# U201 and programming through J201

## Device and mechanical limits

U201 is the Chengdu Ebyte **E73-2G4M08S1F** module, JLCPCB part number **C54337734**. It includes the nRF54L15-QFAA in QFN48 package, two crystals, the RF filter, and a 2.4 GHz PCB antenna. The module solders to the station board through a 40-contact LGA footprint. Table 2.1 from the programmable component's datasheet explicitly identifies QFAA. Ebyte has renamed this variant to `E73-2G4M54LP(15)` and lists the same silicon in the new datasheet. The suffix `FX` denotes the external antenna version. At 13:47 EET on November 15, 2026, the JLCPCB service indicates 104 units available for the old MPN. Stock may be subject to change until the lot is reserved. BOM-associated datasheet · [Announcement of renaming](https://www.ebyte.com/news/4576.html) · [Current Ebyte Datasheet](https://www.ebyte.com/downpdf/2769.html) · Inventory response

Nordic recommends that the distance between the ANT pad area and the ground plane be at most 0.10 mm for QFN packages. A Nordic engineer reports FCC out-of-band harmonics when the QFN die is mounted directly on a two-layer PCB with a greater distance from the plane. The module retains the RF circuit and its geometry on its own substrate. The station PCB remains at 1 mm with two layers. The module's pinout does not expose a 50 Ω RF output and does not incorporate an antenna trace on the station PCB. Ebyte specifies an equivalent impedance of approximately 50 Ω for the antenna in the module. The enclosure size and proximity to the display require radiated emission testing on the assembled product. [Nordic recommendation](https://devzone.nordicsemi.com/f/nordic-q-a/127617/nrf54l15-pcb-design-review-for-iot-device/570252) · [Nordic reference design](https://www.nordicsemi.com/Products/nRF54L15/Reference-layout) · [Module description](https://www.ebyte.com/product/2769.html)

The Ebyte drawing measures 12.0 × 17.2 × 2.4 mm. Mechanical clearance provides ±0.2 mm for the module and 0.15 mm for the solder interface, resulting in a maximum distance of 2.75 mm from the top face of the PCB. The STEP model has nominal dimensions with its origin at the center of the body, at contact level. The antenna contour is shown for orientation. Placement maintains the radiating end directed outward, 4.65 mm beyond the edge. Metallic frame, shield, battery, solar conductors, screws, and other metallic components must leave at least 10 mm of clearance around the radiating end. This distance is a design and test criterion. STEP model · [Placement geometry](../hardware/radio-layout.json)

## Connections and footprint

VMAIN powers pin 9. C201, 100 nF, is the first decoupling capacitor, and C202, 1 µF, provides local reserve. Pads 10, 30, 31 and 37–40 connect to GND. The regulator, switching lines, and high-voltage EPD traces are placed outside the radio footprint. Maintaining ground in the digital section of the module and prohibiting copper under the antenna are distinct rules. [Block diagram configuration](../hardware/radio.json) · Ebyte Guide, §4.1 and §5

The ngspice 47 simulator used the multibranch RLC subcircuits generated by Samsung for the two exact MPNs, biased at 3.3 V at 0, 25 and 50 °C. At 25°C, their susceptance at 1 kHz corresponds to 84.30 nF and 640.73 nF in small-signal mode. Separate polarization curves, measured by Samsung with 1 V RMS, show approximately −3.92% and −27.13% at 3.3 V. The simulator starts from small-signal networks specific to voltage and temperature. The resistances include the VMAIN trace and return GND path extracted from the PCB geometry, with 2.320 mm to C201 and an additional 1.600 mm to C202. The total loop inductance varied between 0.6 and 1.5 nH/mm for unidirectional length. This is a sensitivity assumption, which can be correlated on prototype. Original Samsung models, six files · [Curves for C201](https://product.samsungsem.com/mlcc/CL05B104KO5NNN.do) · [Curves for C202](https://product.samsungsem.com/mlcc/CL05A105KA5NQN.do) · Numerical samples

The 72 simulations isolated the source after pre-charge. In this window, only C201 and C202 powered the current stage, with a 100 ns front. For tolerance sensitivity, the second series reduced capacitive elements in subcircuits by 10% while retaining original resistive and inductive values. Minimum voltages at U201 at all corners in the simulated cases for plate durations of 1, 5, and 10 µs are **3.291 / 3.259 / 3.220 V** at 4.8 mA and **3.264 / 3.130 / 2.966 V** at 20 mA. Integrated operation includes the response of RT6160, its output capacitors, and the module's internal network. Verify VMAIN on C201 with a short probe ground spring and an oscilloscope with at least 100 MHz bandwidth, including during radio and simultaneous EPD refresh. The proposed integration criterion is 3.135–3.465 V after VMAIN stabilization, for the actual station loads. The 72 results and comparison with Q/C · Script, reproducible Python, and ngspice · [Samsung Model Generator](https://weblib.samsungsem.com/mlcc/mlcc-ec.do)

