14
\$\begingroup\$

Some friends of mine recently moved into their new house (in Belgium.) It is located on a small hill, next to a double-track, electrified railway. Some 2-6 trains pass by every hour, depending on the time of the day.

The ground level of the house is roughly on the same height as the power lines feeding the trains. I guess the horizontal distance between the router and the closest power line is some 40 meters. There are no obstacles in the way, except for an outside window and some small bushes.

My friends experience lots (anywhere between 5 and 20) of 'WiFi outages' per hour: periods of 10-60 seconds that not a single device in the house (cellphones, tablets, laptops,...) can make a connection to the wireless router. Horrible to watch series online, apparently.

The router itself is in the upper class of routers: it's a Linksys WRT of some 2 years old that delivers a clear and strong 2.4 and 5 Ghz Wifi-bubble around the house. Works perfect 80%-90% of the time.

We were wondering if the trains or, more likely, their power lines cause these WiFi outages. An engineer I know (working in a completely different field) told me something about 'power inversion' on the lines happening several times per hour, almost always right before a train passes. I am not sure I understood him correctly. :)

This issue lingers around in my head for a couple of weeks now, so I come to the esteemed panel here for answers:

  • Is there indeed some sort of 'power inversion' happening on electrified railways (in Belgium?)
  • Is it possible that the effects of this inversion reach into my friend's house and affect their router?
  • Can they shield their router in any way to prevent the Wifi-outages?
  • Is 2.4 Ghz more or less affected than 5 Ghz? (Should all their devices connect to one instead of the other?)

Thanks a lot for helping solve this riddle !

\$\endgroup\$
  • \$\begingroup\$ Which train line is running past the house? (We can look up the voltage and rolling stock details to understand the likely causes of interference.) Belgium has 3 kV DC and 25 kV AC on the high-speed lines. \$\endgroup\$ – Transistor Mar 21 at 10:39
  • \$\begingroup\$ it's a 'regular' train line: mostly passenger traffic during the day and 3-5 freight trains between 2200 and 0500h. It is by no means a High-speed line! \$\endgroup\$ – BennyM Mar 21 at 10:43
  • 7
    \$\begingroup\$ My guess would be a broken telephone (DSL) line. The vibrations from the train shake it loose enough for a connection loss and re-train cycle. This would explain your 10-60 second connection loss - and also affect wired (ethernet) connections to the router. \$\endgroup\$ – Turbo J Mar 21 at 10:54
  • 2
    \$\begingroup\$ I called my friends: wired PC's & the TV decoder have never experienced the same issues \$\endgroup\$ – BennyM Mar 21 at 11:01
  • 9
    \$\begingroup\$ Many trains offer Wi-Fi, and many of the people on board the train are also probably carrying devices that can talk to the Wi-Fi aboard the train. Having a few hundred wireless devices pass by might cause quite a lot of traffic in the channels you are using. \$\endgroup\$ – user4574 Mar 21 at 18:42

10 Answers 10

24
\$\begingroup\$

The answer is "Maybe". If you look at trains moving along such lines you often see sparking. This is going to cause RF interference. Additionally, the length of the power line itself might act as an antenna radiating the interference.

They need to do some experiments. Record the exact timing of the outages and see whether it correlates with trains. Also, have an AM radio on and see whether they can hear interference at the same time as wifi outages.

The key thing is to find out whether there is a correlation with something visible happening. If there is, contact the owners of the railways. Otherwise, it may well be interference from elsewhere.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thank you for your answer, I will have them inspect this more closely and then report back ! \$\endgroup\$ – BennyM Mar 21 at 11:05
  • 2
    \$\begingroup\$ Experiments are key here. Measurements, the more the merrier. Perhaps the BIPT (Belgisch Instituut Postdiensten en Telecommunicatie) can help as well. This is a telecom problem and if they're like their Dutch counterpart (AT, Agentschap Telecom) they'll have vans with measuring equipment. I'm not sure whether they want to help, but it won't hurt to ask. Policing the ether is their job. \$\endgroup\$ – Mast Mar 23 at 8:56
  • \$\begingroup\$ @Mast Getting BIPT involved would normally be the my first thought as well, but Belgium is in near total lockdown because of Corona. I don't think you can get BIPT to do anything at the moment, unless it is a major outage affecting a whole area. \$\endgroup\$ – Tonny Mar 23 at 15:35
  • \$\begingroup\$ @Tonny True. Unless OP (or their friend) is a HAM or with similar experience and equipment, experiments are going to be tricky at this time either way. \$\endgroup\$ – Mast Mar 23 at 17:02
12
\$\begingroup\$

Many years ago I was asked to consult on a flawed speed-controller in an European electric train system. As result, I acquired some awareness of electric train speed controller waveforms and switching.

Assume 5,000 volts switching in 1 microsecond. Switching 1,000 to 2,000 amps at 1KHz PWM pulse-width-modulated rate. The customer would not tell me the actual switching edge was 1uS, but for survival of IGBT devices with safe-operating-area, and for moderate magnetically-induced aggressors in all surrounding metals including the PCB ground planes of the controller PCB, I computed 1US and the customer volunteered "That is about right."

