LoRa: Field Testing Antennas | Hacker News Reader

LoRa: Field Testing Antennas

sparkfun.com

113 pointszdw36 comments

LoRa seems like a super power for hobbyists. I naively tried to use WiFi to network some MCUs around my place (in different buildings, some close to the mesh but a couple fairly far) and it went terribly. I wound up writing a lot of code to keep the connections alive, but after some data logging I discovered the two furthest boards were offline around 60% of the time. That’s in range to seem pretty broken to me.

After a bit of research I discovered LoRa and I’m in the process of swapping out the WiFi network. While the bandwidth isn’t great, I really don’t need it. So far I’ve found a couple test boards can communicate virtually uninterrupted over the same distance that the WiFi boards typically fail. They also consume way less energy doing so! The code footprint is smaller, the boards will survive on battery power much longer, and expanding the network will be way less hassle.

If you need high bandwidth then I guess it’s not a viable solution, but if you don’t it seems incredible to me. I’m still in mild disbelief that something so cool and useful is so cheap and easy to use.

Antenna type and design can make all the difference in the world with regards to performance. Even plain-old 2.4GHz or 5GHz WiFi can work great over large distances provided that you use a parabolic reflector, yagi, log-periodic, etc. Microwave linking is all the rage nowadays, with WISPs popping up all over the place and large commercial operators using microwave links as backhaul between internet-connected sites (usually cell tower sites). But LoRa is still really awesome for low-bandwidth data with minimal antenna considerations, and having it be such a convenient standalone package with some of the boards available is just icing on the cake.

Do you recommend any resources in particular for learning more about improving WiFi range and reliability, or should I just google some of those terms?

I briefly dug into it but had the sense that it might be a bit over my head. I’m not great with hardware — I just goof around with it a bit and make fun stuff for hobbies. Making WiFi work better would be really useful for some things, though.

Those are all directional antennas, for when your client is at a fixed predictable location.

LoRaWAN is so awesome. It's the sweet spot between Wifi and ZigBee/Z-Wave. Has phenomenal range and networking options. I've got a few devices around 1/4 mile from the base station and they're highly reliable. I wish the home automation market would have adopted it earlier. There are a few brands out there that are decent but it's unfortunate the market is so fragmented. At least with Z-Wave you're not tied to a specific vendor and interop is relatively good these days. LoRa is pretty immature in the market but, at least, it seems to be progressing.

The downside to LoRaWAN is the implementation of mesh. It, generally, requires far more planning currently - but I'll be curious to see how it evolves with more mainstream applications.

I wondered if LoRa is a little immature still, but I wasn’t sure. For how easy it is to get going with, it still requires quite a bit of playing around, non-standard libraries, etc.

I was totally content to accept that once I saw how well it solved my problem. WiFi has a clear cut path to getting what you want accomplished which seems great from the outset, but is totally useless if it’s not reliable. I’ve found that WiFi-based projects generally need to be really, really close to an access point to be properly reliable.

As another comment mentioned though there are probably plenty of ways I could mitigate these problems, so it’s as much an issue with me trying to use a technology wrong as it is a hard limitation of WiFi.

It's pretty decently deployed as far as I'm aware and I've yet to find anything that is in the same class without getting into some of the more esoteric modulation schemes(although LoRA is pretty impressive on it's own, for instance the different spread factors are orthogonal so you can have multiple devices transmitting at the same time without colliding if they use different factors).

I've found the AT modem command[3] based modules[1][2] to be fairly drop-in and usable, I was even able to pair via BT on my phone and use the BT serial TTY to send/recv without any specific software. It's pretty hard to beat ~$20 modules that you can just drop in and run over UART.

Keep in mind these are mostly for point-to-point although if I remember correctly you can also configure them for a standard LoRA deployment through that command interface as well.

[1] https://store.rakwireless.com/products/rak4200-evaluation-bo...

[2] https://www.seeedstudio.com/LoRa-E5-mini-STM32WLE5JC-p-4869....

[3] https://docs.rakwireless.com/Product-Categories/WisDuo/RAK42...

I used a pair of EZ-Bridge Ultra5 5GHz repeater dishes to traverse a 350' link from a main house to an outbuilding. The dishes are rated for a 4 mile link, but there are trees in the way so it worked out. A single linear link, so not a mesh, but FYI...

Every LoRa article I've read always does range tests but never mentions bitrate achieved at that range. I have tried to find a good idea of what the rates are, and I've seen everything from a few hundred kbps in short range to literal bits-per-second for longer range and no good guidance for what is actually typical.

I'd like to see if two LoRa modules could be used as a transparent bridge for a 9600 bps serial connection over about 5km with good line of sight with a moderate sized whip antenna. It seems like 1-2km at that rate is for sure but determining the bitrate falloff vs distance (as it automatically adjusts the rate for the signal) I haven't found a resource for.

Another comment in this thread says "Best case you're getting a few kbps,"

So yeah good for sensors

My experience is it depends. I’m getting ~50kbps at 500m over water. I’m also using it in a mesh so don’t need more than that. With a stub antenna.

They also rarely give error rate.

Having used lora my gut feeling is you’re not going to get more than 3km unless you take advantage of hills or towers. Use the keyword “fresnel zone” for more information.

