# Astra prototype qualification

This plan defines the bench measurements required to accept the first two prototypes. All physical tests listed below remain pending. For each unit, the technician records the PCB serial number, file revision, firmware hash, temperature, equipment used, and date. Store the original oscilloscope traces, SWD logs, and calibrated photographs with the test record for that unit. Use the same cable configuration during comparisons.

Verification uses the schematic `8a362004` and the board `0da00a40`. A total of 300 pads were checked per document. The board has 149 positions, of which 139 components are mounted, all on the top layer. In Gerbv, under the antenna, zero copper pixels were measured on both layers. Under the shielded core of the module, coverage is 96.79% on the top side and 100% on the bottom side. Trace reconstruction from the manufacturing files identified 92 nets without opens or short circuits, 210 vias, and 214 holes in total. Native results and SHA-256 hashes · Gerber/drill reconstruction

## Design assumptions

The selected battery is the LiPol **LP223439 with PCM, NTC 10 kΩ ±1% B3435 and JST ACHR-03V-S** battery pack, drawing FD_3423_10, index 1. The pack provides 250 mAh nominal capacity, 0.925 Wh at 3.7 V. The manufacturer's specifications are 125 mA charging current, 250 mA discharging current, and an operating temperature range of 10…45 °C for charging. The standard charging current is 50 mA. The specified upper adjustment limit is 4.20 V ±50 mV. Internal overvoltage protection serves as the final protection level. The maximum dimensions from the drawing are 34.5 × 41 × 3.5 mm, with a cable length of 50 ±3 mm. The battery, solar panel, and display glass are mounted separately from the components on the top side of the PCB. [LiPol Drawing](https://www.lipobattery.us/wp-content/uploads/2022/11/LiPo-Battery-Datasheet-LP223439-250mAh-with-NTC-and-JST-ACHR-03V-S.pdf) · [Supplier Configuration](https://lipo-battery.com/lipo-battery-lp223439-3-7v-250mah-0-925wh-with-pcm-wires-connector-jst-achr-03v-s-b/)

Specify the following pinout when ordering: **JST cavity 1 = GND/black, 2 = NTC/yellow, and 3 = VBAT/red**. Use the molded triangle on the plug housing to identify cavity 1. The LiPol diagram shows the color order, but does not indicate the cavity numbers. Upon receipt, measure the voltage between 3 and 1 on an unpowered mating test connector using isolated probes. A positive polarity and a value of 2.75…4.25 V is expected. Only after confirming the power pair connection is the resistance between 2 and 1 measured. At 25.0 °C, a value of 9.9…10.1 kΩ is expected, adjusted for the thermometer's accuracy. Another wiring configuration is corrected at the supplier or with a keyed adapter specified prior to assembly. At 0.2 Ω, the new battery pack impedance limit at reference temperature, a 100 mA step incurs only 20 mV within the cell. [JST housing and footprint](https://www.jst.fr/doc/jst/family/pdf/eACH.pdf)

The ABRA **SP-5V-230MA-100X80** solar panel has 5.0 V and 230 mA at the declared maximum power point, 1.15 W at STC, and dimensions of 100.8 × 82.5 × 3 mm. The product page does not specify the cold open-circuit voltage or its temperature coefficient. On receipt, mark the positive lead, measure `Voc` across the operating temperature range, record `I(V)` with an electronic load and check the model at the same irradiance. The proposed acceptance criteria for the test at 1000 W/m², 25 °C cell temperature, AM1.5 are 1.04…1.27 W and `Vmp` 4.5…5.5 V. These intervals represent proposed acceptance tolerances. Use the cold `Voc` to check the SPV1050 input limit and the voltage at each MPP pin. Mechanical prevention of connecting the battery to a PV connector or vice versa is required. External stock proof, 15 November 2026 · [Product sheet for this SKU](https://abra-electronics.com/solar-products/solar-cells-chargers-modules/0-1-watt/sp-5v-230ma-100.8x82.5mm.html) · [SPV1050 Limits](https://www.st.com/resource/en/datasheet/spv1050.pdf)

