(you see, i’m flooding the group with topics )
Now, i’m running a ROUTER in a superior position, about 600m ASL close to a metropole region at about 100m ASL. There must be plenty of RF noise arround and i notice that the link is not as stable as it could be.
The channel utilisation is often at above 20% which cannot be from my mesh.
As a first step to improve the situation is discovered and changed the rebroadcast mode to LOCAL_ONLY, which made a significant difference.
Next i found informations that the LoRa modules are not quite protected against overloading of their RF input for strong out of band signals. Since each input signal is analog, there must be a sampling stage at the input which will have a maximum input voltage. If that is exceeded, ugly things will happen and the communication will suffer.
Thus, i thought about building a band pass filter, in the hope to get cleaner signals and so a better performance of the overall mesh.
Quickly i noticed that such filters are meanwhile cheaply available at aliexpress: https://aliexpress.com/item/32991983002.html
I’ve ordered 3 of them, let’s see if there will be a significant difference.
There is a radio tower in less than 10 km distance, they broadcast near 100 MHz with 25 kW and near 600 MHz also with 25 kW!!!
Did someone else use such filters before?
It is not only a question, it is also a recommendation for node operators in noisy environments.
In my experience bandpass filters work only in big city arias, where is much Lora devices are present in the same freq. If you living in a country area you may skip these filters and save some money for beer.
The ones from my link above can handle 0.2 W (23 dBm). The version named “900” seems to be for 915 MHz, at least if you have a look on the spectrum scope images, you will see that the actual pass band seems to be there.
It may be worth to ask the seller what the actual frequency is.
Also consider the ones you linked have an insertion loss of obviously 2 dB, so some of you valuable transmitter power (and also input signal!) is lost. Also they seem to have just 2 MHz bandwidth, which is less than you may need, depending of your settings (long-slow, long-fast etc).
@D4rk4 Of course, if there is no disturbing signal to filter out then no filter is needed.
I still didn’t get my order and can’t say if there is a significant improvement. But i know that the input stage of a LoRa module must have a finite dynamic range and surely it is possible to overload that input stage. Usually, if you use a poor antenna in a bad location (basement, ground floor) then the module will handle the incoming spectrum. But if you are on a high mountain, using a good antenna with some gain, and have strong local broadcast stations close to you… then it could make a difference. I hope so (for me)
I’m just adding a few more notes here, also for later collection:
The receiver is basically described here: https://www.mouser.com/datasheet/2/761/DS_SX1261-2_V1.1-1307803.pdf
Page 25, chapter 4.3.
Just a low pass filter in front of the digitizer and then 80 dB dynamic range.
Sounds to me like pre-filtering can make sense in some cases.
Here are 2 images showing the stop band attenuation up to 1 GHz and the pass pand attenuation. Looks like there is an insertion loss of about 1.5 dB.
All in all it looks promising and it is time to install it in front of the node in the noisy environment.
For the SX1262, the RF Receiver of this Lora transceiver uses a differential input to suppress common-mode noise. Meanwhile, the external LC balun (in the red box) is a narrowband balun that rejects signals outside the baseband frequency.
Maybe adding another bandpass filter might not be worth it. Because on the one hand, it will increase the insertion loss, on the other hand, the performance improvement may not be very obvious.
Of course, this is just my intuition, and intuition is sometimes unreliable. Looking forward to the results of your experiments.
Yes, the insertion loss is certainly not an advantage But in many cases you can afford 1.5 dB.
For a battery/solar operated system, which is somehow floating anyway, you can interprete the RX as a black box with a coax connector having GND and Input. And that input signal has a certain spectrum, depending on location and antenna size.
It’s good to see that analog filter in front of the digitizer. It will not have a particularly high Q but in most cases this will do the job, i.e. preventing the overload of the ADC.
And as you say, it is hard to precisely determine an eventual advantage of an additional filter since we miss informations from the SX1262, isn’t it? Or is there some data we can request from the SX1262, telling us that an ADC overload occurred? That would help a lot to answer the question! If so, it would be most helpful to be able to have a look at that in debug messages.
In urban area’s, specially in the 868/915 MHz regions, GSM signals can be very strong and are often close to the LoRa frequency.
I imagine this can result in blocking and/or desensitizing of the LoRa_RX chip. I expect that adding a BPF will increase the SN ratio in such situations.
I’am interested in results of anyone who has done the test with a BandPassFilter.
You can avoid the BP filter if you use a tuned loop antenna, such as a folded dipole array or folded vertical ground plane designed for the frequency band. Folded vertical and folded dipole antennas tend to reject strong out of band signals.
Want more gain? setup a folded dipole array, like a 2 or 4 bay system.
That’s right, but this antenna has its specific radiation pattern which is not completely omnidirectional. Anyway it makes sense to do so, in many cases.
Or a bi-quad, Online Calculator .:. Bi-Quad Antenna
I do now, since 2 weks, run the above band pass filter in one of my nodes. It is difficult to say whether there is an actual improvement. Also i upgraded from FW 2.1.5 to 2.1.9 in the same time which is a BIG improvement on all of my nodes
Actually the question (whether there is an improvement) must be determined in a lab, with known artificially generated signals of various levels, frequencies and modulation injected into the signal path between two nodes, giving reproducable results…
Can you tell us more about your channel, encryption, and modem settings?
There’s a chance you are routing other people’s traffic as Meshtastic is designed to still route packets that aren’t part of its own encryption key. The Router role will also give that node priority for rebroadcasts.
)
