Author Topic: New 5G antennas interfere with RFM69 radios (@ center freq. 915mhz)  (Read 9794 times)

akstudios

  • NewMember
  • *
  • Posts: 28
So this is an interesting discovery that can have many ramifications for us going forward or none at all, since it's just my experience.

I have around 100 nodes deployed in various rooms in a university building in Chicago since the past 2 years. All of them have the RFM69HCW radios. They are a combination of Moteinos and custom PCBs that use the same radios @ 915 MHz, sending data back to a central gateway (and to some control nodes that do some simple radiator control). Since January of this year, I noticed that almost all nodes started losing data, and just dropping out entirely. I initially thought it was my gateway, but after some tests I noticed that I was able to receive data from some nodes if I have the gateway in the same room as them, but as soon as I walk out of the room, the packets drop out entirely.

Talked to a few people around and turns out some of the RF engineers responsible for setting up wireless microphones in the university (in the 600 MHz band) had to transition out of it as well due to new FCC regulations restricting that band for TV broadcasts and very recently, 5G networks. They also experienced dropouts and weird behaviors, which they were able to get around by either changing the frequency, using updated firmware provided by manufacturer, or replacing the wireless equipment entirely.

What they say and what I've noticed coincides with the installation of a new 5G antenna on the street corner close to our building. I'm not very familiar with all the frequency bands that 5G networks use, but I do know that T-mobile is the first one to activate their low frequency 5G network very recently, about the time I started losing data.

I did some quick scans with a spectrum analyzer and it does look like there are some strong signals being emitted centered exactly at 915 MHz. My fix to this was to use the setFrequency() function in the RFM69 library and change it to 905 MHz in all nodes like so:

Code: [Select]
radio.setFrequency(905000000); //set frequency to some custom frequency
and that pretty much fixed the issue. I'm getting reliable transmissions from the nodes again, going all the way to the next building as well. I'm not sure if 905 MHz still falls in the ISM unlicensed band in the region or how it affects radio performance with the same antenna length as 915 MHz, but I just wanted to share my experience with what I've observed in my area.

Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6866
  • Country: us
    • LowPowerLab
Re: New 5G antennas interfere with RFM69 radios
« Reply #1 on: January 31, 2020, 10:02:22 AM »
AFAIK 5G is between a few GHz to upwards of tens of GHz.
But googling around seems to suggest there are sub-ghz bands that may be used.
How wideband was that 915mhz signal you observed? Was it a continuous stream?
902-928Mhz is ISM in the US so I would not expect 5G to interfere since there are many 915mhz devices.
Having said all that, 905mhz is totally fine and within the ISM band, and an easy fix. Thanks for sharing your findings and how you got around it.

TomWS

  • Hero Member
  • *****
  • Posts: 1930
Re: New 5G antennas interfere with RFM69 radios (@ center freq. 915mhz)
« Reply #2 on: February 01, 2020, 07:07:22 PM »
This is very interesting.  In one of the US allocated 5G bands (n8), the uplink operating frequencies are 880-915MHz!  Wow... how did they get that through FCC?  Oh wait, the carriers have the big bucks and FCC no longer cares about... uh, never mind.

Good catch, I may have to shift the frequency on one of my networks (either that or turn on CW at full power!)


akstudios

  • NewMember
  • *
  • Posts: 28
Re: New 5G antennas interfere with RFM69 radios (@ center freq. 915mhz)
« Reply #3 on: February 03, 2020, 12:16:10 AM »
AFAIK 5G is between a few GHz to upwards of tens of GHz.
But googling around seems to suggest there are sub-ghz bands that may be used.

Here are some very crude videos and info from T-mobile explaining it: https://www.t-mobile.com/5g
Apparently their plan is to roll out the low band 5G first, which essentially replaces some sub-GHz 3G and 4G networks (you may see "5Ge" on some phones instead of LTE at the writing of this post as a marketing stunt). They would then roll out the mid and high bands over the next few years as more antennas get installed. So sub-GHz networks are definitely taking over, and centered around exactly those frequencies that were once used by common devices like RFM69 and ZigBee radios, as well as wireless microphones and other equipment. One of the "key features" of 5G that was advertised is that it will allow IoT devices to communicate a lot easier, so maybe they want us all to use 5G instead ::)

How wideband was that 915mhz signal you observed? Was it a continuous stream?

It's centered at exactly 915MHz and is a continuous stream. Here are some pics:





There was no way any transmission was getting through unless I explicitly change the frequency on our radios to well away from that frequency. 905MHz looks like a sweet spot for my setup since it looks like they took over 868 and some more around it as well, but I suppose this might vary by region.

This is very interesting.  In one of the US allocated 5G bands (n8), the uplink operating frequencies are 880-915MHz!  Wow... how did they get that through FCC?  Oh wait, the carriers have the big bucks and FCC no longer cares about... uh, never mind.

Hah, not like they haven't been doing this for years! FCC has some docs published about "make sure your devices comply by July 2020" and such, so we know where the big bucks are actually going ;)

Good catch, I may have to shift the frequency on one of my networks (either that or turn on CW at full power!)