Nordic documents a limit of 100 µm between the ANT pad and its return plane for harmonic performance on bare nRF54L15-QFAA devices. The 1 mm board uses the Ebyte **E73-2G4M08S1F** module, rebranded as **E73-2G4M54LP(15)** by the manufacturer. Table 2.1 in the manual specifies the nRF54L15-QFAA. The module integrates the RF network, PCB antenna, 32 MHz and 32.768 kHz crystals, and the shield. In Figure 3.1, the dimensions are 12.0 × 17.2 × 2.4 mm, with a tolerance of ±0.2 mm for single decimal measurements. A revised marking on the shield is acceptable if the supplier correlates it with the same code. For the first batch, the subassembly drawing and label are verified; if a change in marking creates ambiguity, confirmation of QFN48 on the batch or an X-ray of the module is required. FICR.INFO.PART confirms the silicon model, and the Nordic datasheet does not publish a numeric enumeration for QFN48 under INFO.PACKAGE. Variants `(05)`, `(10)`, or `LX/FX` are not equivalent. Measure current and harmonics on the physical assembly with its battery, display glass and cables fitted. [QFN limitation explained by Nordic](https://devzone.nordicsemi.com/f/nordic-q-a/127617/nrf54l15-pcb-design-review-for-iot-device/570252) · [Naming correspondence](https://www.ebyte.com/news/4576.html) · [Ebyte manual and drawing](https://www.ebyte.com/product/2769.html)

In the radio simulation, VMAIN started at 3.3 V and the regulator supplied no current for 10 µs. The local capacitors held the rail at approximately 3.220 V with a 4.8 mA load and 2.966 V with a 20 mA stress load. The model covers local energy storage over the temperatures and pulse tails tested. It excludes the regulator control loop. During prototype tests P01 and R01, probe the C201/U201 pads at radio startup and at the end of the pulse to check the prediction.

For the antenna area, native geometry verification measures the actual perimeter of the board surface relative to the module body, traces, vias, pads, and constraints. Separate CAM verification opens original copper files using Gerbv at 10 µm/pixel. Six known-position pads must appear before a black area can be accepted as "no copper." Zero copper is verified on both faces within the antenna rectangle and ≥85% coverage is required on the ground plane beneath the shielded module. Results include hash values of measured files; the option `--receipt verification/manufacturing/gerber-vs-native.json` rejects a combination between a board and different revision plot files. Native geometry · Gerber result · Top copper detail · Bottom copper detail

For return paths sensitive to noise, inspection separately checks traces, vias, through-hole pads, and foreign polygons on the bottom layer, then measures in Gerbv the actual copper exported into four regions. Under the RT6160 loop, SPV1050, display boost, and RTC crystal, the copper photoplot coverage prior to drilling is 100%, with no other nets in native geometry. These protected rectangles do not measure the EMC impedance of the entire board.

## Checks by calculation and simulation

The independent verifier maps symbol → physical pad via the library from the native schematic file and directly reads `contactref` from the native board. It compares power supplies, both CC connections, diode orientations, cell terminals, switch ports, /CE, wake-up domains, ADC interlock, SWD, MCU, and FPC pinouts against a separately transcribed pin map. The 300 pads on 63 components are verified both in the schematic and on the board. Values of components, actual trace layers, zone areas, and via placements, native nominal stack, and component orientations are also checked. The script `python3 verification/independent/review_design.py --require-board` is executed. Negative controls from `verification/independent/negative_controls.py` temporarily invert the battery pins, reinsert a total of 8.2 MΩ divider, shift the NTC threshold, mirror the module, add an internal layer trace, and introduce a parallel path across the photovoltaic shunt R126. The verifier detects all six deviations. A pad test does not prove trace continuity or copper clearance; these require separate layout and CAM verifications. Primary sources and reproduced inventory.

