In this article I add an outdoor antenna to the RAK2245 Raspberry Pi 3B+ 915MHz gateway I created a month ago.
This article is the third in my LoRaWAN series that started here. This part of my LoRaWAN adventure required much quiet contemplation and planning, often aided by fine whiskey.
In the month that my (presently indoor) LoRaWAN gateway has been online, it has received the following number of packets from LoRaWAN sensors other than mine: ZERO. So my motivation is not to service a burgeoning local LoRaWAN community. My motivation is purely to entertain myself and see what sort of range improvement I can achieve. That said, if I increase the range, maybe I catch some new third-party traffic. Update Jan 3 2020: With the outdoor antenna I do indeed now receive traffic from at least two third-party sensors.
In creating an outdoor LoRaWAN gateway, there are two major approaches:
- Put the gateway into an IP67 enclosure and put it on the roof with the antenna directly attached to the enclosure, OR
- Put just the antenna on the roof and run coaxial cable back to an indoor gateway.
The first strategy will generally produce the best range because it minimizes the length of cabling, and hence the signal loss, between the antenna and the gateway. However, it has a multitude of disadvantages:
- cost of an IP67 enclosure.
- need to get power to the roof; uncommon for there to already be power available on a residential roof.
- have to be sure that the gateway can handle the temperature extremes.
- any maintenance on the gateway requires a trip to the roof; effort and risks to personal safety.
Conversely, the second strategy is easier to setup and easier to maintain. So long as the coaxial cable run is achievable and not too long.
As much as I would like to be a purist and seek maximum range, I think we already established that I’m not that motivated. For me, it’s always going to be the second strategy when in a hobbyist residential scenario.
- A cost of approximately 50 Aussie pesos.
- In stock.
- The biggest gain I can get that meets all of the above.
Now we get to the true horror.
My gateway is presently located in my study. My house is constructed with clay bricks and there is no cavity wall in the study that is safely accessible. Hence there is no easy or tidy way to run a new 50 Ohm cable to this room. Thus my gateway needs to go somewhere else in the house.
Or I can be creative and impure. I already have RG6 running from a wall plate in my study to the roof space. This is for my terrestrial TV. However, I don’t use the TV point in the study. This cable is chased into the brick wall and plastered over. Therefore it cannot be replaced.
Thus I had the kooky idea to re-purpose the RG6 75 Ohm cable run for my 50 Ohm antenna. I of course understand that this impedance mismatch will cost me a few dB. And it seems I’m not the only person to consider this pragmatic route: https://forums.radioreference.com/threads/lmr400-to-rg6.376919/.
Lauri-Coyote‘s comments fed in to my confirmation bias:
50 Ohms v.s 75 Ohm feed line to a 75 Ohm or 50 Ohm antenna is not going to be an earth shaking mismatch. And if your a purist, wanting that 1:1 SWR, the antenna can be tuned to the new impedance- this may mess up its radiation pattern slightly, or not be easy depending on the way the antenna is constructed. I wouldn’t get up a sweat over it, especially since a 1.5 to 1 match is only a 4-5% reflection- unless the transmitter is very very Very finicky, the world will continue to spin very nicely about its axis.
(If my opinion matters; I did/do this all the time, both professionally and as a ham, and I’ve never had any issues.
My lab had six, seven, eight ?, one thousand foot spools of Belden 8213, 75 ohm coax- these were left from the previous “administration.” The bean counters admonished me to use them up first before I order’d more……. so years later when I finally left there was still plenty of that cable. We used it for all general things RF; 50 or 75 Ohms didn’t matter. )Lauri-Coyote, forums.radioreference.com, Sep 22 2018.
So I resolved to proceed. Dang the naysayers! Onward and upwards! That is, into my roof cavity, to find my TV signal splitter.
I disconnected the study cable from the splitter and fed it out of my tiled roof.
So I have a nice RG6 cable sticking out of my roof. Now what? Tiles are not a good surface to mount on. I paid attention to my TV antenna. It’s screwed into the timber of a valley in my roof.
Monkey see monkey do; I bought a valley-mount antenna from Altronics for just $23.
And I followed the valley-mount installation instructions that a great guy named Boris blogged. Voila! Behold my outdoor antenna!
I’d like to say it all went smoothly however I inadvertently disturbed a wasp nest on my roof and received a sting for my troubles. Those wasps are now in wasp heaven.
The Nearson antenna ended up closer to my TV antenna than I would have liked, but I was limited by the length of the RG6 cable. And please give me credit for the expert bending of my galvanized steel mounting pole, achieved with steel-cap boots and some steely determination.
Per my previous article, the humble little 5cm antenna that comes with the RAK2245 was achieving a maximum range of ~150m. If I gave it every chance. This was using the RAK7204 as the transmitter (+5dBm SF7 BW125) and with the gateway and antenna indoors.
Now with the Nearson outdoor antenna, my gateway achieves a maximum range of over 700m and easily reaches at least 250m in every direction I’ve meaningfully tested. A clear improvement. However, I can’t hide my disappointment at these still-modest figures. Hmm, that RG6.
In my next article I will map the range using TTN Mapper and experiment with the RAK7204 transmit power, spreading factor and bandwidth settings.
I performed an indoor experiment of three different antenna scenarios to try to quantify where I stood with this RG6 plan and Nearson antenna:
- Humble little 5cm antenna.
- Nearson antenna connected directly to the RAK2245 using an SMA-to-N-type cable.
- Nearson antenna connected indirectly to the RAK2245 via my RG6 cable (approx. 12m), wall plate and a collection of adaptors (in other words, my intended long term solution).
Scenarios expressed in photos (the indirect scenario shown in two photos):
I used the RAK7204 to send packets (LoS from 5m) and used the TTN console to observe the RSSI at the gateway for each of the three scenarios.
|Scenario||Average RSSI (dBm)|
I was surprised that the Nearson indirect scenario was 12dB worse than the Nearson direct. With reference to the following links, I only expected to lose 2.5 dB for the 12m RG6 run.
So it seems there is another 9dB or loss due to the 75/50 Ohm mismatch, the various adaptors (N-type to F-type) and the wall plate. This was a higher price than I expected to pay, but still not a high enough price to convince me to run 50 Ohm cable just yet. Clearly though, the range could be significantly improved with better cabling.
Note however that the RG6 cable itself has essentially identical loss to LMR-200 50 Ohm cable. So the 2.5 dB lost to the 12m RG6 cable length could not be recovered by replacing it with a 50 Ohm cable run. To recover that 2.5 dB it would be necessary to put the gateway on the roof.
Lightning is a vexed subject. It seems it’s often raised on TTN but not often addressed with much clarity. As always, when there is fear to prey upon, there are people willing to sell a solution: https://store.rakwireless.com/products/lightning-arrestor.
I was not keen to pay for this kind of solution. It costs more than a quarter of what a new RAK2245 would cost me. That ratio is sadly typical of modern insurance and I hold nothing but disdain for modern insurance.
I really don’t know how serious an issue lightning is for a 900 MHz antenna. However I again went to my TV antenna for guidance; it’s not grounded. Not grounded to anything, much less my house earth. All my RG6 cables and TVs have spent 13 years connected to that antenna. If it’s good enough for TV, I reason it’s good enough for LoRaWAN, but don’t quote me on that. Make your own choices at your own risk.