Split-core DC current sensors: openable Hall-effect transducers that close around an installed cable or busbar, chosen by window, output type and response time
Openable Hall-effect current sensors · fitted over a live cable or busbar without disconnecting it; raw ±4 V output at 10 kHz versus a 4–20 mA loop at 100 ms, window size, return-conductor spacing and the offset that dominates at low current
A solid-core current sensor has to be threaded onto a conductor before the conductor is connected. A split-core sensor is hinged or separable: the magnetic core opens, closes around a cable or busbar that is already carrying current, and is fastened shut. Makers sell it for retrofit work on battery banks, PV strings and combiner boxes, traction and auxiliary loads, where cutting a cable is not an option. The decision is less about the Hall element than about what comes out of the box — a fast raw voltage for a controller, or a slow, conditioned 4–20 mA signal for a PLC — and about how the sensor sits on the conductor.
Split-core (openable) Hall-effect current sensors and transducers for DC and mixed DC/AC currents, fitted over existing cables or busbars: window size and rating, output type and response, offset and drift, insulation, and mounting rules. Hall elements, closed-loop and fluxgate current transducers and Hall IC sensors are covered on the Hall effect sensor page; split-core and other AC current transformers for metering and protection on the Current transformer page; shunt-based measurement on the Precision current shunt resistor and Current-sense amplifier pages; test-bench current and voltage front ends on the Measurement module page.
CNFX editorial research · Sources checked October 13, 2026 · 3 sources
A raw ±4 V signal at DC–10 kHz or a 4–20 mA loop that settles in 100 ms
Split-core DC sensors are sold in two quite different forms. LEM's HTR 50…500-SB is an open-loop Hall transducer that outputs ±4 V into a load above 10 kΩ from a ±12…15 V supply, with a bandwidth of DC to 10 kHz (−1 dB) and a delay under 10 µs to 90 % of nominal current. NK Technologies' DT Series combines the Hall sensor with a signal conditioner and outputs 0–20 mA, 4–20 mA, 0–5 V or 0–10 V (±10 V on bidirectional models) from a 24 V AC/DC supply, with a 100 ms average response to a 90 % step and a DC-only frequency range. [1][2]
The two forms suit different loops. LEM lists switched-mode power supplies, UPS, choppers and battery-supplied applications for the HTR; NK lists battery banks, traction and auxiliary loads, welding and PV string and combiner monitoring for the DT. [1][2] A 100 ms transducer reports a PV string or a battery charging current well but will not see a converter's ripple or a fast fault; a 10 kHz open-loop sensor needs a bipolar supply and an input that accepts ±4 V. NK's current outputs are limited to 23 mA and its 0–10 V output to 11.5 V when the input exceeds range. [2]
Published item
LEM HTR 50…500-SB
NK Technologies DT Series (split-core)
Accuenergy HAK21 / HAK40
Range
IPN 50–500 A; measuring range ±100 to ±1,000 A [1]
0–50 A min. to 0–400 A max., jumper-selectable ranges [2]
50–200 A (21 mm window), 400–1,000 A (40 mm window) [3]
State the measured current range and direction (unipolar, bipolar or bidirectional), the bandwidth or response the control loop needs, the supply available, and the input the sensor will feed (controller ADC, PLC analogue card, panel meter).
02
21 mm or 40 mm windows, a return busbar 2.5 window-lengths away, and a primary below 100 °C
Accuenergy's HAK split-core sensors are sized by the cable window: 21 mm diameter for up to 200 A and 40 mm for up to 1,000 A, marketed for retrofit on existing wiring without any disconnection. LEM's HTR has a 21 mm primary through-hole and fastens with two 3.2 mm holes 50 × 36 mm apart, or one hole and a 2.5 mm spigot 36 mm apart. [3][1]
LEM's mounting remarks on the HTR carry the rules that a split core makes easy to break when it is clipped onto existing wiring: the return busbar and any primary conductor elbow should sit at least 2.5 × the window length away from the case; dynamic performance is best with a busbar that completely fills the aperture; and the primary conductor should not exceed 100 °C. Output is positive when current flows in the arrow direction. [1] NK's DT split-core case comes with DIN-rail adapters and built-in mounting brackets, and its terminals take 12–22 AWG wire. [2]
Insulation is rated on the sensor, not on the cable it clamps. LEM tests the HTR at 3 kV RMS for 1 minute, with impulse withstand above 6 kV, 28.1 mm creepage and 17.1 mm clearance, giving 1,000 V basic or 500 V reinforced insulation under EN 50178 and IEC 61010-1 at overvoltage category 3 and pollution degree 2. NK's DT is UL listed to 1,270 V AC and tested to 3 kV. [1][2]
Give the conductor type and size (cable diameter with insulation, or busbar width and thickness), the position of return conductors and bends, conductor temperature, system voltage and the insulation level required.