Simply invert 1us, and the frequency becomes 1MHz. We use this in our prediction.

The ability of this high power signal to correlate with your 2,000GHz WiFi has 1/Frequency dropoff. Thus 1GHz will be down 1/1,000 or 60dB weaker.

Thus the 5,000 volts becomes 5 volts at 1GHz, or 2.5 volts at 2GHz, which will enormously overload your WiFi.

However the regulator agencies forbid this, and require filtering on the trains to prevent such interference.

But----- train filters might break due to constant vibration.

And your house is not adjacent to the train's speed-controller power-inverter.

But:

You might determine if train "old # 4959", passing several times a night (Belgium is small area, so lots of shuttling by that one engine, each night) causes the dropouts.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ In other words you are instructing OP to become a trainspotter. (historically referring to UK citizens who literally log train engine numbers, and considered harmlessly daft). Suddenly trainspotters seem like the smart ones... \$\endgroup\$ – Harper - Reinstate Monica Mar 23 at 20:20
  • 2
    \$\begingroup\$ @davidbak - A Singaporean company did just that by doing data analysis on a similar problem: blog.data.gov.sg/… \$\endgroup\$ – slebetman Mar 24 at 11:28
11
\$\begingroup\$


First off: do really try to be 100% sure that the wifi isn't available, and not just the internet connection through that wifi. It'd be sad to figure that out later on. Smart phones do report Wifi outages a bit strangely, with connections sometimes not even being displayed, before contact with the internet has been established. It's a pretty sad but usual thing that commercial internet providers have repeated downtimes at night.


The frequency on train power lines is in the regions < 100 Hz. In fact, Belgium is a bit unique in that it even uses DC, so 0Hz, for its trains, aside from the HSL lines, which run at 50 Hz, like your normal power grid.

Your WiFi is at > 1 GHz. During normal operation, since these frequency ranges don't overlap the least, there will be no influence whatsoever.

I really don't know what this "power inversion" would be, but if it does anything with respect to RF, it would be a short impulse when something is switched.

So, the problem with the Wifi would last for a few microseconds, which means a single lost wifi packet, which is totally normal and you wouldn't even notice.

So, the only option would be if there was a lot of switching going on for 10 to 60 seconds. Indeed, modern high-speed trains do have what is called a switch-mode power supply that converts the train supply grid voltage from DC to a controlled set of sine waves by switching it off and on at a high speed. That's awesome, because it allows the train motor to be a "normal" AC motor excited by a sine of the frequency it needs to go exactly at the speed (or acceleration) the train operator wants.

Now, such constant switching does indeed always cause a wideband electromagnetic emission. That's normal, but think about this: in the trains, the Wifi works nicely. Same goes for stations. And these are far closer to the source of the interference than your wifi devices. Also, if the trains themselves were to blame, there would be no daytime/nightime difference. So we can rule the trains out.

So, what could be the case is that during daytime / nighttime, the way the train power network gets its power changes, mostly due to French and Belgian nuclear power plants being in overproduction at night, and thus, power from the regular grid becomes cheaper than what the train companies produce in their own power plants. (This is speculation.)

If that's the case, there might actually be a converter station at the end of the power lines passing by the house, which has the job of converting grid voltage to 3kV DC, or vice versa. They also might house switch-mode power supplies that cause problems. However, these should really have been tested for emissions very strictly, and it would surprise me if that was the reason. However, for verification that industrial electrical equipment doesn't cause harmful interference, there's an official body in every country. In the case of Belgium, that's the BIPT, and they offer advice and who to turn to in case you're seeing interference, which you very possibly do.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ This is definitely a very logical and complete answer. I will ask my friends to perform the 'router in open room'-test (cfr supra) first and then report back. Thank you for your time & input ! \$\endgroup\$ – BennyM Mar 21 at 11:03
  • 1
    \$\begingroup\$ As a side note, in early 90ies back at our uni, we had Internet by a direct visibility radiolink (yes, not WiFi if course) to another building some 400m away. The link was positioned at first floor window right above a tram line end station. Whenever a tram arrived, the link went down for the duration of tram stopping there. Caused no end of cursing in our MUD sessions. :) \$\endgroup\$ – Gnudiff Mar 21 at 18:31
6
\$\begingroup\$

I think you may have a misunderstanding or a mis-hearing about this "power inversion" term. "power inversion" isn't really a thing in terms of power flows reversing or doing anything strange when a train passes.

Btu it may be helpful to understand what's happening on the train. Power is supplied to it as single phase AC, something like 25 kV and several hundred amps at the European 50Hz frequency.

But in that form it isn't really useful for driving a motor (except at one fixed speed, which really doesn't suit a train!). It is rectified to DC, then chopped up into a waveform precisely suited to drive the train's motors at the right speed at any given moment.

The unit that does this is called an inverter. (sometimes a Variable Frequency Drive or VFD.

It switches at high frequencies and may indeed emit enough high frequency interference to jam a WiFi router.

