How vulnerable are our gateways really to be destroyed by e.g. nearby nodes?


#1

Recently one of our gateways was destroyed. This one was safely high at a roof far away from nodes, and probably something physical has happened to it during disassembly. I was aware that it's tricky if nodes are nearby a gateway but in the following communication with the vendor i was warned that cell phones and a lot of more things could also blow up a gateway.

It may be coincidence but I did notice that since yesterday another window-located gateway also seems to have issues with coverage. I didn't dig into it yet but I will do so in the near future.. perhaps it's something else. But it did made me think. I always considered the risk of blowing up gateways quite theoretical but it may be more realistic than that.

Everything on the rooftop seems safe to me, but obviously things like balcony mounted gateways or gateways located next to a window are very likely to meet at least cell phones nearby.

-Did anyone experience similar issues?
-Is it possible to determine useful guidelines that are still workable? It would be a pity if an unrealistic big safety margin would render 75% of possible locations unusable.
- Are there differences between gateway brands?


Affordable ESP8266 based gateway build instructions
Repeater behaviour
Placing two LoRa gateway antennas on the same roof?
Best position for a gateway
(Gonzalo Casas) #2

If you have more info, I'd love to hear about it.

I've also been on the skeptical side when it comes to hardware damage caused by radio signals, but if there's a real chance of this happening, it would be good to know with enough time in advance to plan accordingly.

For what is worth, almost all Zurich gateways are using a Wi-Fi dongle connected on the RPi, and my dev gateway (also an imst board) usually sits directly on top of my router; and we haven't had any damaged board so far.


#3

One thing I can imagine: If there is some resonance frequency that interacts with the electrical circuit of the gateway and an ongrowing oscillation starts. This oscillation probably could destroy parts of the circuit.

But I think its even more likely blowing up a gateway due to Overheating. Especially Outdoor Gateways at hot and sunny spots.


#4

The overheating theory sounds most likely to me. I know we've had a gateway go defective when it was inside an enclosure that got very hot in an environment that was already quite warm.
We're using a multitech conduit as a gateway. They allow you to measure the inside temperature by reading in a file somewhere in /sys/class/hwmon/.. I think it would make sense to keep an eye on this temperature if you can, for example by publishing it to an MQTT server somewhere hourly or something like that.

If you have a different gateway without a built-in temperature reading, perhaps you can make one using some kind of i2c or spi temperature sensor.


#5

@bertrik
Didn't know that. Thanks for that information.

Do you know if the have some mechanism implemented that shut down the gateway if it gets to hot?


(Gonzalo Casas) #6

for RPi based gateways, @tamberg published a simple script to push CPU temperature readings to ThingSpeak: https://bitbucket.org/tamberg/iotworkshop/src/tip/Bash/thingspeak-cputemp/


#7

I think wifi is no problem, since that is 2.4ghz. I was warned explicitly about the cell phones since they are near the LoRa band. In the meanwhile I figured out that the second suspected gateway was not affected. It was an issue with the node during the measurement.

Our community is quite young and the network is not used that much. This made me fear that not only the second gateway would have died but also a few others without being detected, but I'm glad I was wrong. It also means that most of these other gateways also faced cell phones at least 1 meter nearby without problems, and one of them probably even less than 50cm since it's next to a window where people walk by and it runs for over a month without issues. And I checked the sensitivity and it's still working as it should.

So then the question is; did someone's gateway got blown up?


#8

lightning?


(Arjan) #9

Here's a summary of a June 2016 discussion on Slack (which seems to have been partially purged already, due to the 10,000 messages limit for free account):

@nmap

Has anyone tried putting two LoRa mCards in a MultiTech Conduit?

@Ludo

I think we'll face the same problem as with the kerlink gateways. Both gateways will slowly break each others op amps. So I don't recommend that approach

And:

@matthijs

Anyone know what a safe distance between gateways would be? The Lorank8 manual says to keep at least 2m between node and gateway, to prevent overloading the gateway RX stages. Now I want to compare a few different gateway positions by putting two gateways nearby each others, but I'm worried that gateways will TX more loudly than nodes, so one will damage the other.

[...]

@telkamp

matthijs: i just did a quick experiment: at 3m you seem to be just safe (-14 dBm rx)
matthijs: gateway transmits at 14dBm in RX1 and 27 dBm in RX2, so quite a bit more than nodes can/do (TTN nodes never transmit at the 27dBm frequency)

[...]

@jpmeijers

telkamp: I just did the calculation for a 37dB loss at 868MHz using the Friis equation. The result I got was 1.94m, which sounds about right compared to your measurement. Of course reflections and antenna gains also have an influence. I assumed isotropic antennas as the most devices (CE certified ones like the RN2483) will take the antenna gain into account and turn their output power down to obey the regulations.

[...]

(I don't know if the above is a good summary.)

Also, @BoRRoZ once found a support article about protecting the “RF Explorer” device (unrelated to LoRaWAN), which apparently can even be damaged when not powered up:

However, other RF Explorer models have a maximum safe input power of +5dBm, which is 3mW in the RF SMA port. As an reference, that is the power captured by the RF antenna from a cell phone a distance of 10cm, or from a Microwave oven at about 1 meter distance. Not all the ovens or cell phones work the same, so take this as an indication only.