In the footprint library, pin 1 is square and located on the left, immediately below the antenna tip, as viewed from above. The Ebyte diagram shows the **bottom** contacts. The control script verified the numbers in both manual editions, read the pad records from the manufacturer's Altium library, and compared the 40 center positions against the assembler's independent library. The 36 peripheral pads have the same center positions as the Ebyte library. Group 37–40 from the old Altium library was offset up to 0.311 mm from the component's explicitly stated dimensions. The delivered pad centers use the dimensioned clearances of 3.50 mm and 3.65 mm with a 1.40 mm pitch. The copper diameter of the central lands remains 1.20 mm. The maximum deviation from the independent JLC geometry is 0.056 mm. The minimum copper-to-copper distance is 0.20 mm, and the nominal mask bridge is 0.15 mm with an expansion of 0.025 mm on each edge. Numerical result · All 40 lands · [Original Ebyte library](https://www.ebyte.com/pdf-down/3930.html) · [C54337734 part's JLCEDA record](https://easyeda.com/api/products/C54337734/components?version=6.5.51)

On port P1, TWIM21 handles SDA/P1.12 and SCL/P1.11. Both positions accept clock signals. SPIM00 uses P2.01/P2.02 for SCK/MOSI. AIN0 on P1.04 reads the voltage from the divider commanded through P1.09. The versioned table explicitly maps each pad number to the physical pin on the QFN and to the carrier net. Mapping of the 17 signals · [Pinout Nordic QFAA](https://docs.nordicsemi.com/r/bundle/ps_nrf54l15/page/chapters/pin.html-qfn48) · [Dedicated Pins Nordic](https://docs.nordicsemi.com/r/bundle/ps_nrf54l15/page/chapters/pin.html-dedicated_pins)

Ebyte includes passive crystals of 32 MHz and 32.768 kHz. Both master and slave configurations, supplied by the manufacturer for distance testing, use internal capacitances of 15,000 fF for HFXO and 17,000 fF for LFXO. These values are derived from the manufacturer's **compiled** device tree. Properties and hashes of the two configurations · [Ebyte firmware archive](https://www.ebyte.com/pdf-down/3950.html)

## Power-up behaviour

Nordic 100 and 103 anomalies are relevant because the station frequently interrupts power supply. In the first 10 µs, multiple analog pins can be temporarily connected between them. After a brownout drop shorter than 50 ms, the GPIO level may remain indeterminate for approximately 10 µs. I²C pins have limited pull-ups, button inputs have series resistors, the power-off command passes through the wake-up block filter, and firmware activates battery measurement. R201 and C203 on nRESET have a nominal τ of 0.33 µs for an abrupt front and a τ of approximately 0.40 µs with additional 7 pF input capacitance. The fixture keeps the reset in high impedance until VMAIN is stable. On the prototype, DONE, VMAIN, and nRESET are monitored including during rapid reapplications. [GPIO Anomaly 100](https://docs.nordicsemi.com/r/bundle/errata_nrf54l15_rev2/page/err/nrf54l15/rev2/latest/anomaly_l15_100.html) · [GPIO Anomaly 103](https://docs.nordicsemi.com/r/bundle/errata_nrf54l15_rev2/page/err/nrf54l15/rev2/latest/anomaly_l15_103.html)

For SPI mode 0 with a prescaler greater than 2 and MSB=1, the firmware must use Nordic’s workaround for anomaly 8. Anomaly 39 relates to the order of PLL/HFXO startup. Anomaly 41 documents degradation of sensitivity on certain channels, while anomaly 47 affects debugging in System OFF. The external alarm and the fixture’s VMAIN hold allow debugging to continue after initialization. Record the SoC revision on the first unit and apply the corresponding errata for the lot. [Errata for revision 1](https://docs.nordicsemi.com/r/bundle/errata_nrf54l15_rev1/) · [Errata for revision 2](https://docs.nordicsemi.com/r/bundle/errata_nrf54l15_rev2/)

## Test fixture and first-article tests

J201 pads are 1.45 mm ENIG contacts, on a 2.54 mm grid. Looking at the top board **in the local orientation of the footprint**, with the absent position at the bottom-right, the top row is `1 GND`, `3 SWCLK`, `5 nRESET`, `7 VAON`. The bottom row is `2 VTref/VMAIN`, `4 SWDIO`, `6 WAKE_MDRV`, followed by an open position. Verify continuity with an ohmmeter before using a new fixture. The adapter cable is mirrored when viewed from that side.