| Limit | Calculated result at extremes | Implication for prototype |
|---|---|---|
| SPV1050, 6.8 MΩ / 1.5 MΩ / 2.7 MΩ, resistors ±1% and reference ±1% | Total 10.89…11.11 MΩ, within the 10…20 MΩ range of ST. EOC 4.887…5.138 V, UVP 3.150…3.294 V. | The lowest EOC, after the typical hysteresis of 1%, remains at 4.838 V, above the VINDPM max. of 4.63 V before path losses. This does not guarantee solar current under low light conditions. |
| MPPT Input | The divider sets the operating target to 80.33…80.96% of Voc. The MPP pin voltage is 21.14…21.82% of Voc measured after the input diode. | For acceptance, require **Voc ≤ 9.0 V at the panel terminals under cold conditions**; the analytical limit at the harvester input for MPP ≤ 2.1 V is 9.625 V under adverse resistor values. Verify current and power separately. |
| Select USB, 301 kΩ / 100 kΩ | Threshold 3.605…4.427 V with VREF 0.92…1.08 V, resistors ±1%, current PR1 up to ±0.1 µA. Nominal 4.01 V. | At 4.75 V, the USB source has priority. With excessive cable drop or a slowly falling USB supply, USB can remain selected below the 4.35–4.63 V VINDPM threshold. The battery can supply the shortfall. |
| Charger | RISET 8.66 kΩ ±1%, coefficient 797…975 AΩ → 91.1…113.7 mA before total limitation USB100 at 90…100 mA. In steady state, no input can accumulate current from the others. | Charge current can fall below about 100 mA while the board draws power. The pack limit is 125 mA. For this cell, 125 mA is 0.5 C and 100 mA is 0.4 C. In USB100 mode, R123 sets a calculated termination threshold of 4.26…6.64 mA. R114 sets a fast-charge timeout of 6.73…11.45 hours before dynamic timer slowdown. Dividing 250 mAh by 95 mA gives an ideal constant-current charging time of 2.63 hours. Actual charging also includes precharge, the constant-voltage phase and the application load. |
| NTC Window | Resistors and R25 ±1%; sensitivity for B3435 ±2% (design margin, not guaranteed tolerance as per pack drawing), threshold error ±20 mV at 3.20 V. Cold: **12.09…15.67 °C**; hot: **37.54…42.63 °C**. | D103 isolates the push-pull outputs. In the test, CE must be ≥1.4 V for an open or shorted NTC or an excessively high or low temperature, and ≤0.4 V when charging is allowed. Repeat the check while slowly reducing CHG_IN toward UVLO. |
| ADC 1:2 | At 4.20 V, ±1% individually on each of the two resistors results in ±42 mV before SAADC errors. | The ±100 mV bench criterion includes the ADC, filter, reference and cables. An ADC connected to BAT cannot detect the physical presence of a pack: with the pack absent, a capacitor or the charger’s detection pulses can still produce a nonzero reading. |

For the USB→solar transition, the complete and unmodified **TI TPS2116 Final 1.10 / SLVMDI4A** model was simulated in ngspice 47. The simulation started with both sources valid, then ramped USB down from 5.25 V to 3.55 V in 200 µs, with 5.011 V available at the solar input, 0.5 Ω in series on each source, 3.5 µF effective with 55 mΩ ESR at the output and load of 180 Ω. The output before transition was 5.234 V; after switching, 4.996 V. The minimum values from three runs with different step sizes and tolerances were **3.8816…3.8830 V**. Thus, the mux output drops below the VINDPM of the BQ24074. The TI model omits diode mode when VBUS and MODE reach zero. The calculation also excludes the BQ24074, battery, MPPT and display optics. Cold start remains unverified. Run tests P04 and P05 on the physical board. [TI Public Model](https://www.ti.com/lit/zip/SLVMDI4) · Circuit and script · Results, assumptions, and hashes · Simulated oscillogram data

Simulated voltages before and after transition from USB to solar

## Energy budget

The script `python3 verification/independent/energy.py` generates `energy-cases.csv` and `energy-inputs.json`. It sums the currents at VMAIN and reports battery consumption via the formula `Qbat = Qmain × 3.3/(Vbat × η)`. It adds the charge drawn by the always-on circuitry over the full day, then separately includes the assumed self-discharge of the battery. It assumes 96 daily updates, with five partial updates between each complete update. Scenario names describe the assumptions. Actual values depend on lot, temperature, area modified, and objects near the antenna.