What may not be obvious to all users is this is a limit in any RF frequency band, even if the RF Explorer is powered off.

(Which makes me wonder if that device might even be destroyed when not even unboxed...?)


Four Channel RPi Gateway Board
< 200€ Plug and Play 868Mhz Gateway?
(Arjan) #10

Anyone any idea if some distance should be observed towards lightning rods, in case lightning strikes such rod?

(Actually the gateway on our office’s roof seems to have been grounded to the same solid wiring that the lightning rods are connected to. This was done by a professional engineer.)


#11

That's a thing of my concern.

Being the owner of a Lorank8 and a licensed hamradio operator I switch off and disconnect the antenna from my Lorank8 when I make 70cm transmissions (max 20W). The reason I have done that is that the 2nd harmonic of 70cm (430Mhz - 440Mhz) is right in or to close to the the 868Mhz ISM band.

I still have plans to make some measuring on the outdoor antenna when transmitting on the 70cm band (measure the dBm picked up by the antenna).

Personally I wouldn't hook up the ground of the outdoor antenna directly to the solid wiring used by the lightning rods. There is a reason why the lightning rods aren't connected to the earth wire of your power net and are built like a Faraday cage. When the outdoor antenna is connected / grounded to the solid wiring of the lightning rods, you still would have a very large surge spike on antenna outer braid of the antenna cable into your ICT environment when the lightning strikes onto your building. IMHO: you don't want to have that.


(Arjan) #12

Thanks for the notes on the lightning rods, @lex_ph2lb. See also a new topic, Gateway placement tips, that has some (different) discussion on that:


(Arjan) #13

2 posts were merged into an existing topic: Gateway placement tips


(Matthijs Kooijman) #14

I don't have any real knowledge about this, but I can imagine that grounding the antenna will increase the chance of destroying the antenna and gateway on a lightning strike, but I've heard that a nearby lightning strike is likely to kill nearby electronics anyway. On the other hand, I can imagine that connecting the antenna to the lightning rod grounding would make sure that if lightning strikes the antenna directly, most of the energy can be led away through the lightning rod's grounding cable (and stay outside of the building), instead of being led away through the gateway and its ethernet or power cables (into the building).


#15

Lightning protection for antennas is not really for protecting the antenna or equipment, but rather for the building it is on. If a lightning strikes a grounded antenna, other lightning protectors or just strikes something in the neighborhood, your gateway will be fried. But with proper grounding, your building wont catch fire :wink: .

Another reason to ground antennas is when you have weatherproofing around your antenna, e.g. a PVC tube with your antenna inside. The blowing wind creates a static buildup on this insulator which reduces the sensitivity of your antenna. By shorting this to ground you avoid this. These are so called DC-grounded antenna's.

I think the easiest way for lightning protection is putting you gateway very close to the antenna and using an isolated power supply and ethernet isolator. If lightning strikes it, your gateway will be fried (but it will be fried anyway) but the rest of the electrical and IT equipment of your building will be saved.


#16

is this ok ??


or this one ??


#17

The receiver on the Semtech based gateways is good to 6dbm or 400mW although 100mW input is the normal range. I presume the reference design has some transmit / receive switching that probably terminates the receive port during transmit to prevent overload. That still gives a risk from other signals

I would hope the receive path has a protection diode on the input to rectify a signal falling beyond about 0.3V and ensure the receiver is protected from overload. Unfortunately I don't have any gateway schematics to check

What can be done though is put a decent band pass filter on the input from the antenna. This removes transmit harmonics and protects the receiver from other signals. The preferred way to do this would be a shorted coax stub as this grounds any static charge picked up by the antenna too


#18

Some time ago I asked the same question, also with a burning house in mind.

https://www.thethingsnetwork.org/forum/t/is-extending-the-antenna-worth-the-effort-and-lightning-risk/2236

based on what I read there I think the additional risk of being hit caused by the antenna is very limited, but if lighting already decided to hit your house anyway it will do so by hitting the antenna.

That leaves the question if there is a rational reason to spend a serious amount of money on a lighting rod.


(Arjan) #19

...but if, on a high building, tall lightning rods are hit (by design, that's why they're there) then I'm somehow afraid that would damage any gateway that is also connected to the same wiring, or is very close to it... (But I'm not an expert at all.)

As gateways could fail for other reasons as well, I guess putting money into network redundancy is much better spent.


#20

It's a common misconception that lightning rods 'attract' lightning. The impedance of the air is so high a lightning only 'senses' the rod when it's a couple of meters away. This means that if lightning hits a rod it was going to hit the building anyway, but in the last few meters the rods provided a lower impedance path to ground than the building structure.

If your grounding is good, no other grounded equipment should be damaged. Many professional radio masts are grounded through the metal casing of the radio equipment. That's why they have such big grounding screws :slight_smile: This equipment does have proper surge protection though, like PIN diodes on the RF output. I don't know what surge protection the TTN gateway or the IMST gateways have, as @tkerby mentions.

@BoRRoZ no idea if these offer good protection. I'm a bit worried by not seeing any grounding pins, which any decent lightning protector should have. These look more like normal surge protectors, where the excessive charge is dissipated through the TVS diodes. But these will blow or fuse if lightning strikes.

@TijnOnlijn How are the results (range) with your antenna setup?