An incremental encoder reports motion as pulses; the controller must count them and, after every power cycle, re-establish position by homing. An absolute encoder removes that fragility: it outputs a unique position word the instant it is switched on, with no reference move. Within one revolution the single-turn track supplies that word; across many revolutions a multiturn stage must remember how many times the shaft has turned even while the machine and its 24 V rail are off. That single requirement - retaining a revolution count for months or years with the main supply absent - is why a whole class of multiturn encoders carries an internal backup source, and why their power architecture splits into two very different domains. This first paper on nickel-metal hydride backup for absolute encoders explains the working principle and the resulting load profile: the active domain, where a magnetic or optical encoder ASIC and an SSI, BiSS-C or EnDat line interface draw tens of milliamperes from a 5 V or 10 to 30 V supply; and the retention domain, where a low-power counter must be kept alive on microamperes from a backup cell through an automatic, glitch-free supply switch. Real devices anchor the numbers - magnetic encoder chips spanning 3.0 to 5.5 V with integrated switchover to a backup battery and error flags for low battery, overspeed and CRC; compact magnetic modules with 12-bit single-turn and 16-bit multi-turn resolution; and industrial encoders rated to 12,000 rpm and minus 40 to 100 degrees C - before the later papers turn to source selection and qualification.
The single-turn track resolves angle within one shaft revolution - 12 bits gives 4,096 steps, high-end optical or magnetic encoders extend single-turn resolution toward 17 bits. The multi-turn stage counts revolutions, commonly 12 to 16 bits and, in premium optical encoders, up to 29 bits combined. Concatenated, the two values form one absolute position word that is valid immediately after switch-on, so a driven axis, a crane hoist or a rotary table knows exactly where it is without travelling to a reference switch.
While the encoder is powered, counting revolutions is trivial; the difficulty is that the shaft can still be turned - by hand, by gravity, by maintenance - while the control system is off. Whatever retains the multi-turn count during that interval defines the encoder's memory architecture: a battery-buffered electronic counter, an energy-harvesting stage that generates its own counting pulse from rotation, or a physical mechanical gear train. This paper focuses on the first, the battery-buffered design that still dominates mainstream servo and automation catalogues.
In normal operation the encoder draws its active current from the controller: published magnetic modules specify a 4.5 to 5.5 V supply with a no-load current on the order of tens of milliamperes (a compact BiSS-C module quotes about 70 mA unloaded), while wide-input encoders accept 10 to 30 V DC and draw less current at the higher voltage because internal regulation is more efficient. The interface - SSI clock and data, BiSS-C, EnDat or an RS-422 line driver - adds load only while clocked.
When the external supply disappears, an internal switch transfers the multi-turn counter and its low-power sensing front end to the backup source, where the current falls to the microampere range needed only to detect slow rotation and increment non-volatile or retained count registers. The first animated figure layers these functional domains; the second contrasts the active-milliampere and retention-microampere regimes across repeated power cycles, illustrating why backup sizing is dominated not by energy during running but by quiescent self-discharge and retention current over years.
Modern magnetic encoder chips integrate the supply architecture rather than leaving it to discrete diodes. A representative device operates from 3.0 to 5.5 V, includes integrated supply switching to a backup battery, emits synchronized multi-turn and single-turn data over SSI with error, parity and synchronisation bits, and raises an error output on overspeed, low battery and CRC failure. Integrating the power-path removes the voltage drop and uncertainty of an external diode-OR and guarantees that no revolution is missed in the milliseconds during which the main rail collapses and the backup takes over.
That switchover has a hard real-time aspect: if a powered-off axis is rotated faster than the retention front end can count on its tiny backup current, the count becomes ambiguous - hence the explicit overspeed error flag. Designers must therefore match the backup source's voltage and available current to the chip's specified backup window (one family permits a 3 to 5.5 V backup) and treat the low-battery warning as a planned-maintenance signal rather than an alarm to be ignored until position is lost.
SSI remains the simplest robust point-to-point clock/data interface; BiSS-C adds faster, bidirectional digital communication; EnDat and proprietary fieldbus variants carry position plus diagnostics, including the encoder's own backup-health status. Resolution and speed climb together - compact modules reach 2 MHz SSI clock and 12,000 rpm - but the interface current is active only when the controller clocks the encoder, so it belongs entirely to the powered domain and does not drain the backup.
The diagnostic channel is precisely what makes a managed backup source valuable: rather than discovering a dead backup after a lost homing position, the controller can read the low-battery flag over BiSS-C or EnDat during normal operation and schedule replacement or recharge before the multi-turn count is at risk. This converts an invisible, failure-prone coin cell into a monitored, predictable element of the maintenance plan.
Independent comparisons of multi-turn architectures are blunt about the battery-buffered design's weakness: it retains position only as long as its internal cell holds charge - typically three to ten years per element for a primary lithium cell - and when that cell is exhausted the encoder suffers complete loss of position and must be re-homed. Energy-harvesting (Wiegand) designs remove the external cell by generating counting energy from shaft rotation, but they cannot count the first fraction of a turn after a long power-off and their storage capacitor ages on a roughly decade-scale; mechanical gear trains avoid electronics but introduce gears that wear.
Understanding this failure mode frames the battery design problem correctly. The backup source is not a miniature traction battery delivering work; it is a long-life, ultra-low-current retention reservoir whose worst enemy is self-discharge and whose required output is a guaranteed microampere-level current at the lowest rated temperature, with enough warning margin to avoid an unplanned re-homing operation.
The analysis yields a layered specification for the backup source: a voltage matched to the encoder chip's backup pin with a clean, glitch-free switchover; a retention current budget in the microampere regime sustained for the required off-power interval; operation across the encoder's full temperature span, which for industrial magnetic encoders reaches minus 40 to 100 degrees C; a readable low-battery diagnostic; and a defined service strategy that prevents silent total position loss. The second paper turns these requirements into a concrete source selection, comparing a rechargeable nickel-metal hydride backup with primary lithium, a capacitor and energy harvesting; the third maps the EMC, environmental and battery-standard evidence the encoder vendor must assemble.
Treating multi-turn memory as a dual-supply design problem - rather than as 'add a coin cell' - is what separates an encoder that powers up to a correct position after years of intermittent service from one that forces an unplanned, sometimes dangerous, reference move.
Weijiang Power designs and manufactures sealed nickel-metal hydride cells and matched industrial packs for remote, off-grid and safety-related equipment, and supports OEM partners with IEC 61951-2 performance files, IEC 62133-2 safety evidence, pulse-load characterisation, wide-temperature testing and charger/pack co-validation. Tell us your duty cycle, peak current, temperature envelope, autonomy target and the standards your product must meet, and our engineers will specify a cell-and-pack combination that protects runtime, reliability and service life. Review the range on the products page.
