September 2021 Station Update
Unlike summer 2020, I found significantly less time for radio projects this year. Between mentoring a team of interns at my day job and twice as many weddings as usual, I rarely had time for more than Monday net control and occasional check ins.
But with autumn on the horizon, I’m jumping back into some old and new projects that I’m excited to share.
A distributed system for working with PNWDigital.net CallWatch:
- daemon for ingesting live callwatch data into a postgres database
pnwusersrepo is automatically updated each day with active DMR users seen on the PNWDigital network (~9k)
- API server and web app, “QSO Watch” for showing who talks to who over a longer period of time (currently private alpha)
There’s still a lot more work to be done here, and I’m not sure it will ever be released. Last summer, I got kind of stuck working on the persistence layer because it required changes across the code base, but also killed performance and wasn’t fun to work on.
The W7DG-5 high level digipeater and igate had been offline since October 2020 after the Raspberry Pi 2 hosting it ended up with a fried USB controller. Last month, I finally got around to replacing it with an Raspberry Pi 4 (8GB) running raspbian OS, direwolf software model, and YAAC (APRS client).
The new interim station still needs work, but it’s now providing 2-way internet connectivity to local stations over RF and operating in both autonomous and manual mode.
- New antenna, relocate station and clean up wire mess
- 12v/5v converter for backup power
- Separate user accounts for W7DG and N7DEM to operate independently with same RF link
- Operator guide
- Link WX station via RF
Home packet station
Packet is infectious, so I’ve also been working on my own personal home station. After fighting against my Yaesu FTM-400 with 10-pin connector, I realized it made way more sense to move the Yaesu with builtin TNC+GPS into the mobile station (KF7HVM-5) and bring the Kenwood TM-V71A into the shack (KF7HVM-1).
I’m running the station from a single board computer I had laying around, the rock64, using the standard Kenwood OEM cable kit and a cheap USB sound card. So far there are a few problems:
- rock64 is not binary compatible with raspi, so I have to compile many packages myself (thanks docker build environments!)
- The Kenwood OEM Echolink cable only handles Audio In/Out, relying on the serial connection for PTT. While hamlib supports this rig, there is no CAT command I’ve found to key the data side only and revert to the previously active control side. Worse than that, keying with CAT control doesn’t open the data line, but rather the microphone, so it’s really not usable at all.
- Kenwood’s recommended approach is to put the radio into “Echolink Sysop Mode”
[PF2]while powering on. This has its own problems: 1) CAT control doesn’t work in this mode, so no remote control and 2) Audio output is always controlled by channel squelch, so using Voice Alert (PL 100Hz) is broken.
I’m still happy with the radio for packet use and will solve all of these problems by building a custom 6-pin data cable that breaks out the PTT and COS pins for attachment to GPIO of a Raspberry Pi Zero.
The complete setup with pictures will be posted after I finish the next iteration.
After being down for about 4 months (and Rick continuing to advertise it), I finally got a day to get my original RTL-SDR scanner back online at https://scanner.kf7hvm.com and functional.
There are three main issues that still need to be solved to optimize the scanner:
- Two colocated receivers that can be toggled at 50% duty cycle to avoid thermal issues that reduce performance. After running for 1 week straight, the signal becomes degraded to the point that ham2mon can’t reliable decode the FM signal.
- Increase audio level without distortion - partly due to thermal issues, partly additional tweaks are needed in ham2mon, partly these cheap SDRs just leave something to be desired.
- Reduce CPU load at capture point. Currently I’m spinning the
CPU between 200-300% on a multicore desktop workstation with
32GB of RAM to decode 4 concurrent FM signals from 1 MHz of
spectrum and while it could handle more, I suspect the constant
load is not ideal.
- Current plan is to investigate sending raw IF to cloud servers for decoding, or leverage USB over IP to virtually bring the SDR into the data center.
In the short term, I expect to solve the 2 receivers tradeoff problem in the next week!
Updating this blog has been overdue and I’m definitely happpy to be back to tinkering with radios again. While none of this work seemed to warrant its own post (yet), it is starting to pile up, so I wanted to at least get a status update out. 73.