor element has to be foil or hermetically sealed metal-film, and the switched contacts contribute proportionally less drift than the element itself. IET Labs' HRRS-5 kV and 10 kV high-resistance decade boxes cover 1 GΩ up to 11 TΩ at 5000 V or 10000 V maximum, using "the most stable and accurate resistors and custom high voltage switches" [S1].
ATSP's published spec table for the same HRRS-5 kV family lists "Accuracy: As low as 20 PPM, Stability: < 5 PPM per year" on the decade-resistance-box line, dropping to "< 1 PPM per year" on the fixed resistance standard line that covers 100 µΩ to 10 PΩ [S3]. The VPG Foil Resistors M-ADR6105, a precision foil decade standard, sits in this same tier and is the type of unit specified as a transfer standard rather than as a working bench source [S6].
The GenRad 1413 precision decade capacitor, still in distribution as a refurbished unit, lists "Excellent Stability 100 ppm per year" on capacitance, a useful reminder that the ppm/year envelope is a function of the resistive or reactive element technology, not the decade-switch mechanism [S1].
Side-by-side comparison of representative units
For procurement, four decision criteria separate the tiers: resistance range, accuracy, long-term stability, and temperature coefficient. [S4]
On range, the HARS spans 1 mΩ to 111 MΩ with 1 to 11 decades [S2], the 1067 covers a narrower 10 mΩ to 12 kΩ in six decades [S4], the DB62 covers six selectable decades in the DC and audio band [S7], and the HRRS-5 kV starts at 1 GΩ and runs to 11 TΩ for high-voltage high-resistance work [S1][S3]. On accuracy, the HARS-X, 1067, and DB62 all land at ±(0.01% + 2 mΩ); the HRRS-5 kV is specified at 20 ppm accuracy with < 5 ppm/year stability [S1][S2][S3][S4][S7]. On stability, 20 ppm/year is the working-bench floor and 5 ppm/year marks the high-stability tier; on temperature coefficient, 5 ppm/°C on the HARS and DB62 is the supporting number that keeps the annual drift figure valid in a 20 °C to 25 °C lab [S2][S7]. The 1067 carries a wider TC budget at < 10 ppm/°C above 1 Ω and < 20 ppm/°C below 1 Ω, which is why its low-ohm stability is the looser 100 ppm/year figure [S4][S5].
The implication for users: a 20 ppm/year decade box that sees only 5 °C of annual lab-temperature swing adds another 25 to 50 ppm of TC error on top, and the combined uncertainty is what gets entered into the calibration budget, not the headline stability alone [S2][S4][S7].
Where a decade box is, and is not, the right tool
Decade resistance boxes earn their place where the resistance value must be set, read, and re-set by a human or a programmed switch: simulating a Pt100 sensor for a temperature-indicator calibration [S4][S5], building a Wheatstone or Kelvin bridge arm, calibrating a megohmmeter, or substituting a known resistance into a process loop. A separate reference article covers the adjustable-reference design pattern in more depth at Decade resistance box as adjustable precision reference. For high-ohms work above 1 GΩ, the same datasheets show that a switched decade is workable up to about 11 TΩ but that the TC and voltage-coefficient contributions start to dominate, and a fixed standard in an oil or air bath is the more honest reference [S1][S3].
Decade boxes are not the right tool for AC applications above audio, where skin effect and switch capacitance matter more than element drift; they are not the right tool where a four-wire Kelvin connection to a fixed standard is already available; and they are not the right tool where the calibration budget is dominated by switch contact resistance rather than element stability, which is the case for any sub-10 mΩ step [S2][S4][S5].
Verification cadence and what to watch
Manufacturer recommendation is to recertify a 20 ppm/year box on a 12-month cycle against a reference standard whose own annual drift is at least four times lower, so a working standard at < 5 ppm/year, and to log the residual end resistance at the time of calibration because the silver-alloy contact set is the most likely wear item [S2][S4][S5]. The Time Electronics 1067 datasheet offers NPL-traceable and UKAS calibration certificates as ordered options, which is the practical implementation of that recommendation [S5].
Two trackable signals to watch: the migration of foil-resistor elements (VPG M-ADR6105 class [S6]) into programmable decade form factors at sub-5 ppm/year, and the widening of high-voltage high-resistance decade coverage past 10 TΩ as 1500 V PV insulation test sets demand higher source impedances. Both will move the published ppm/year floors down over the next product cycle.
For component-level specifications, see insulation resistance tester, and rtd pt100.