How to verify RF transmit power

1

Hi all, I am interested to learn if anyone has developed a low-cost way to check RF transmitter power.
I am using an old UHF power meter (previously used for testing radio microphones) and I think its working OK.
The power meter is non-radiating (there a load inside), with an analogue gauge. It only really works at SF12 and its a bit slow to stabilise.
I ‘calibrated’ the meter to +14dBm using a known end-device (RN2483), interestingly when I checked the RF output of my test gateway, it was about 3dB lower, (I guess its setup for a 3dB gain antenna)
I wonder if anyone has tried this one?:

https://www.immersionrc.com/fpv-products/rfpwrv2/

I can see the RF power changing if I manually set it in the end-device, its not quite linear but at least I can verify the 25mW level.

Cheers
A

2

Actually if you set 14 then the device should/will output 14 if good, as that is determined to allow nominal 2.15db gain for (EMEA) EIRP =16.15dbm. I find (across the range) that as the silicon is effectively ‘right by design’ as is the ref design of matching components/rf switch etc, as verified by the millions of units built, tested and certified (RED, etc.) then if there is a problem then it’s because the suppliers haven’t followed design and bodged it or a mis-build (if regular occurrence would soon get called out!), the usual culprit is the Antenna! There are lots of fakes, poor build or simply bad designs/implementations out there and variance due to that, especially if bad VSWR implementation, is usually far greater than say minor tolerance issues in matching components etc. That plus folk doing poor connector or feeder implementations - connector and Ant joins should be ‘tool tight’ not just finger tight etc. :wink:

Also if you haven’t put the device into a test mode for evaluation (as done for certification tests and device characterisation) would get the time and thermal averaging and response of a low cost rf meter likely doesn’t capture the relatively fast energy burst of the LoRa chirp and likely averages it down. Suspect you may be seeing that with the fact you only get close to decent result with SF12, where time on air typically approx 32x that for SF7. Perhaps also try using FSK in ‘short duration continuous mode’ to give meter time to respond, or use LoRa mod with longested permitted payload to extend time?

Amended to add: IIRC there some issues with early RN2483’s , with SF12? So make sure test with 83A version :wink:

3

There are low cost RF power meters that you can buy circa £20. No good for measuring power if the packets are very short air time, but its easy enough to make a LoRa packet last a second or more.

I went to the trouble of checking the power level of some LoRa modules on a lab calibrated spectum analyzer. I keep those LoRa modules for referance, so that I can check the el-cheapo RF power meter is still accurate.

4

Thanks for taking the time with these interesting comments.
I am able to set both gateway and end-device into test mode which produce a CW tone.
Interestingly the gateway I am using (Tektelic) has what looks like, a circulator inside and an internal PCB antenna as well as an N-type external antenna port. I assume this introduces some loss (in both directions?)
Noted on the RN2483, I have started using the Seeed Lora-E5, (much nicer to work with). My test setup here is conducted to avoid the antenna vagaries. Cheers Jeff

5

Agree, I’m a bit in the dark without a calibrated reference in the lab, actually I also tested, (rather call it compared) a few devices TX power output, interesting to see that my commercial field tested TX output was 2dB lower than the Seeed Lora E5, when both set to +14dBm! Thanks for your feedback!

6

I used one in one of those Barclays Eagle labs, they are usually co-operative.

Local college also has a techie lab for ‘developing businesses’.

7

Because they have so much experience of doing that, the government gives them lots of money to do it.

8

That works, interestingly (I’m not an RF/Cert tester!) I believe conducted tests are used in US (for e.g. FCC), where in EMEA its usually/often radiated test IIRC.

Perhaps at some point when you have built up some test/eval data you can report here for others to benefit? (With notes on test set up etc for repeatability/context for users? I know Stuart has done that several times in the past e.g. for ant comparisons, as have others (even if just links to their blogs etc for intrested parties to follow :+1:

10

Yes, happy to share my data, I will do a full write up in the next weeks.
Its quite a setup, including (amongst other things), RF shielded test enclosures made out of a stainless steel catering containers! and a programmable attenuator.
Basically its an end-to-end test rig, using Chirpstack4. The plan is to be able to verify the PHY and MAC performance, looking at the measurement data (rssi/snr etc) and test end-device ADR and Join behaviour etc.
I use Node-Red to drive the Seeed Lora E5 modem, and collect the NS measurement data via MQTT. Eventually a suite of automated tests.
I appreciate this discussion and your input!
Cheers