| Scenario | Full / partial refreshes per day | Daily charge drawn by the board | Daily charge including estimated self-discharge | No solar input; derated battery capacity | Light for daily balance |
|---|---:|---:|---:|---:|---:|
| Example at room temperature | 16 / 80 | 0.540 mAh/day | 0.707 mAh/day | 283 days, at 80% useful capacity | 2.1 minutes if measuring +20 mA net |
| Test envelope | 16 / 80 | 2.045 mAh/day | 2.295 mAh/day | 71 days, at 65% useful capacity | 6.9 minutes if measured at +20 mA net |
| Cold (+5 °C, cell at 3.35 V), full refreshes only | 96 / 0 | 7.636 mAh/day | 7.886 mAh/day | 16 days, at 50% useful capacity | Charging inhibited outside the temperature window |
| Below the display threshold, cell at 3.10 V, no external sources | 0 / 0; 96 wake-ups only for measurement | 1.022 mAh/day | 1.272 mAh/day | 20 days only if effectively 25 mAh usable remain | Scenario without charging |

The room-temperature example assumes 5.6 mA during the display command, 0.42 s for partial update and 3 s for full update. The upper-bound scenario assumes 15 mA and refresh times of 1 s / 6 s. Boot time, bus latency, MCU waiting time, and factory BLE advertisement overhead, RT6160 efficiency, and useful capacity are enumerated as assumptions in JSON. At each periodicity, journal reading and transfer of two planes of 15,000 B are accounted for. For example, at 2 Mbit/s, transferring 30,000 B takes 120 ms. Explicit waits during panel initialization and shutdown add 312 ms before including boot, RTC, sensor, renderer, and journal. Thus, time outside the waveform is 0.60 s in the room temperature example and 0.70 s in the upper-bound scenario. The three BLE windows in the upper-bound scenario extend three out of the 96 wake-ups; if separate wake-ups are introduced, their costs are added. Self-discharge of 2% or 3% from 250 mAh over 30 days is a sensitivity assumption and is not derived from a battery pack measurement. The 2–7 minute light duration assumes a measured net charge current of 20 mA. These do not describe minimum illumination at the device window or the ability to power on in darkness. [Manual Waveshare V2, §6.2](https://www.waveshare.com/wiki/4.2inch_e-Paper_Module_Manual) · [Regulator RT6160A](https://www.richtek.com/assets/product_file/RT6160A/DS6160A-05.pdf)

The calendar test on the released firmware covers all 86,400 possible scheduling seconds and 96 alarm rearms with varying active durations. No software time drift accumulated relative to the 15-minute grid in these cases. Physical crystal drift still needs to be measured in W01. Firmware test results

In the case of 3.1 V firmware disables the display and BLE advertising, but the RTC still wakes up the MCU to read the sensor and battery. The open-circuit voltage alone does not determine the remaining capacity; the 25 mAh scenario is merely a sensitivity value.

In battery-only mode without sources, BQ24074 is specified at 4.3 µA typical and 6.5 µA maximum at 85 °C without load on OUT. PCF85063A consumes typically 0.22 µA at 3.3 V/25 °C with CLKOUT disabled, TPL5111 around 35 nA, and RT6160A at most 1 µA in shutdown. The current of wake-up circuits, resistors, protection circuits, PCB contamination, and cables is measured collectively at a threshold of 30 µA at 25 °C. With 96 updates per day, a journal of 512 slots with at most two rewrites per line per rotation implies an arithmetic ceiling of approximately 73 years until reaching 10,000 writes per line. Retention, actual usage duration, and abuse through rapid presses are distinct limits. At 96 updates per day, five years equates to approximately 175,320 updates; the V2 sheet indicates one million updates or five years under its conditions. LiPol indicates 500 cycles with ≥80% capacity under package test conditions. [BQ24074](https://www.ti.com/lit/ds/symlink/bq24074.pdf) · [PCF85063A](https://www.nxp.com/docs/en/data-sheet/PCF85063A.pdf) · [TPL5111](https://www.ti.com/lit/ds/symlink/tpl5111.pdf) · [Endurance RRAM Nordic](https://docs.nordicsemi.com/r/bundle/ps_nrf54l15/page/chapters/abs_max_ratings/doc/abs_max_ratings.html)

