iiinova — where your digital self keeps evolving

23n27.bsky.social

@23n27.bsky.social

break apart to build anew • unrepentant eureka moment junkie • aspiring abstraction-leak plugger and black box elucidator n621.de / The other network: dgc.social/@23n27

Posts Replies Media

23n27.bsky.social @23n27.bsky.social · 5h

I have spent *way* too much time thinking about this blog.n621.de/2026/10/a-go...

blog.n621.de

endlos schleifen | A good(?) serial number scheme

000

23n27.bsky.social @23n27.bsky.social · 02/10/2026

Needs more pixels :(

000

23n27.bsky.social @23n27.bsky.social · 02/10/2026

:3

100

23n27.bsky.social @23n27.bsky.social · 30/09/2026

They're great! Lost the ESD pin on one of them though :(

010

23n27.bsky.social @23n27.bsky.social · 30/09/2026

Clicky!

101

23n27.bsky.social @23n27.bsky.social · 30/09/2026

A project from the bucket list: Some lab automation tooling. (Bottom side stacking connector is still in the mail so I'm bodging this with an IDC cable)

A stack of two PCBs, the top one containing 7 relay channels

View of the top board. It just feeds data and power from the bus interface (RJ45) into the stack but can also inject power into the bus and has an USB interface to optionally drive (and optionally power) it, so it can be used in a self-contained fashion too.

Relay board with 7 channels, each containing a DPDT relay that connects to a shared bus.

100

23n27.bsky.social @23n27.bsky.social · 16/06/2026

Neat! Yeah I originally wanted to build a PoE version that can just wholesale live in a little outdoor enclosure but getting the antenna signal through my window is actually much easier than a full-blown Ethernet connection

Multiband GNSS antenna mounted on my balcony with a little 3D printed shroud

20cm long FPC with 2 u.Fl connectors mounted at either end. 50 Ohm nominal impedance.

021

23n27.bsky.social @23n27.bsky.social · 14/06/2026

I decided to go with a STM32C0 instead of an FTDI because the prices are frankly ridiculous and I also needed some minor glue logic so some uC was necessary anyway. Did a test board here and it works pretty well! Can get 460800 baud without any tricks on the FW side, which is more than enough.

Breakout test board based on the previous design. It just has a USB-C port, some headers, the microcontroller, and the mezzanine connector

The same PCB, but in reality. A GNSS module with a Quectel LC99T is plugged in.

010

23n27.bsky.social @23n27.bsky.social · 14/06/2026

Got hung up on some of the mechanical design aspects because the Hammond enclosure has its screws in a very inconvenient place for making a nice front panel though.

020

23n27.bsky.social @23n27.bsky.social · 14/06/2026

It has provisions for hooking up an off-board OCXO instead (including some extra IOs for controlling it) of the TCXO in case you want to go all-in. Aux u.FL on the PHY for making use of the CLKOUT functionality. Form factor is roughly Hammond 1455C802. (so pretty smol)

120

23n27.bsky.social @23n27.bsky.social · 14/06/2026

USB->Ethernet adapter to a DP83640 which does the HW time stamping. Plus a STM32C0 as a USB<->UART bridge for the GNSS module. GNSS is pluggable so you can pick and choose between a expensive timing-grade module and something cheaper. PPS goes to the PHY for syncing the PTP clock.

120

23n27.bsky.social @23n27.bsky.social · 14/06/2026

Here's how the "PTP GM on a stick" project looks like right now. Have had this sitting around for months now, really need to push it to a fab some time.

KiCad 3D view of a PCB. USB C connector, Ethernet port, ClikMate expansion header. A mezzanine board with a u-blox LEA GNSS receiver is plugged into an Amphenol BergStak 0.8 connector.

A footprint for a TCXO (with a PCB cutout as a thermal moat) is visible in the back

331

23n27.bsky.social @23n27.bsky.social · 28/02/2026

It probes with the wrong PHY driver because both the 83640 and 83848 drivers have a phy id/mask setting of 0x20005ce0/0xfffffff0 :< But I can poke it such that the correct driver attaches and then I get /dev/ptp0

010

23n27.bsky.social @23n27.bsky.social · 28/02/2026

Maybe I should try projectposting on here? Anyway, this seems to be working, so I guess I need to do the fully integrated version with a GPS module now.

Two PCBs connected with a flat flex cable. The left one has a USB-C cable plugged in, the right one an Ethernet cable.

The left one is a breakout for the AX88772C USB->RMII chip.

The right one is a breakout for the DP83640 PTP Ethernet PHY. It has a bunch of GPIOs broken out to pin headers and has two SMA jacks (one for PPS input, one for the frequency generator output of the PHY).

A bunch of cables are soldered to the EEPROM and (invisibly) the MDIO test points on the AX88772C board and go off to the top of the picture into a basic FX2 logic analyzer.

Screenshot of dmesg:

usb 3-12.4.3: new high-speed USB device number 66 using xhci_hcd
usb 3-12.4.3: New USB device found, idVendor=0b95, idProduct=772b, bcdDevice= 0.02
usb 3-12.4.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
usb 3-12.4.3: Product: grandmaster-on-a-stick
usb 3-12.4.3: Manufacturer: isentropic.systems
usb 3-12.4.3: SerialNumber: 000000
PTP clock support registered
asix 3-12.4.3:1.0 (unnamed net_device) (uninitialized): PHY [usb-003:066:01] driver [TI DP83620 10/100 Mbps PHY] (irq=POLL)
TI DP83620 10/100 Mbps PHY usb-003:066:01: attached PHY driver (mii_bus:phy_addr=usb-003:066:01, irq=POLL)
asix 3-12.4.3:1.0 eth1: register 'asix' at usb-0000:80:14.0-12.4.3, ASIX AX88772B USB 2.0 Ethernet, 00:00:10:00:1d:02

270