Record

What the 1.0 rig is, and why it is built this way.

Connections

The rig

A passive survey listener in a Hammond 1590XX aluminium case. A Seeed XIAO ESP32-S3 logs. A Heltec WiFi LoRa 32 V3 hears LoRa. A u-blox NEO-M8N class GPS, with a compass on the same module, places a row when it has a fix. A DS3231 keeps time before that fix, and takes a time from the GPS when its own clock is blank. A microSD card is the sealed archive. The Field page is the live view, and Retrieve log is how the file comes out.

The radios listen. Wi-Fi scan is passive. BLE scan asks for no scan response. LoRa is receive only. A LoRa row keeps length, frequency, spreading factor, sync, RSSI, and SNR. The payload is discarded. Until a GPS fix, the place columns stay empty. Time and the hears still record.

This case build is Ares-XX Tactical revision 1.0. An earlier breadboard used the same name and a different board. Another project reuses the title for a retro computer in a different Hammond box. This record is the 1590XX listener.

Why a carrier

Dupont jumpers shake apart in a car. The carrier is one board on half the floor of the box. Modules snap into sockets. The panel harness and the GPS lead land on screws. Male pins go on each module once, so a tired module is a pull and a swap.

The outline is 100 × 78 mm, corners chamfered 8 mm, 7672 mm². Inside the box is about 139 × 115 × 35 mm, so the board is under half the floor. The open half holds the GPS module and the antenna leads. The board is 1.6 mm and two layers. The stack of sockets plus the Heltec is about 20 mm, inside the 35 mm, when the Heltec spring is not pressed into the lid.

Four M3 holes are bare, on nylon standoffs. The board does not bond the aluminium. The antenna bulkhead shells are that bond. Switch frames need to be insulated from the panel, or the aluminium becomes part of the switch.

Why the power is a pigtail

The case is a 5 V load. A power bank or a computer feeds one USB-C port. The rig holds no cell and charges nothing. The XIAO battery pads and the Heltec battery socket stay empty.

The data pins of the XIAO are on its USB-C connector. They are not on D0 through D10. The same connector carries the CC resistors that wake a USB-C power bank. The pigtail mates the XIAO, and that jack is also the log when a computer is plugged in. The log is serial. The card stays the archive.

The panel power switch is one switch in the pigtail, between the VIN/SWA pair and SWB. SWB feeds the XIAO 5V pin and, through a through-hole ferrite, Heltec J2 pin 2. Switch open, and the pigtail is dead. Switch closed, and both boards are live.

A full USB cable on the XIAO stays live and back-feeds the Heltec. The panel switch cuts the pigtail. Once the Heltec is on the carrier, its USB stays empty. While a USB cable is in a board, that board's 5V pin stays unused.

The ferrite is only in the Heltec 5 V feed. Ground is the pour. The two 3.3 V regulators stay separate. The card, the clock, the GPS default, and the later sensors take 3.3 V from the XIAO.

Earlier ideas put cells inside the box: two protected 18650s, then one cell and a boost charger. Those were set aside when an external power bank was chosen. The box stays a 5 V load, and the charging chips stay unused.

Why the lamps are split

Six lamps, on the panel, where a closed box can still show them. Wi-Fi, BLE, LoRa, and Log are solid while that job is running. Traffic blinks on each hear. GPS stays dark until a fix, then goes solid. The 10-segment bar was removed. Its pins stay empty.

The 1 kilohm resistors and the screws are on the carrier, so the panel harness is LED wires plus one switch. Cathodes join and land on KILL. The lamp switch is the only connection from KILL to GND. Open switch, dark lamps. The radios keep running.

The sketches on the boards today leave the lamp pins alone. A wired lamp stays dark until a later sketch is asked to drive them. The XIAO onboard LED and the Heltec charge LED are outside that kill.

Why the layout is quiet

SPI stays between the card and the XIAO. The clock's I2C drops on the bottom, east of the XIAO pin row. The GPS serial rides the bottom under the Heltec. Sensor 3.3 V leaves the west edge and hops those climbs. Lamp current stays in the bottom-right, clear of the card clock. Heltec 5 V stays on top, so it crosses the clock drop on the other layer. The bottom copper is ground. Mounting holes stay bare.

