Dachshund70

Dachshund70 2.4 GHz dongle

The receiver for the keyboard's 2.4 GHz mode (SPEC phase 14). It is a small desk puck: a USB-C cable runs to the computer, and the puck sits on the desk near the keyboard. KiCad 10, 2 layers, 1.6 mm.

Pins come from the dongle section of boards/Dachshund70/board.json, as for the keyboard.

Board

• 40 × 24 mm, 2 mm corners. Two unplated M2 holes, both at the USB end, so no screws sit near the antenna.
• U2, Raytac MDBT50Q-P1MV2: the same nRF52840 module as the keyboard, with a certified PCB antenna. The antenna end sits 0.4 mm inside the right edge. A keep-out covers the whole strip from the antenna to the edge, across the full board width, on both layers: no copper, pour or vias there.
• J1, USB-C (HRO TYPE-C-31-M-12): the same part as the keyboard's daughterboard. It is on the left edge, at the opposite end from the antenna. USB 2.0 sink: 5.1 kΩ Rd on CC1 and CC2, no PD.
• Protection: U1 USBLC6-2SC6 ESD clamp on D+, D− and VBUS; F1 750 mA PTC on VBUS. FB1 ties the receptacle shell to GND (the daughterboard does the same).
• Power: VBUS feeds both the module's VBUS and VDDH pins (high-voltage mode). VDD (REGOUT0) is +3V3. DCCH is unconnected, so REG0 runs as an LDO. There is no 32.768 kHz crystal, so firmware uses the internal RC oscillator. Decoupling: 4.7 µF + 100 nF on VBUS/VDDH (C1, C2) and on VDD (C3, C4).
• User parts:
  • SW1 PAIR: P1.06, active low.
  • SW2 RESET: P0.18.
  • D1: RGB status LED, common anode on +3V3. Red P0.08 through 1 kΩ; green P1.09 and blue P0.12 through 330 Ω. All active low, about 1–2 mA each at standard drive.
  • The buttons sit along the top edge, with RESET on the left.
• J2, SWD: a 1×5 1.27 mm header in the keyboard's pin order: 1 VDD, 2 SWDIO, 3 SWDCLK, 4 nRESET, 5 GND. Pin 1 is the probe's voltage reference. Power the dongle from USB while debugging.
• PCA10059 compatibility: the button and LED pins match Nordic's nRF52840 Dongle. The dongle firmware can be developed on a PCA10059, and the existing nrf52840dongle/nrf52840 build target drives the same button and LEDs. Adafruit's pca10059 UF2 bootloader uses P1.06 as its DFU button and P0.12 as an LED, so it should run here unchanged: hold PAIR while plugging in to get the UF2 drive. This hasn't been tried yet.

Routing

• D+/D−: routed by hand from U1 to the module as a side-by-side pair, about 23.5 mm each, on F.Cu over the B.Cu ground plane. The pair approaches the module pads from below.
• USB-C fan-out: no vias on the data lines. The two D+ pads are joined by a short loop under the receptacle body, and the two D− pads just behind the pads. U1 is flow-through: each line runs straight under the package between its two I/O pins.
• Supplies: short straight tracks, hand-routed and locked: +3V3 from C3/C4 into VDD; VDDH down to the VBUS row from C1/C2.
• Everything else: Freerouting 2.5.0, then GND pours on both layers and stitching vias. B.Cu is almost entirely ground.
• Checks (kicad-cli): ERC 0. DRC: 0 errors, 0 unconnected items, 0 schematic parity issues. Two warnings remain: the stock receptacle footprint's silkscreen runs past the board edge, the same as on the daughterboard.
• Pin check: a script compared every module pad with board.json and followed each net to its button, LED, resistor, SWD or USB pin: 0 mismatches.
• One +3V3 track runs on B.Cu under the D+/D− pair on its way to J2. At 12 Mbit/s this doesn't matter, but a re-spin could move it.

Polling rate: 1000 Hz

The nRF52840's USB peripheral is full speed only (12 Mbit/s). A full-speed interrupt endpoint can be polled at most once per 1 ms frame (bInterval = 1), so 1000 Hz is the ceiling for this dongle, and for the keyboard over its own USB cable. SPEC already asks for a 1 ms interval.

The radio isn't the limit. An ESB packet at 2 Mbit/s with a ~16-byte report is about 100 µs on air, plus 40 µs for the fast radio ramp-up. The link could carry well over 2000 reports per second.

4000 Hz or 8000 Hz would need High-Speed USB (480 Mbit/s), which polls every 125 µs. That means a different receiver chip, such as Nordic's nRF54LM20A, which has a High-Speed USB peripheral. It's out of scope for v1, since the stack is pinned to the nRF52840.

Keeping latency low

Rough budget for one key press over the dongle, at 1000 Hz polling and with no retransmits. These are estimates to be measured in phase 14; SPEC already lists logic-analyser latency tests.

Typical total: about 1 ms, worst case about 2.5 ms without retransmits. BLE is slower: its connection interval is 7.5 ms at best, and many hosts use 11.25–15 ms.

What the hardware does for latency:

• Puck on a cable, not a stick: the puck can sit next to the keyboard, away from the computer's metal case and from USB 3 ports, which emit noise in the 2.4 GHz band. Fewer lost packets means fewer retransmits.
• Clear antenna: the antenna is at the board end with a full-width keep-out, and there is no metal near it. Use a plastic case, with no screws or labels over the antenna strip.
• USB-powered receiver: the dongle can keep its radio in receive and its crystal oscillator running all the time, with no wake-up delay.
• Host to keyboard: send lock LEDs and config data back as ESB ACK payloads, so the keyboard never needs a separate receive window.

Bring-up checks

1. Power from USB, then check the rails: VBUS ≈ 5 V, and VDD ≈ 1.8 V on a blank module (the factory UICR default). Firmware must set REGOUT0 to 3.3 V, as the keyboard's board.c does; green and blue barely light at 1.8 V.
2. Flash over SWD: either the Adafruit pca10059 UF2 bootloader (then hold PAIR while plugging in for the UF2 drive), or the dongle app directly.
3. Check that it enumerates as 1209:0002 "Dachshund70 Receiver", with both plug orientations and from a USB-C to USB-C cable (the Rd resistors).
4. Check that the LED shows red, green and blue, and that PAIR and RESET work.
5. Range and latency: measure them with the keyboard in phase 14, with the puck next to the keyboard and then plugged straight into a USB 3 port, to see the difference.

Not done

• No case design yet. The case needs a light pipe over D1, button caps over SW1 and SW2, and a USB-C cutout. The M2 holes are a placeholder.
• 3D models: the HRO receptacle is a generated model, not HRO's (hardware/3dmodels/), and D1 uses KiCad's Würth PLCC4 3.2 × 2.8 mm RGB LED model, the same package size. Swap in the manufacturers' STEPs before final case work.
• A Zephyr board definition for the dongle (with the UF2 flash layout) is phase 14 work. Until then, CI builds the dongle app for nrf52840dongle/nrf52840.
• BOM: every soldered part except the SWD header has an MPN. LCSC numbers are filled in only where they were checked (J1, U1, R1, R2).