Meshtastic to connect isolated villages

I’ll look at the soft AP. It shouldn’t have done that.

Maybe I’ll calculate the % have it toggle between both signal % and Rssi. Hmmm…

Thanks for all this amazing detail!

1 Like

Antenna tuning

Here’s the result of our first attempts at tuning a ground-plane:

First thing we did was leave the vertical radial and a single side radial. With close to 80mm the resonance was at around 860Mhz. We found it strange that we had to cut the radial sizes to around 68mm which is 20mm smaller then the calculated 78mm for 915Mhz. According to this Return Loss vs VSWR table we’re at 1.4 SWR which is pretty good, am I interpreting this right?

The graph isn’t looking reallly good yet, but it was good practice to understand the relation between the radial sizes, angles and quantity.

@TitanTronics the measurements your app gives differs from the reference I used. The vertical monopole it the same, but the radials are very different.

The radials are different because you used a square plate that joins everything together by changing the conical function, the antenna needs to be redesigned slightly or their length simply needs to be changed according to what your nano VNA says :wink:
Trust your nanoVNA :grin:
It will do its job for what it was designed for

1 Like

That is great. an swr of 1.4 means about 97% of the rf energy is making it toward the radiating part of the antenna. That is about as good anyone would achieve, note that subtle changes in the temperature, humidity and nearby obstructions will cause the swr to vary slightly, so don’t be alarmed or concerned.

1 Like

@TitanTronics you’re right, the square plate measures 21mm, so taking that into account the radials should measure 67mm, which is close to what they have now. I also wasn’t measuring the vertical radial from the base, and it really is 78mm now. So everything is matching :+1:

Thanks for the tip @ac8dg!

I started testing the new v1.1.1 firmware, and the consumption in the morning now has gone down from 70% to 50%. Much better, thanks @geeksville! Also tried the router mode, but the command line doesn’t show battery information, so can’t really tell. Had forgetten to put a battery :confused:


Directional antenna

The two 20dbi Yagi antennas have arrived. It’s 2,55m long, 21 elements Yagi.

Here’s what the NanoVNA tells me about them:

SWR at 915Mhz is at 2.2, which is not good.

This antenna has it’s resonance at 832Mhz. Using a Yagi calculator set to 915Mhz, I get a Boomlength of 2.13m, it’s actually 2,55m.

The directors should have from 149mm to 136mm of length, they have all around 140mm.

I’m not sure how I should go about tuning it… Any tips?


I’m guessing an impedance of around 40Ohm gives you around 1.5 swr?! Did you connect the VNA directly to the pigtail (how long is the pigtail) from the antenna? If you take the front end off, do the measurements change? Is the dipole fixed or can you slightly change its position on the beam?
The relector at the back looks allright, is it fixed tight? At what angle? Can it be taken off and test etc.
To make production cheaper they may have made all directors the same size and added a few more to compensate. Since these are designed for 100Mhz range to resonate somewhere in the middle is expected.
My 900 Mhz yagi is also not “perfect” at 868mhz but works pretty well from the tests I did a week ago. Got a bigger model coming in this week and I am considering to put a reflector like you have onto them but getting the material cheaper then the actual antenna is a challenge lol.

1 Like

Hey @PA7John! Thanks for the info.

I’m using 10cm pigtails to connect the antenna directly to the VNA.

By taking the extender off, at 916Mhz the return loss goes to around -6dB and smith chart value goes to around ~31 ohms ~5pF (donno how to interpret these values, only the chart itself).

There’s a screw holding the dipole, so it’s easy to unscrew it and move it around, to fix it would just need to drill new holes.

The reflector is fixed pretty tight at around 45 degrees I would guess. They are easy to take off and test.

Which parts should I be tweking with in what ways to tune it?

1 Like

Really awesome bamboo casing a local artisan from my village made :slight_smile:

I’m at the indigenous territory where the implementation will happen… will be documenting offline, and posting whenever a have a chance. Thanks everyone for the help so far, which made this possible.


Ooooh, nice.

I’d totally buy half a dozen of those if they want to make more and a non awful shipping arrangement could be organsed?

My setup is with a TTGO t-beam Lora dev board. Following all the instructions I’ve tried to setup with the python API to run Wifi in AP mode with the command:

meshtastic --set wifi_ap_mode true --setstr wifi_ssid "*****" --setstr wifi_password "*****"

But I get this in return:

meshtastic --set wifi_ap_mode true --setstr wifi_ssid mywifissid --setstr wifi_password mywifipsw
File “”, line 1
meshtastic --set wifi_ap_mode true --setstr wifi_ssid mywifissid --setstr wifi_password mywifipsw
SyntaxError: invalid syntax

I read that there might be a bug on Mac OS X however I have also tried Windows 10 and Ubuntu to the same affect.


In the first day we left a node with one of the DIY ground-planes at Manuel Alves village, which is closer to town and furthest from Serra Grande. But it seems we’ll need Yagis for these ~20Km links between hills.

In the first three days we managed to get a 18Km LoRa link working between two hills using the large Yagis I bought, and from them to two distant villages: Serra Grande to Galheiros.

Galheiros was holding the meeting of all chiefs. They saw the experiment working and were thrilled. Many came asking me to visit their villages to connect them.

The meeting started warming up and we had no time to experiment for the next few days. After a few rainy days we came back to testing with two nodes in the same house, but only they would show up, not the two hill nodes.