Clearance target is 0.20 mm. On 12 October 2026 the checker passed: 193 pads, 152 traces, 7672 mm². The silk was checked against the copper after that. KiCad is not installed here. The SVG is the picture to review. Order copper after the Heltec row is measured and the clock silk is checked against the module.

What the sockets are for

XIAO, two 1×7. S3 now, C5 later, same outline. Wire by silk D0 through D10. The chip numbers change. The S3 sketch stays off the C5.

Heltec, two 1×18. The drawing uses 22.86 mm between the rows. Measure the real board before copper. USB to the left, IPEX to the right.

Clock, 1×6. Working order GND, VCC, SDA, SCL, SQW, 32K. SQW and 32K open. Turn the module if its silk reads the other way.

Baro and motion, two 1×4, empty. 3V3, GND, SDA, SCL, modules toward the open half. Barometer BME280 or BMP280 at 0x76. Motion LSM6DSOX at 0x6A. The clock owns 0x68, so an MPU-6050 stays off this bus. Neither sensor needs a hole. They are sockets, not a purchase, until asked.

Card latch, Hirose DM3AT-SF-PEJM5, the one surface-mount part. The card pushes toward the USB wall. Power is XIAO 3V3 only. Everything else on the board is meant to be through-hole.

GPS lead, six screws. One shunt, drawn on 3V3. TX on the screw is the module TX and it lands on XIAO D1.

Radios and the case

Antennas that must hear stay outside the aluminium. The XIAO U.FL is the only 2.4 GHz antenna. Wi-Fi and BLE take turns on it. The Heltec IPEX is LoRa only, with the antenna that shipped on that jack. The spring stays idle. The GPS patch sits face up in the middle of the lid, on its own grommet. Two whips stand vertical on the lid, one near each short end. USB leaves the near long wall, low, clear of the nuts. The box sits wide, lid up.

LoRa walks 433, 868, and 915 MHz, 125 kHz, spreading factors 7, 9, and 12, sync words 0x12 and 0x34, about 700 ms on each step. A wider signal can be missed. The megahertz on the Field page names the unmarked can.

The log columns are time, lat, lon, kind, id, name, rssi, extra. Kind is wifi, ble, or lora. A hear is written again when it is about 20 seconds old, when the signal moves about 6 dB, or when a fix has moved about 25 m.

What is still open

Measure the Heltec header row. Check the clock silk against the module. Then copper can be ordered.

The single-cable harness is not built. On the bench, each board has its own USB cable and an empty battery socket.

GPS still wants a sky view. The card still wants a confirmed seat and a FAT32 format.

A XIAO ESP32-C5, a Proxicast 8 inch U.FL to RP-SMA female bulkhead (B09GJLK7DP, pin in the nut), and a Bingfu 2.4/5/5.8 GHz RP-SMA male whip (B082SHKT3Q, hole in the plug) are ordered. The antenna in the C5 box is 2.4 GHz. 5 GHz uses the Bingfu. The Field page follows the S3 USB port until the C5 is in.

The sketch that copies the card when Retrieve log asks was compiled and not flashed. Until it is loaded, Retrieve log returns the rows this computer has already saved. A survey that ran on a power bank stays on the card until that copy works.

Revision 2.0, parked

6 GHz, the Alfa AWUS036ACS, and a Linux USB host stay out of this case. The C5 radio stops at 5885 MHz. Wi-Fi 6E starts at 5925 MHz. The cheap host, when that day comes, is the workshop PC with an EDUP MT7921AU stick, or a Raspberry Pi 4 1 GB with the same stick. A Pi 5 plus an AX210 is the thicker, costlier path and does not fit the 35 mm inside height. Prices were written down in October 2026 and will go stale. Nothing on that list is a 1.0 purchase.

Zigbee, a second LoRa radio, a camera, a microphone, and air sensors stay out. The shell is closed aluminium. The compass stays with the patch on the lid. Aluminium is a poor magnetic shield, and the compass can be calibrated in the finished box.

Files

The working copies live under field-listen on this machine. The links are the same bytes, served from the Field shelf.