Start with three evidence categories that separate “coverage” from “quality” and “device behavior”: (1) link quality (RSRQ/SINR trends, not just RSRP), (2) uplink stress (Tx power climbing to max, retries/HARQ/reattach counts), and (3) mechanical dependency (orientation/hand/battery proximity changing KPIs). If RSRP is stable but SINR/RSRQ collapses, interference or antenna detuning is likely. If Tx power saturates with poor results, RF loss or antenna efficiency issues dominate.

• Capture a snapshot at each drop: RSRP/RSRQ/SINR, Tx power, band/channel, cell ID, retry counters, min VBAT, temperature.
• Run A/B in the same location: rotate device, add/remove enclosure parts, compare KPI deltas.
• Route next steps: antenna/mechanics first, then RF chain budget, then log-based separation of network vs device.

The most common causes are detuning and efficiency loss driven by ground-plane and nearby metal/absorbers. Moving a battery or adding a metal bracket changes the effective counterpoise and shifts the antenna resonance, increasing mismatch loss and reducing radiated power. Cable/connector routing can also create common-mode currents that distort the radiation pattern. These effects often show up as longer scan time, higher retry counts, and higher Tx power for the same throughput.

• Compare time-to-attach and KPI deltas between “open frame” and “final enclosure” builds.
• Check ground continuity and “keep-out” distance around the antenna; re-validate placement tolerances.
• Use a controlled A/B: temporary known-good antenna path to confirm whether mechanics dominate.

Average current hides the real cost when cellular behavior is bursty: short high-current TX windows cause voltage droop, which triggers retries, longer attach time, and additional wake cycles—multiplying energy per report. Measure a time-aligned window that includes sleep → wake → attach → TX burst → idle/eDRX → return to PSM. The key quantities are peak current, burst duration, and the minimum VBAT/VSYS during the burst (not the average over minutes).

• Measure current and VBAT simultaneously; compute energy per message (∫V·I·dt) across the entire report cycle.
• Correlate spikes with retry counters and attach duration—energy loss is often retry-driven, not payload-driven.
• Validate that PSM/eDRX settings match the reporting schedule; misalignment can increase wake overhead.

Use a fast triage split into three branches: Band/region mismatch shows up as repeated scans with no viable cell on the expected bands, clock instabilityantenna/mechanics

• Log band/channel attempts and attach cause; if failures are region-specific, start with H2-2 band/profile verification.
• If failures cluster right after wake or at temperature edges, check oscillator warm-up/stabilization behavior (H2-9).
• If small mechanical changes create big KPI swings, prioritize antenna/ground-plane coupling (H2-4).

LTE-M typically supports more responsive reporting and mobility use-cases, so the strategy can tolerate more frequent short uplinks. NB-IoT is optimized for deep coverage and low throughput, with higher latency tolerance—so frequent wakes are often expensive due to longer attach/transaction overhead. In practice, NB-IoT designs benefit from batching (send more per wake), longer sleep spans, and careful alignment of report interval with PSM/eDRX cycles to reduce overhead.

• Pick reporting intervals that minimize attach overhead: fewer wakes with more data per wake is often better for NB-IoT.
• Use event-driven wake where possible; avoid periodic wake that conflicts with eDRX/PSM timing.
• Validate the “energy per report” budget rather than relying on average current only.

Charge-and-transmit stresses the power-path because TX bursts demand high peak current while the charger input path may be current-limited. The first two drop points to check are: (1) the modem/system rail near the modem VBAT pins (post power-path FET/ORing), and (2) the source-side rail (charger output or input rail) that may fold back or sag under combined load. A stable average voltage can still hide short brownout windows that cause reboots.

• Probe close to the modem supply pins and at the charger/system node; trigger on TX burst to catch minimum voltage.
• Check if droop correlates with Tx power ramps (poor signal increases burst stress).
• Confirm reset thresholds and recovery behavior match real burst waveforms, not bench steady-state.

eSIM introduces a provisioning and personalization pipeline that must be stable before PVT: correct eUICC form factor, profile delivery method, factory programming flow, and operator acceptance prerequisites. The most common blockers are mismatched profile logistics, unclear ownership of EID/profile inventories, and late discovery of operator-specific onboarding requirements. If these are not locked by DVT, production can stall even when hardware is stable.