I’ve been using BATMAN protocol for meshing with #libremesh and #librerouter, but I’m not sure how it works, I know there are originators and neighbors. I’m also not sure how I should be testing #meshtastic networks. I’ve been using mainly the Python cli. All nodes are using v1.1.6 and the two hill nodes are using router-mode which should be helping saving battery, and indeed it is, but what I noticed is that they don’t seem to wake up, but disabling doesn’t seem to help either.

Should I be using the v1.0.0 as it’s more stable? Is there a reason why messages aren’t waking up the two hills nodes?

Yagi Antennas

We started experimenting with DIY Yagis. They have meters of spare 4mm copper wire, so that’s what we have. We used this folded dipole calculator and this yagi calculator. But the dipole never matched, we had to do some guess work to get it to around 900Mhz. I’m guessing the diameter of the material used isn’t matching with the calculator, but without Internet I couldn’t really do any research. The dipole we ended up using was quite smaller then the one the calculator suggested.

Any recommendations for folded dipole and yagi calculators? Any good Android apps?

Copper will brittle, is there any good way to protect it, or any other material we could use?


If the copper wire is too brittle, you could try aluminum or brass brazing rod. It is inexpensive and easy to bend. Most welding rod has a durable coating(flux) and usually corrosion resistant.


Since you use a fixed small bandwith try an open dipole, totally oversize the wavelenght (45cm’ish) and slowly cut it down untill you get a good reading in VNA.
Build up the antenna with a few elements every time after that.

Much of it is trial and error as the materials are the “unknown” ,calculators will help but as ac8dg says if you got (alu)materials around just give it a go. Even chicken fence will work as a reflector.
If you have a stable signal that you can allign the antenna with and test the changes, even easier.
I ordered two of these
so I can do some signal testing with my antenna’s, at some point RSSI and auto reply will be in Meshtastic but not soon I think.

1 Like

Any updates on you project?

1 Like

Meshtastic Review :hammer_and_wrench:

Meshtastic was the initial focus of this whole project and we got working on setting up the nodes on the first days. As noted before, the first experiment was a big success with a 18Km link. But one of the objectives was testing the limits, to really understand how far LoRa can go with our setup.

We organized an expedition to get to the top of the highest accessible hill on the 302,550Ha territory. Wasn’t an easy one, and we almost gave up when we couldn’t find a way to the peak. But we risked climbing a crack and with some risky maneuvers got up there.

The mountaintop was filled with bees all over, they kept trying to get into our noses and ears all the time, but were nice enough not to sting, they just really wanted our sweat.

With only a couple of days left for the experiment we were able to put a link at the top of an easier to access hill, a few kilometers from a village which is close to town. We got a 43Km link, which was a good surprise. But unfortunately we ran out of time and equipment for connecting the villages in a meaningful way.

Most of the time was spent visiting different villages and talking to the people to learn about their needs and realities. Another good part was spent just living life as they do and helping with their big meetings. About 10 days out of 35 might have been spent giving small workshops on our plan for the LoRa mesh and on building antennas, and setting up nodes at strategic places.

We’ll only really know how our equipment held out after a next visit. These are my impressions now:

What worked :white_check_mark:

  • LoRa can be really useful for providing an emergency communication or delay-tolerant information network
  • Such a communication system can really help the Krahô organize themselves and even save lives in emergencies
  • The coverage simulator is great for finding where to place nodes across the territory
  • With some decent antennas LoRa can really travel far
  • Connecting the 10W panels to the USB controllers proved to be easy, and seemed they could last
  • Setting up the autonomous nodes was fun and easy to learn and execute for them
  • The knowledge and experience on working with radio and solar setups are really valuable for them

What’s being worked on :construction:

  • Ability to remotely change configurations in a node thru LoRa
  • Video tutorial in Portuguese documenting how to setup a autonomous LoRa node
  • A practical method for debugging the network, like this script for sending messages (still not clear)
  • Videos in Portuguese showing to how to update firmware using different operating systems and methods
  • Finding computers for installing or updating firmware is hard, so ways to update using a phone and Bluetooth while offline
  • The Bluetooth connection necessary for connecting the radio and phone is a barrier, and limits the number of users; a UI accessible thru the browser for everyone connected to the network is awesome

What needs working on :construction_worker_man:‍♂

  • Building antennas is really a hard art, need more learning resources and practice
  • Need to learn more about buying from Aliexpress from Brasil; everything I bought arrived well, but the moxons, which returned and I got my money back
  • Audio thru LoRa (like qmesh)
  • DIY end-user devices with audio input/output
  • (Donno if possible) Update firmware thru LoRa
  • A monitoring tool for the network

Future plans

I plan to edit and release small multimedia stories telling bits of the experiment and showing the peoples impressions thru interviews. Thru Coolab we’re looking at financing the documentation part of the experiment, but it would be good to use the video to direct people to crowd-fund-back the equipment and the trip.

During the next months we’ll be drafting a second phase for this project together with representatives from the involved communities. A consensus has yet to be reached on which strategy we’ll use for financing. There were also representatives from the Apinayê territory which have shown interest and would also like be become part of the experiment.


That is amazing. Thanks for the update.

1 Like

Seriously. That is so great! One thing I would recommend is coating your PCBs in a few layers of paint to protect those modules from moisture. For example (german link):


Luandro do you need help with Portuguese content?

1 Like

Here’s a little video I made about this project:

It’s all in Portuguese, but yall might enjoy the various images showing Meshtastic setups.