| J201 | Nordic DK P19 Adapter | Role |
|---|---|---|
| 1 | 3 and 5 | Select the external target and establish the common ground |
| 2 | 1 | Measure VMAIN as reference for I/O level |
| 3 | 4 | Clock SWD |
| 4 | 2 | SWD Data |
| 5 | 10 | Reset active low, optional after power-on |
| 6, 7 | Isolated relay or switch on fixture | Bridge the two pads to keep the main power domain on |

The FICR PART field identifies the nRF54L15 device. The Nordic header for this family lists only the value `Unspecified` for the PACKAGE field. QFN48 package traceability is established from the MPN on the reel, the assembled lot, and Table Ebyte 2.1. [Nordic definitions in the provided library version](https://github.com/zephyrproject-rtos/hal_nordic/blob/7858281d843468fe53c829995fb63f45a227387a/nrfx/mdk/nrf54l15_types.h#L9586-L9615)

Power the station from its power source. Leave the P5 power output of DK disconnected and use a cable of no more than 10 cm. Close contact 6/7 after the first valid rise of VMAIN, or, if VMAIN is not monitored, after at least 500 ms of stable VAON. This delay accounts for the possible monostable pulse at power-up and for reading REXT. VMAIN must remain within the nominal window of the regulator. Set SWD to 1 MHz, read `FICR.INFO.PART` at `0x00FFC31C` and verify that it reads `0x00054B15`. Also record `INFO.VARIANT` at `0x00FFC320`, `INFO.PACKAGE` at `0x00FFC324`, the device identifier and the lot label. Write and verify the delivered image, perform a power cycle, and repeat the procedure for three cycles. For a session exceeding 15 minutes, clear the RTC alarm indicator and rearm it before opening contact 6/7. Verify return to normal cycle. [Header P19 and target power supply](https://docs.nordicsemi.com/r/bundle/ug_nrf54l15_dk/page/ug/nrf54l15_dk/program_debug/debug_out_ext_board.html) · [FICR Registers](https://docs.nordicsemi.com/r/bundle/ps_nrf54l15/page/ficr.html-topic)

At VMAIN 3.3 V, Ebyte publishes typical values of 4.8 mA during transmission at 0 dBm and 3.4 mA during reception at 1 Mbps for the module. For an isotropic receiver at 10 m, free-space calculation yields path loss of 60.06–60.34 dB across the three advertising channels. With a design margin of 12 dB for efficiency, directionality, and enclosure, the estimated received power is −72.06 to −72.34 dBm. Ebyte's cited typical sensitivity is −96 dBm. The loaded antenna, housing effects, reflections, and real receiver sensitivity are determined by the overall physical setup. Calculations at 1, 3, and 10 m · Ebyte radio specifications

**Physical tests not run.** For first-article testing, use both the exposed board and the assembly with the e-paper display, battery, solar panel and enclosure in their final positions.

1. With B3 held during power-on and the measured battery voltage within the firmware’s allowed range, verify reception of the BLE name `Astra weather` for each of 10 consecutive wake-up events. Use a receiver with continuous scanning, at a distance of 1 m, and record RSSI, orientation, voltage, and any restarts.
2. For spectral and sensitivity testing, use Nordic DTM firmware for the same module and BLE tester with calibrated OTA coupling. Measure at 2402, 2440, and 2480 MHz at 0 dBm commanded and at least 1000 packets at 1 Mbps per frequency. Prototype threshold at an equivalent level of −85 dBm allows no more than 1% error, with no link loss at 10 m in free space and with documented orientations. For current measurement, separate the EPD and CPU loads from the radio window.
3. With a setting of 0 dBm, measure the carrier offset including after cold and warm starts. Proposed internal threshold ±50 ppm from the nominal frequency. Scan the second and third harmonics in a calibrated RF chamber, using the applicable standard's limit and detector. Retain captures, distances, and uncertainty budget. Integration tests do not grant product certification.
4. Scope DONE and reset at power-up with B1/B2/B3 held separately and together, and after a short VMAIN drop. Check for no false DONE pulses, no repeated resets, stable readings after I²C bus recovery and a return at the next RTC window. Measure timings at 0, 25 and 50 °C within the ratings of the complete assembly and the battery.

Numerical verification of the footprint is repeated with `python check_radio.py --pdftotext /path/to/pdftotext --catalog /path/to/C54337734-EasyEDA.json` from `verification/radio`, with `olefile` installed. Without the `--catalog` argument, the test runs autonomously from the two PDF files, the Altium library and the numeric extract of the 40 center positions from the project. The optional `--catalog` argument reperforms and verifies the extraction from the original JLCEDA response. Verification script