## Power states

This table gives the expected behavior for P04. For solar, `1` means an illuminated panel has charged STORE to EOC. Test a connected panel in darkness separately. A pack with tripped PCM protection cannot supply the board. Across the eight states, exercise all 24 directed transitions that change one source at a time. Also disconnect both inputs simultaneously, at cold start and during a refresh.

| USB | PV | Pack | Expected behavior |
|---:|---:|---:|---|
| 0 | 0 | 0 | All power domains shut down. The clock loses its time, but RRAM contents and the bistable display image are retained. |
| 0 | 0 | 1 | BQ powers SYS from the pack; RTC and wake-up functions remain active; the panel is updated only if the firmware allows voltage and temperature. |
| 0 | 1 | 0 | Start by charging STORE and it can power SYS. At each MPPT sampling, SPV stops transfer for 0.3…0.5 s, even under strong light. With approximately 35.5 µF effective capacitance, a 20 mA load would drop the voltage from 5.011 to 4.63 V in about 0.68 ms. Measure the autonomous recovery after each window; continuous operation of RTC/panel solely from PV is not guaranteed. |
| 0 | 1 | 1 | Once the harvester starts, PV powers SYS and surplus power charges the battery within the NTC temperature window. In shade, the battery takes over the load. |
| 1 | 0 | 0 | With the NTC open, USB powers SYS and permits SWD and basic tests without attempting to charge an absent cell. |
| 1 | 0 | 1 | USB priority, maximum 100 mA on the BQ path, plus low consumption on input components; the board's additional load can be powered by the battery. |
| 1 | 1 | 0 | USB priority. Removing the USB without pack may temporarily interrupt the RTC; the report must distinguish between reset and continuity, including the validity of the time. |
| 1 | 1 | 1 | USB priority, single charging path via BQ. PV switching may temporarily reduce the charging current. With a usable battery, the update phase and RTC must remain valid. |

## Bench tests

Use a USB-C power supply with dual orientations, bidirectional cell simulator, two isolated power sources or a programmable PV emulator, an electronic load, a power analyzer with µA range and sufficient bandwidth to capture boot peaks, oscilloscope, 10× probe with short ground lead for VMAIN, high-impedance differential probe for display voltages, logic analyzer, SWD with VTref input only, contact thermometer, camera, and a neutral target for optical comparison. A simple power supply that cannot absorb current does not replace a cell simulator for charging tests. Record the measured values for each case on the worksheet for that board. For E01, analyze the synchronized CSV file using reduce_scope.py: SI units columns `time_s,vbat_V,ibat_A,vmain_V`, positive current during discharge. The script separately integrates charge and discharge phases and rejects data gaps exceeding the configured threshold. Peak transients are measured separately using the probe's and shunt's documented bandwidth.

