Shack automation layout: what should live on the Pi and what belongs on an ESP32 or Pico?
- Replies 2
- Views 30
- Created
- Last Reply
Top Posters In This Topic
-
Mod_Chrissy 1 post
-
Mod_Nancy 1 post
A better way to browse. Learn more.
A full-screen app on your home screen with push notifications, badges and more.
I'm sketching a shack control setup and want to sanity-check how the work is divided before any boards get ordered. The goal is one small network of devices that handles rig control over CAT, antenna switching, a rotator, an SWR/power readout and an APRS/packet node, without turning the desk into an RFI generator.
The split I'm leaning toward looks like this:
Raspberry Pi (Linux side): runs Hamlib's rigctld (default TCP port 4532) and rotctld (default 4533) so logging and digital-mode software can share the rig and rotator over the network instead of fighting over a serial port. It also runs Direwolf as a soft TNC, exposing KISS over TCP (8001 by default) and AGW (8000) to whatever APRS or packet client is in use. Anything that needs a sound card, a filesystem or a network stack goes here.
Pi Pico or ESP32 (real-time side): anything that has to act predictably within milliseconds or has to be safe if the Pi crashes. That means relay drivers for the antenna switch, PTT interlocks, ADC reading for rotator position, a CW keyer, and the detector for the SWR meter. The Pi sends high-level commands such as "select antenna 3" or "rotate to 270", and the microcontroller enforces the rules.
A few specific questions where I'd like opinions:
1. Antenna switch safety. Should the microcontroller refuse to change relays while PTT is asserted? I'd want that check in hardware or firmware close to the relays, not only in a Python script on the Pi. Latching relays versus normal relays with a holding current is also open. Latching saves power and heat, but you need to know the state after a power loss.
2. Rotator position sensing. The ESP32's built-in ADC is known for non-linearity and noise, especially near the rails. Is an external ADC like an ADS1115 over I2C worth it for reading a rotator's feedback potentiometer, or is averaging on a Pico's 12-bit ADC good enough? And is emulating the Yaesu GS-232 protocol the easiest way to make the controller look like a standard rotator to rotctld?
3. CAT levels. Different rigs use different electrical interfaces: TTL-level serial, RS-232 levels, or Icom's single-wire CI-V bus. The 3.3 V logic on a Pico or ESP32 is not 5 V tolerant, so level shifting and isolation need thought. USB-serial adapters with proper isolation are tempting, but ground loops between the PC, rig and station ground are a real concern.
4. RFI discipline. Wi-Fi on an ESP32 sitting a few feet from an HF antenna feed line is one more thing that can go wrong, and so are the cheap switching supplies powering all of this. What do you do with every cable that leaves an enclosure (ferrites, bypass capacitors, shielded cable, a metal box)?
I'm also trying to decide whether a Si5351 (as a VFO, a test source or a beacon driver) belongs in the same project at all, since its square-wave output needs low-pass filtering before it gets near an antenna. Tell me where this architecture is overcomplicated or fragile.
Link to comment
https://www.hamradiobase.com/forums/topic/11163-shack-automation-layout-what-should-live-on-the-pi-and-what-belongs-on-an-esp32-or-pico/Share on other sites