ERS-2026-003Engineering Readiness SeriesReleased

Why Carrier-Board Module-Current Measurements Can Be Misleading

Off-state current attribution at the supply header of two off-the-shelf cellular carrier boards

Log-scale bar chart of supply-header current against module power-off and PSM figures for two boards. BG95-M3: 1.539 mA header current against 12.99 µA power-off leakage and 3.89 µA PSM, about 118 times. SIM7080G: 1.886 mA header current against a 15 µA power-off maximum and 3 µA PSM, at least 125 times.

0.000 mA

BG95-M3 Off-State Header Current

≈0×

BG95-M3 vs Power-Off Leakage

≥0×

SIM7080G vs Power-Off Maximum

Primary Instrument
Nordic Power Profiler Kit II, 100 kS/s
Boards Evaluated
DIP40CORE-A (Quectel BG95-M3), Waveshare SIM7080G Cat-M/NB-IoT HAT (SIMCom SIM7080G)
Constraint
No hardware modification
Raw Recordings
Zenodo, doi:10.5281/zenodo.23399520
Engineering Readiness Outcome
Module-level off-state current not attributable at the supply header of either board

Assessment

A current measurement at a cellular carrier board's supply header records the current entering the board through that point. Where the module shares that supply path with other loads, the reading includes their consumption.

ERS-2026-003 examines two off-the-shelf carrier boards from different vendors, carrying modules from two manufacturers, with the same instrument and no hardware modification. It asks whether supply-header current can be attributed to the module alone, and defines a procedure for assessing attribution before a header reading is reported as module current.

Sleep and Power Saving Mode current were not measured, and no energy-consumption result is reported.

Engineering Findings

  1. 1DIP40CORE-A (Quectel BG95-M3). With the module confirmed off by its STATUS and VDD-EXT outputs, the supply header drew 1.539 mA over a 30-minute capture, about 118 times Quectel's published BG95-series power-off leakage figure of 12.99 µA (Quectel publishes no separate LA-variant figure). The result reproduced the earlier 1.53 mA measurement within 0.6 %, and a module-side UART disconnection changed the mean by less than 0.13 %.
  2. 2Waveshare HAT (SIMCom SIM7080G). Re-measured at 3,800 mV in two 5-minute recordings, the module stayed off after the supply was applied, was started with PWRKEY and was then powered down with PWRKEY. After power-down the supply header drew 1.886 mA, at least 125 times SIMCom's published maximum power-off current of 15 µA, and within 0.6 % of its level before the module was first switched on in both recordings.
  3. 3Board-level paths. On the SIM7080G board, the schematic places an always-on power LED on the shared VBAT net. On the BG95-M3 board, two resistive paths account for an estimated 0.3 to 0.4 mA, and the on-board regulator's unconnected input was back-powered from the supply rail to 3.33 V. The current through that path was not measured, so the remaining current has not been accounted for.
  4. 4Attribution outcome. Module-level current could not be isolated at the header of either board. The results do not establish a board fault; they show that a header measurement includes board-level paths.

Evidence Basis

  • Instrument: Power Profiler Kit II checked at three points with deviations below 1 %; the off-state operating points (source meter, 1.5 to 1.9 mA) were not separately checked.
  • Module state: STATUS and VDD-EXT outputs read independently of the current reading, with an AT-interface cross-check (BG95-M3). PWRKEY power-down with the STA LED dark when off and lit when on, STATUS not metered (SIM7080G).
  • Raw data: a 30-minute off-state recording and a 5-minute UART disconnection recording (BG95-M3), and two 5-minute recordings at 3.8 V (SIM7080G), analysed from the raw samples.
  • Topology: vendor schematics for both boards, with every other source and sink on the shared supply net identified.
  • Comparators: module manufacturers' hardware design documents, cited by version, table and statistic.
  • Reproducibility: raw recordings, analysis scripts and SHA-256 values published in the supporting repository and Zenodo dataset.

Procedure

  • Cite the module manufacturer's power-off, PSM and sleep figures by document version, table and statistic.
  • Establish supply topology from the board schematic, and identify every other source or sink on a shared node.
  • Establish module state from signals independent of the current reading.
  • Measure off-state header current at the specified supply voltage for at least 10 minutes, hands off, and analyse the raw recording.
  • Compare like with like. If the ratio to the module figure is much greater than one, report at system level.

Engineering Decision

Following independent engineering evaluation, Radianode considers that a supply-header reading on an off-the-shelf cellular carrier board should not be reported as module current unless three conditions are met and recorded: the off-state header current has been compared, like for like, with the module manufacturer's published off-state figure; module state was established independently of the current reading; and results are labelled at the attribution level they support.

System-level header-current measurements remain available on both boards. Module-level attribution requires a board that exposes the module supply in series, or a modification that this assessment's constraint excludes. On the DIP40CORE-A, a header measurement in network-confirmed PSM is the recommended next step.

Does not evaluate

  • Sleep, PSM or eDRX current under a live network
  • Any board other than the two evaluated
  • Statistical behaviour across multiple units of either module
  • The share of each board-level consumer on either board
  • Any modified board, or any energy consumption result
  • Temperature dependence

ERS-2026-003 · Version 1.0 · October 2026 · 25 pages

Publication Details

Report
ERS-2026-003 · Radianode Engineering Readiness Series
Author
Oluwanifemi A. Ogunjemilua · ORCID
Issued by
Radianode Ltd

Radianode provides independent engineering validation for connected and wireless systems. If a current or battery-life figure in your design rests on a header reading, talk to an engineer.