Also Lora is one-directional. It’s not even half-duplex out of the box. So you either need 2 sets of radios or you need to implement half-duplex communication in software.

Tangent: U.FL connectors. Gawd, they suck. I was pleasantly surprised to see U.FL removal tool in the write up. The female connectors have at best 10 mating cycles in my experience. By using the such removal tool you may get dunno 20 if you are super careful. I killed GSM and GPS antennas with single or second try detaching.

From wikipedia article of Hirose U.FL:

Female U.FL connectors are not designed with reconnection in mind, and they are only rated for a few reconnects (approximately 30 mating cycles[5]) before replacement is needed.

I just got off of a big LoRaWAN project for remote data collection.

Antenna elevation is your friend. At long ranges, be aware of the fresnel zone!!

Does it make sense too look into it for high speed data telemetry? Like sending a reasonably sized array of floats 50 times a second, or would there not be enough throughput?

Probably not. Best case you're getting a few kbps, but generally less - especially if you want long range.

Ah shame I guess what I've heard is true about throughput. With the almost gigahertz range frequency I'd have expected a more like roughly a third of wifi speed honestly, but I guess that's not the reality.

There’s a 2.4Ghz version that does higher speeds (up to 2Mbps) at shorter distances, but can fall back on slower speeds as well. It can’t get the same distance as the sub-GHz version but it’s still a nice product.

Short answer: nope.

Long answer: https://www.thethingsnetwork.org/airtime-calculator

When I saw the ufl connector removal tool, I was thinking that it was neat. But I saw the price of 25$ for this simple piece of bent metal, I thought that someone fell on his head trying to sell this one.

Why not using a NanoVNA to measure the antenna performances and parameters? It's quite cheap (less than $50) and it will provide a more objective methodology to test an antenna.

Probably because they had fun doing it?

A VNA still doesn’t tell you range an antenna gives over another (ie gain). For that, a legit antenna test range is necessary, or a hike out in the woods works too.

I did that a while ago (with more professional tools though):

https://www.coredump.ch/2017/04/13/lorawan-868mhz-antenna-te...

https://www.coredump.ch/2017/04/30/lorawan-868mhz-antenna-te...

A 50-ohm resistor will appear to a VNA like a perfectly efficient antenna...

If you're knowingly measuring a 50 ohm resistor with a VNA you're either calibrating it, or need to do more reading about a VNA.

I know it says Field Testing, but the results aren’t very helpful to find what you’re looking for. The antenna ‘strength’ in every direction. We built a €200 setup that combines a rotating platform, a drone power meter (for 868MHz or 915 in US) hooked up to a yagi antenna, and a network testing device as a reference. At least gives you a decent impression of how well the antenna emits its supplied power, without spending >10k on professional equipment.

Fun idea, but why did they travel in different directions for each test? The environment and line of sight are everything in this type of testing, so it doesn't make sense to wander in different random directions and then try to compare numbers.

For the final test they went up in elevation and had what appears to be line of sight back to the base station. That alone was probably more impactful than any antenna change.

Just a guess, but if you know about how far your anticipating getting a signal, you can draw a circle on the map and pick the direction that will give you the best chances. They clearly state that line of sight is suggested. At one distance, there might be obstructions along the same line as a previous test.

They could also just be bored with the first path and looking for something different.

>For the final test they went up in elevation and had what appears to be line of sight back to the base station. That alone was probably more impactful than any antenna change.

For the final test, they were 6.4 miles away. If you think the elevation was the only thing that helped, you're just being obtuse. Is it deliberate? I'm asking for my friend Andy Dufresne

For those with a keen sense of hacking adventurism, picking up one of these to just drive around with to see what signals are available is fun if you're into that kind of thing. Once you find a signal that looks interesting, you can then see if you can reverse/decode the signal. There's been a few HN posts about decoding signals pulled out of the air, and I always find them interesting.

That will be pretty difficult for LoRa messages without having access to the relevant decryption keys. LoRaWAN is a pretty secure standard.

You'd be amazed at the number of unencrypted analog signals still in use. They may be boring AF, but it requires some skill to know if it is boring or not. There's a lot of existing stuff out there that hasn't been / won't be updated because it's paid for and it still works.

My one bit of advice is do it all passive. If you try to start talking back, you could find yourself in a heap of trouble regardless of how benign your intentions.

Just to be clear, you're talking about driving around with something like an RTL-SDR and listening out for any kinds of signals?

Field comparisons of antennas is tricky to do correctly- you really want to use the same location, ideally far away from human settlements (who knows how much background noise was there on a particular test day) and ideally under the same weather conditions.

Distance basically drops to zero when you lose line of sight too. You can find references to Lora being used from a balloon with 40km distance and even from orbit but you will never reach that even on a perfectly flat plain (the ground affects the signal unless each antenna is quite high off the ground).

Great stuff! Of note, I'm using LoRa (2.4Ghz SX1280) mainly for ExpressLRS (https://www.expresslrs.org/3.0/) remote aircraft control link. The antennas common there are very different; usually thin whip antennas, with or without T. Probably due to weight and cost saving. I will try those and see the difference.

Reply on news.ycombinator.com