| ID | Setup and stimulus | Proposed criterion and evidence |
|---|---|---|
| M01 | Before applying power, inspect the board with AOI or a microscope and X-ray the WLCSP, LGA and hidden pads. Start with the panel and battery disconnected. Check FPC pin 1, polarities, the two independent CC connections, pogo access and assembled height above the PCB. | Check for solder bridges and opens on GND, signal nets and CC. Verify both CC resistors against the BOM and the 1.00 mm PCB thickness against the order tolerance. All components must be on the top side. The highest point must be ≤4.10 mm including solder, the battery plug and the portion of the PV pigtail above the PCB. Secure the pigtail to the enclosure for strain relief at J103. Include a ruler in photos of the FPC, side access, antenna area and final assembly with the battery pack and PV panel. |
| P01 | Set the cell simulator to 3.7 V with a 120 mA limit for first power-up, with USB and PV disconnected; then set the limit to 250 mA for the peak-current test. Set an alarm, give the board time to finish, then disconnect the fixture. Measure VMAIN, VAON and current for 60 s at idle. | VMAIN active 3.135…3.465 V without exceeding 3.6 V. VMAIN after shutdown ≤ 0.30 V within ≤10 s. VAON and RTC remain available. Median current at terminals ≤ 30 µA at 25 °C and separately ≤ 50 µA at 40 °C. VMAIN does not rise upon connection of a programmer with VTref as input only. No spontaneous discharge of the main domain through data lines or BUSY lines. |
| P02 | With simulated cell voltages at 2.75 / 3.0 / 3.7 / 4.15 / 4.20 V, apply USB-C in both orientations and separately test the two CC pins. Use 4.75, 5.0, and 5.5 V at the connector at 25 °C. Record the total USB current and charging current including dynamic load. | At P01/E01, R124 and R125 allow temporary insertion of a current-measurement shunt at the regulator's input/output; the configuration is restored prior to the final measurement. On the BQ branch in USB100 mode, IIN ≤ 100 mA; at the connector, IVBUS ≤ 101 mA including upstream dividers and comparator (measured independently or subtracted from measured branches). Cell charge current must be ≤125 mA. Below precharge threshold, a reduced current is applied according to RISET. In CV mode, `VBAT` maximum ≤ 4.25 V; charging ends upon dropping below threshold and can be reactivated after discharge. VMAIN and the clock survive insertion/removal as long as the battery can support the load. |
| P03 | With simulator and resistor box on the NTC input, measure the curve for USB only, solar only, and both. Sweep slowly through equivalent temperatures of 5/10/15/25/40/45/50 °C, open-circuit NTC, and short-circuited NTC. At 10 °C and 45 °C, use the NTC tolerance and device offset; perform real thermal validation on the pack within the safe operating range. | At 5, 10, 45 and 50 °C, average positive current into the cell must be ≤0.1 mA after stabilization, including with both sources present. At 25 °C charging is active. An open or shorted NTC inhibits charging. A window narrower than 10…45 °C is acceptable; charging outside those limits is not. For final qualification, report the switching temperatures on rising and falling temperature for both units. |
| P04 | Test all eight states of `(USB, PV, battery)` and the unique transitions between each state, then simultaneously apply USB+PV until both are absent. On the PV path, measure current between R126.2, `PV_IN` after D102, and R126.1, `PV_IC_IN` which powers U101.20, C103.1, and R105.1. With sources disconnected, temporarily replace R126 with a 50 mΩ Kelvin shunt or a current analyzer with equivalent insertion load. At 230 mA, the shunt introduces an additional 11.5 mV. Record the voltage drop and repeat the final transitions with a 0 Ω jumper connected. Test cold start with capacitors discharged, stepped light, rapid shadowing, peak irradiance, and current limiting, with the cell at 3.2 V and 4.1 V. At 3.7 V and 4.1 V, repeat source cut-off at the peak of the refresh cycle; at 3.2 V, firmware must inhibit the refresh. | With all sources disconnected, the rails lose power. With a usable cell, the RTC must not reset and an interrupted partial update must not be marked complete; VMAIN on refresh ≥ 3.0 V. USB valid without pack must allow bring-up on USB. With weak PV without pack, controlled shutdown is permitted; verify autonomous recovery when energy increases. USB return currents without cable and with PV covered ≤ 10 µA stabilized at 25 °C in battery states. Capture inputs, STORE, mux output, SYS_AON, VMAIN, DONE signal. |