• Confirm eUICC variant, manufacturing flow (when/how profiles are loaded), and the “bootstrap connectivity” assumption.
• Align identity chain artifacts (device ID, EID, credentials) with production traceability requirements.
• Plan certification/operator milestones early if the target market requires operator acceptance steps.

Both mechanisms can dominate, so separate them with measurements. Battery internal resistance increases at low temperature, causing deeper TX/attach voltage droop, which triggers retries and longer awake time. Clock stabilization issues appear as slower acquisition after wake or increased sensitivity to warm-up time, even when supply voltage is stable. The fastest split is: if min VBAT dips during attach/TX and correlates with failures, power-path dominates; if VBAT is stable but acquisition time worsens after sleep/temperature change, clock behavior dominates.

• Record min VBAT at attach/TX and compare across temperature; correlate with retry counters.
• Log time-to-attach after wake; if warm-up time improves behavior, clock stabilization is implicated.
• Use temperature as a first-class KPI in the evidence snapshot.

Use terminal-side KPIs to split the problem. If Tx power is maxed and SINR/RSRQ are poor, the limiting factor is often interference or cell loading (network quality). If Tx power is high but SINR is acceptable and performance is still poor, suspect uplink path loss or antenna efficiency loss that forces higher Tx for the same link budget. Also watch for repeated retransmissions (high retries/BLER), which inflate awake time and reduce effective throughput.

• Compare throughput vs SINR/RSRQ; poor quality with max Tx points to RF environment or detuning.
• Check RF chain loss/mismatch indicators: enclosure sensitivity, frequency/band-specific collapse, large Tx ramp behavior.
• Use the evidence snapshot to separate “network conditions” from “device radiated efficiency.”

The most common pitfalls are (1) assuming global band coverage when the module is region-optimized, (2) discovering late that target operators have different readiness or acceptance requirements for RedCap categories, and (3) enabling the wrong band set or feature profile that changes test outcomes. RedCap sits in an evolving ecosystem; selection must be anchored to the target region’s band plan and a realistic certification/operator milestone plan. Locking these early prevents “works in lab, blocked in launch” scenarios.

• Verify band/frequency variants and regional SKUs early; do not rely on marketing names alone.
• Align DVT/PVT milestones with certification/operator acceptance timelines.
• Freeze the enabled band/profile set before entering long certification queues.

A minimal yet powerful set is: timestamp, PLMN/cell ID, band/channel, RSRP/RSRQ/SINR, Tx power, attach state + cause, retry counters, min VBAT, temperature, and reset reason/uptime. This set enables three splits: radio quality vs coverage (RSRQ/SINR vs RSRP), uplink stress vs normal (Tx power + retries), and power/thermal-induced instability (min VBAT + temperature + reset reason). Trigger capture on failure events for maximum diagnostic value.

• Trigger snapshot on: drop, attach fail, TX fail, reboot; include the “before and after” KPI window.
• Store units/scale and sampling rules to ensure cross-build comparability.
• Use consistent naming so field data can be aggregated and compared across regions.

Module certificates reduce scope, but they do not guarantee instant end-device approval. Any change that alters radiated performance or stability under burst conditions can trigger re-evaluation: antenna type/placement, ground-plane size, enclosure material near the antenna, battery/metal bracket position, RF matching changes, cable routing that adds common-mode radiation, or power-path changes that create brownout windows during TX. The practical rule is simple: if a change can move TRP/TIS or create resets, treat it as a retest risk and run delta checks before queuing labs.

• Maintain a “golden” mechanical + antenna build for correlation; compare deltas after any change.
• Use a RED/AMBER change map: antenna/mechanics changes are usually the highest retest risk.
• Lock evidence pack and change-control gates before entering long certification queues.