Thanks, also FYI - I had my HCWs at maximum power and was barely getting any coverage outside the room. I mean a very strong -30 RSSI, and then.. nothing. Those new antennas are powerful.

DootDoot

  • NewMember
  • *
  • Posts: 7
Re: New 5G antennas interfere with RFM69 radios
« Reply #4 on: May 25, 2020, 06:12:37 AM »
AFAIK 5G is between a few GHz to upwards of tens of GHz.
But googling around seems to suggest there are sub-ghz bands that may be used.
How wideband was that 915mhz signal you observed? Was it a continuous stream?
902-928Mhz is ISM in the US so I would not expect 5G to interfere since there are many 915mhz devices.
Having said all that, 905mhz is totally fine and within the ISM band, and an easy fix. Thanks for sharing your findings and how you got around it.

Is there a way to listen to the frequencies between 902-928 with the RFM69HCW to determine which frequencies are most noisy?

Something similar to this. Used to scan wifi bands.
https://forum.arduino.cc/index.php?topic=54795.0

Is there a way to do it with the RFM69 library?  I was planning to use 915mhz so it would be good to have a way to avoid a frequency thats full of interference
« Last Edit: May 26, 2020, 01:34:58 PM by Felix »

Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6866
  • Country: us
    • LowPowerLab
Re: New 5G antennas interfere with RFM69 radios
« Reply #5 on: May 26, 2020, 01:37:21 PM »
Is there a way to do it with the RFM69 library?  I was planning to use 915mhz so it would be good to have a way to avoid a frequency thats full of interference
Perhaps a frequency scanner/sweeper that outputs RSSI at each frequency. But it would probably be very slow compared to proper frequency analyzing tools.
The real way to do it is using a spectrum analyzer. Some photos of that in above posts by akstudios.
A cheap and still decent way to analyze frequencies is "software defined radio".

DootDoot

  • NewMember
  • *
  • Posts: 7
I have an RTL-SDR, but I'm just thinking about ways to help deployed units avoid congested frequencies.

My devices transmit every 60 seconds (at the moment) so perhaps if there's no ACK from 915mhz then I should do a re-transmission 20 seconds later on 910mhz. and if no ACK from 910mhz, then 20 seconds later I should do another re-transmission on 920mhz. If I get an ACK from any of them then I stick with that frequency for subsequent transmissions.

I'm concerned about staying within the permitted 902-928mhz frequency range to remain FCC compliant. I've done some tests with my RTL-SDR with the default bitrate (55555) with your RFM69 library using an RFM69HCW radio.

This has probably been discusses before but I'll share my results in case anyone is interested.
When the transmitter is positioned next to the SDR antenna (50cm) I see additional signals down at 913.85, 908.0, 907.55, 907.25 but they move around when I scroll the window. So I think they're artifacts caused by excessive RF gain in the SDR. They disappear if I reduce the gain.
If I put my transmitter 5 metres away from the SDR antenna then I don't see anything outside 914.7 - 915.3. Even with higher RF gain values. I also tested at 2m but the results were basically the same. So I think that with the 55555 settings I can probably use frequencies 903 - 927.

I wish I had slightly more professional equipment than a $30 SDR!

I changed my radios to use the HopeRF default settings and repeated the test.

Code: [Select]
    
    /* 0x03 */ { REG_BITRATEMSB, RF_BITRATEMSB_4800}, // default: 4.8 KBPS
    /* 0x04 */ { REG_BITRATELSB, RF_BITRATELSB_4800},
    /* 0x05 */ { REG_FDEVMSB, RF_FDEVMSB_5000}, // default: 5KHz, (FDEV + BitRate / 2 <= 500KHz)
    /* 0x06 */ { REG_FDEVLSB, RF_FDEVLSB_5000},

The signal has a single peak at 915. Almost all the signal is between 914.7 - 915.3. When I scroll to the left I see another weak signal at 913.6. To the right I see one at 916.4

So I think this means I can probably safely hop around the frequency range from 904 to 927. I'll need to verify the signals from all the other frequencies of course.


Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6866
  • Country: us
    • LowPowerLab
You effectively cut the bitrate by a factor of more than 10. That also means your throughput is 10 times lower.
The deviation is also cut by a factor of more than 10. So of course there will be far less harmonics.
That also means if you have a radio with a crystal that is a little off compared to others, at 900+mhz carrier with 5khz deviation it will be far easier to miss the main signal and that mote will never "find" a working frequency. So such a "lost" mote will produce noise in adjacent channels trying to find the gateway.

There are ways to get around that with features offered by the RFM chipset (frequency drift compensation, automatic frequency correction, etc). Or you could try to obtain the RFM variant with a TCXO which is more accurate and low drift. But why bother when my standard settings have worked so well since creation and you have RFM69_ATC which reduces your transmission power to only the needed levels.
« Last Edit: May 27, 2020, 09:10:40 PM by Felix »

DootDoot

  • NewMember
  • *
  • Posts: 7
wow. I assumed I was just extending the range. THANK YOU!