Richard G7LEE

Richard G7LEE

2m 70cm QRP diplexer for APRS and Lora

3 minute read

I have a 2m APRS igate on my Diamond X510 2m/70cm collinear when I’m not using it, and I want to add a LORA-APRS unit to it. LoRA is low power and fits on single boards, so a two-way igate is possible. The existing 2m APRS rig uses Direwolf on a RTL-SDR dongle, so I know for a fact there won’t be any transmissions on 2m to leak into the LoRA unit. The other way, however, could be a problem, hitting the RTL-SDR1 from the LoRA 0.1W (20dBm). You can buy diplexers2, but it hardly seems worth it for a QRP signal.

G8MWR VHF and UHF Diplexer

Enter the G8MWR Homebrew VHF and UHF Diplexer - from this article derived from PW October 19883. This uses four series tuned resonant LC circuits, two to pass the antenna signal for each band and two more to ground the unwanted channel.

original G8MWR Homebrew VHF and UHF Diplexer - I omitted the 2m trap C4 and L4, see article for values

I left out the 2m series LC on the 70cm output. I would recommend G8MWR’s layout shown in the QSL article. I missed a trick here, I suffered crosstalk between the main coils because mine were parallel, so I had to make a screen. I built mine from PCB material rather than an eddy box4, and I probably should have given myself more space. G8MWR didn’t say how big his eddy box was, but I’d say he had more than twice as much real estate as I used.

my version of this. I would have had an easier time using thinner wire

I used a NanoVNA to test the through response.

through response antenna to 2m
through response antenna to 70cm

and then to investigate the bleed from the 70cm to the 2m port (with a 50Ω term on the antenna port).

70cm to 2m port
70cm to 2m port - zoomed

I had to zoom in to get more resolution for the trap. LoRA APRS is on the slightly oddball frequency of 439.9125MHz in the UK, which is why I aimed the trap at 440MHz. The spread is ~ 125kHz, so I’d say I have 45dB loss setting the LoRA signal at -25dBm at the input to the 2m RTLSDR.

In practice I could probably have saved myself the 70cm trap on the 2m output. I’d probably have got about 15dB attenuation of the 70cm signal if the notch wasn’t there, just due to the passband of the series C3 L3. I may pick up some 2m APRS packets with the notch I’d miss otherwise due to LoRA QRM.

This is not a problem at the moment because the LoRA station rarely beacons, and has never heard another station in the last week, other than my own tracker. That at least confirms it is working and has an acceptable Rx range. https://lora-aprs.live/ can take the syslog output from CA2RXU’s igate software which carries more information than is taken by APRS-IS, such as RSSI and SNR data. This would help establish if the lack of Rx is simple due to a dearth of local LoRA APRS activity other than mine. The Tx beacon range is ~32miles as received by GW4WVO-10 near Cardiff.

  1. It’s hard to find a max spec for the RTL-SDR. From the tests they did it appears they felt comfortable injecting +8dBm for the test. ↩

  2. for about £30, just as well I had all the bits for this otherwise arguably that would have been a better way to go ;) The capital cost of the trimmer caps alone would be a fair part of that. You get more for your money, too. Most commercial diplexers seem to be more ambitious in their passband filtering. I have a fixed pair of frequencies to deal with, so the notch method works well. A Diamond MX72 for £45 would do the job at transceiver powers. ↩

  3. Original October 1988 Practcal Wireless article at https://www.worldradiohistory.com ↩

  4. Generically called diecast enclosures, they used to be made by Eddystone Radio, and the Eddystone brand was bought by Hammond in 1988 ↩