| P05 | Record 60 s across the transition from darkness to light, including at least three 12…20 s MPPT windows. With the cell at 3.7 V and the NTC at 25 °C, measure ampere-hours over 30 min at STC and separately at 10%, 1% and partial shade. Repeat the transition test as USB drops from 5.0 to 3.5 V and then to 0. | Under STC, the proposed target is ≥20 mA net into the cell with the board idle, measured over 30 min after startup. At low irradiance, report the actual startup threshold and the light level at which charging balances consumption. Check whether a sagging USB supply retains priority over the voltage range where the charger limits current. Inputs must stay within the SPV and mux limits, startup must not oscillate continuously at STC, and the cell must stay at or below 45 °C. |
| P06 | Test charging with a cell simulator at the transition from precharge to CC, the application active and USB at 5.5 V. Run for 20 min or until thermal equilibrium. Measure the top temperatures of the charger, inductor and BME280. Use a thermocouple on the pack and a separate ambient reference. | No thermal throttling in qualified mode. The component temperature is below the datasheet limit after accounting for measurement uncertainty. Adjust ventilation/sensor position if the temperature deviation at steady state after loading exceeds 1 °C above the reference air temperature. A simple calculation `(5.5 − 3.0) × 0.1 = 0.25 W` represents the approximate power dissipation envelope for the USB100 charge branch, prior to additional load and internal current. θJA published at 44.5 °C/W depends on the TI test board. |
| W01 | Power from battery and measure RTC at each wake-up time for 24 h at 25 °C. Test 23:59→00:00, date change, one of three keys then all three, pulses of 100 and 500 ms, hold for 5 min, alarm during key press and key press during alarm. | On SCL/SDA at U403, proposed rise threshold of 30…70% ≤300 ns with bus at 100 kHz, without injection on VMAIN in standby. For RTC drift, OSCI/OSCO is not observed with probe; verification is done via alarms or temporary activation of CLKOUT, then deactivated. Alarms occur at second 00 in minutes 00, 15, 30, and 45 of the RTC calendar, either relative or UTC. A total difference of at most 5 s from reference over 24 h at 25 °C must be maintained. A press of at most 2 s before the quarter mark must cover that threshold. Each event triggers a single startup, and the clock continues after VMAIN is turned off. Holding the key or AF active does not maintain main power supply after the pulse ends. Upon artificial firmware freeze, the domain drops no later than the TPL timeout of 2 min +20%, then can be recovered by a new press. |
| W02 | Use the seven fixture contacts. After VMAIN first rises, or after VAON has been stable for at least 500 ms, close dry contact J201.6–7 and connect SWD. Read address 0x00FFC31C and require `FICR.INFO.PART = 0x00054B15`. Record VARIANT, PACKAGE and the module label. Program the image, verify readback and check reset. Release the fixture hold before measuring current. | Program each board twice and verify identical results both times. Support the PCB while contacting all pads simultaneously and check for backfeeding through VTref. If the next alarm elapsed during the session, restart the firmware via SWD or press a button to rearm it. After removing the fixture, verify return to sleep. Check the power sequence and data retention over 100 consecutive manual or RTC-triggered starts. |
| D01 | Insert the 24-pin FPC with power off and record the confirmed display revision: Waveshare 4.2″ B/W V2 / SSD1683. Probe VGH/VGL with a high-impedance probe, and monitor VMAIN, BUSY and SPI. At 25 °C display an asymmetric pattern with start/end markers and a test rectangle with white and black edges. | No X/Y inversion or polarity swap; all corners correctly aligned. During full refresh, for glass specification verification, internal contrast ≥8:1 and white reflectance ≥30% under the conditions specified in §8 of the Waveshare datasheet. For power supply, VGH positive and VGL negative during boost phase, without exceeding individual limits of the capacitors, diodes, and transistor. For FPC, SPI writing: CS active before SCK ≥60 ns, maintained after the last descending edge ≥60 ns, CS inactive between transfers ≥100 ns, MOSI setup ≥10 ns and hold ≥40 ns at 25 °C (SSD1683 Table 12-1). Check edges with a low-capacitance probe; no exceeding pin limits. BUSY complies with firmware-defined time limits; display enters sleep mode prior to shutdown. The display does not initiate panel update outside the 0…50 °C temperature range. |