03
Offset and drift in millivolts: why a 50 A open-loop sensor misreads amperes at light load
LEM gives the HTR's electrical offset at zero current as ±45 mV (typ.) and ±65 mV max at 25 °C; its offset variation from −10 to +70 °C as ±70 mV (typ.) and ±240 mV max; and a magnetic offset after an overload of 3 × IPN of ±10 mV (typ.), ±20 mV max. The ±2 % error figure excludes the electrical offset. [1]
Editorial calculation, taking the ±4 V output as the value at nominal current: on the HTR 50-SB (IPN 50 A) 4 V corresponds to 50 A, or 80 mV per ampere. The ±65 mV maximum offset is then about ±0.8 A and the ±240 mV maximum temperature variation about ±3 A, against a ±1 A error band (2 % of 50 A) at nominal current. On the HTR 500-SB the same millivolts are ten times more amperes, about ±8 A and ±30 A. [1] A sensor rated far above the normal load reads small currents poorly, so the rated current should sit close to the current that matters, and a zero reading should be taken with the conductor de-energised if the controller can store it.
Editorial calculation for the conditioned type: NK's 2.0 % FS on a 0–400 A split-core range is ±8 A anywhere in the range, so a 20 A auxiliary load would read with up to 40 % error; selecting the lowest jumper range that covers the load narrows it. [2]
Ask for the offset, offset drift and accuracy at the actual load current and temperature range, not only at nominal current, and whether accuracy figures exclude offset.
04
Data for a split-core current sensor quote
Paste it into your enquiry below or into an email to a maker, then replace every bracket.
Current: [normal and maximum A; DC, AC or both; direction]
Signal: [±V, 4–20 mA, 0–10 V; bandwidth or response time]
Accuracy: [at the working current and temperature; offset included or excluded]
Conductor: [cable outer diameter or busbar size; return conductor position]
Insulation: [system voltage; basic or reinforced; certification]
Supply: [±15 V, 24 V AC/DC, meter-powered]
Mounting: [panel, DIN rail, hanging on cable]
Explore company information and manufacturing capabilities, then discuss the scope of your project with a relevant supplier.
Share this page
Compiled and source-reviewed by the CNFX editorial team.
Each manufacturer above links to the page on that company's own website where the product is listed, with the date we checked it. How we source and review
· Page last updated 13 October 2026
Buyer questions
What is a split-core current sensor used for?
It opens and closes around a conductor that is already installed; Accuenergy markets its HAK sensors for retrofit on existing wiring without disconnection, and NK lists battery banks, traction loads, welding and PV strings for its DT Series. [3][2]
How fast does a split-core DC current sensor respond?
It depends on the output type: LEM's open-loop HTR has DC–10 kHz bandwidth and under 10 µs delay, while NK's conditioned DT Series averages 100 ms to a 90 % step. [1][2]
How accurate are split-core DC current sensors?
Published figures include <±2 % at nominal current with offset excluded (LEM HTR), 2.0 % of full scale (NK DT split-core) and 0.5 % (Accuenergy HAK). [1][2][3]
Where should the return conductor be relative to the sensor?
LEM's HTR datasheet asks for the return busbar and any conductor elbow to be at least 2.5 × the window length from the case. [1]
Sources and what still needs a direct check
Sources checked 2026-10-13. Three manufacturer documents from three publishers: LEM's HTR 50…500-SB datasheet (version 9, 2019), NK Technologies' DT Series 4-wire split-core datasheet (04/2024), and Accuenergy's HAK21/HAK40 cut sheet.