| D02 | After a full update, remove SWD and monitor C304 (VCI/VDDIO) and C305 (internal VDD panel) with high-impedance inputs; **both** must remain below 0.30 V simultaneously for at least 60 s. On the next power-up, change the test zone, with a minimum **180 s between display commands** (or at the usual 15-minute alarm); repeat through five partial updates. Repeat at 15, 25, and 35 °C under fixed lighting, locked exposure, the same camera, linearized histogram, and regional marking. | The SPI capture shows the loading of old and new images prior to activation. Measure contrast on the white and black regions of the fifth partial image. The threshold is ≥6:1 at 25 ±2 °C. For contrast after a complete refresh, D01 is applied. Mean drift in stable regions must be ≤5% of the range between white and black reference, on neutral target. No global blink is visible during partial update. After the maximum number of cycles or at unconfirmed state, the next refresh is complete. This is the optical test for the refresh capability after actual power supply failure. |
| D03 | Disconnect all power sources at four points: before writing intent, during RAM transfer, during BUSY, and between the end of BUSY and journal confirmation. Corrupt the CRC of the last record at fixture and rerun with a different version of renderer. | The last old image is used only when the last state is confirmed and compatible. In uncertain cases, a full update is applied before any differential. Optically verify deterministic results at ≥10 repetitions/position with no infinite BUSY waiting. |
| S01 | Read BME280 ID, perform forced conversion and battery voltage measurement at 3.0/3.7/4.2 V in parallel with DMM. With battery removed and USB power present, compare the reported value with the BAT terminal voltage; maintain the "pack absent" condition set by operator. In a stable environment, verify temperature, relative humidity, and pressure using nearby and condensation-protected reference standards. | BME ID 0x60. Temperature error ±1 °C at 25 °C, RH ±4 points at 20–80% RH, pressure ±2 hPa under specified sensor conditions. In battery domain, measurement error ≤0.1 V versus DMM after filter stabilization. With VMAIN down, apply a 10 µs transition to the ADC control. Also test an abrupt power cut during sampling: confirm VBAT_SENSE ≤0.1 V within 10 ms and absence of phantom power on VMAIN. Use an oscilloscope probe with an input impedance of at least 10 MΩ. USB and antenna can be connected without covering the sensor opening. |
| R01 | In open position, then with the enclosure/solar panel and glass mounted, test BLE on channels 37/38/39 at 1 m from the calibrated receiver. Repeat in three orientations. The bring-up firmware emits for approximately 2 s on BTN3; repeat button presses or use the fixture firmware to count the number of events below. For spectrum testing, use a separate test mode with fixed power and channel. | Identify packets on all channels, >95% of captured events in a continuous scan window test and ≥100 events/channel. RSSI reports (distribution, not a single value) must show ≤6 dB degradation of the median in the final enclosure compared to the open-mount condition at the same orientation. Qualification of the 2.4 GHz emissions, harmonics and antenna requires an RF laboratory and jurisdiction-specific limits. A phone near the board allows only functional testing. |
| E01 | Record `Ibat(t)` with the simulated pack at 3.7 V, with USB, PV and the SWD probe disconnected, for at least 24 h. Use 15 minutes between cycles, 5 partial refreshes per 1 full refresh, and three BLE advertising events. Integrate numerically and compare mAh and Wh with the profile. Check sleep separately at 3.1 / 3.7 / 4.2 V and repeat with a single stuck button. | ≤2.5 mAh/24 h at 25 °C for measured electronic consumption, ≤30 µA in sleep at 25 °C, battery current peak <250 mA. Holding a key activates two pull-up resistors separately: approximately 28 µA through a 100 kΩ resistor and up to 60 µA through a 47 kΩ resistor. A sleep current increase of up to 100 µA is permitted under this defect condition; VMAIN must still be turned off. Record consumption for both a full and a partial update, the measured efficiency, minimum voltage and instrument loading. |

Check the sensor readings under the conditions specified in the Bosch datasheet. A BME280 mounted near an active charger or a heated solar panel can measure the local temperature of the enclosure; test P06 monitors this effect as a whole. [Bosch BME280](https://www.bosch-sensortec.com/products/environmental-sensors/humidity-sensors-bme280/)