If you listen to FM radio when a train passes, can you hear interference on the weaker stations?

| improve this answer | |
\$\endgroup\$
  • 2
    \$\begingroup\$ Most of Belgium appears to be 3 kV DC. That will still use choppers, however, so your answer still stands. I reckon that any interference would occur when the house is between the train and the DC sub-station so if there is an effect it could be over many kilometers or related to the high currents due to acceleration from a station stop. i.e., It won't just happen as the train passes the house. \$\endgroup\$ – Transistor Mar 21 at 13:05
  • \$\begingroup\$ @Transistor you're right of course, hence "something like" in my answer. If it's an express on a TGV) line, probably higher voltage, no? As the interference is reported up to a minute it may happen when the train is on the block passing the house. \$\endgroup\$ – Brian Drummond Mar 21 at 13:24
  • \$\begingroup\$ Some European lines also use low frequency AC, such as 16.67 Hz. This is less insane than it sounds; railways are inherently single-phase with wildly varying load; sticking them on one phase of a 3-phase supply would cause all manner of trouble. Thus some sort of 3pAC-DC-1pAC conversion (rotary or electronic) is very much warranted. \$\endgroup\$ – Harper - Reinstate Monica Mar 23 at 20:44
  • 1
    \$\begingroup\$ Apart from some Italian railways a long time ago, AC railways, whether 16.67 or 50 Hz, supply single-phase AC to the trains, using one overhead contact wire. For the sake of the supply grid, some degree of phase balancing is achieved by feeding track sections from different grid phases. \$\endgroup\$ – Michael Harvey Mar 24 at 7:54
  • \$\begingroup\$ Not forgetting the London Brighton and South Coast Railway's 6.6 kV 25 Hz single phase system installed 1909, replaced by 660v DC third rail in 1930. – \$\endgroup\$ – Michael Harvey Mar 24 at 13:27
3
\$\begingroup\$

Before you even begin to consider whether or not the train tracks have anything to do with the problems, you need to eliminate other, more common problems.

The most common is simply poor signal propagation in the house.

Consider the construction of the house:

  • Concrete and steel - blocks a lot of the signal. You'll probably need a WiFi repeater for each floor. Newer homes are often built this way.
  • Cinder blocks (or stone) and plaster - Blocks less of the signal. Older homes are often built this way, but are often more labyrinthine and so block more signal because there are more walls.
  • Wood frame - blocks less of the signal. Not often used in Europe.

Where is the WiFi router located?

  • In the basement - the signal has to go through the floor to reach the devices. The router should ideally be on the same floor as most of the devices (or at least, the most commonly used devices.)
  • In a cupboard or behind the TV - the router needs to be in the open where it can radiate its signal. On top of a cupboard is OK, inside the cupboard will block the signal. TVs (and other appliances) contain a lot of metal, and can block the signal or reflect it in directions you don't want.

It's also possible that there are simply too many WiFi routers active around the new house. Even if it's a standalone house instead of an apartment building, you can get too much interference from other routers if the houses are close together.

Finally, your router could be broken - despite being somewhat new and of a good model.

All of that is more relevant than the train track. If you position the router and the devices optimally, and still have dropouts then it might be interference from the trains.


Put your router in the open in a room. Get a couple of devices, and watch how they perform over an hour or two. If it's good, then positioning is the problem.

If it's bad, get a new router and repeat the test. If it's still bad, it might be the trains.

Watch the router when the trains go by. Does the router shut off and reboot? That would be disturbances in the AC power causing the router to reboot. A UPS for the router would help.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ This seems like solid advice: I will let them know this should be the first step to 'debugging' this issue. I will report back here. Thanks ! \$\endgroup\$ – BennyM Mar 21 at 11:02
  • \$\begingroup\$ I agree. Ambient conditions would be much more likely to knock out WiFi if it's already running at the margins. \$\endgroup\$ – Harper - Reinstate Monica Mar 23 at 20:46
2
\$\begingroup\$

Is it normal for trains in Belgium to use onboard wifi? If yes, it is possible to get some interference between wifi networks, but complete outage is less probable.

The train power system works in kHz range, the power line is probably 50Hz. I doubt it is capable of radiating something in GHz bands enough to be detectable. Sparking is not a probable source of GHz radio waves either.

It is possible that the train and the railway infrastructure have some communication in/near wifi bands. I know that such systems exist, but I am not an expert and those that I am aware of use licensed bands.

-- most probable cause --

The router is dual-band 2.4 / 5 GHz device. If BOTH bands disappear together, it is not an interference, but the router rebooting itself for one reason or another (overheating, out of memory, kernel crash, etc...).

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ A very good point. There are many channels in each band, and WiFi routers do a dance to try to place themselves in a frequency not used by neighbors. Suddenly a train shows up with many frequencies in use both by on-train WiFi and many passenger mobile hotspots... Can your friend go to 5 GHz? That helps. \$\endgroup\$ – Harper - Reinstate Monica Mar 23 at 20:50
2
\$\begingroup\$

@TurboJ had what I suspect is the right answer in the comments, but I will elaborate a little.

The train or the cables powering it may cause radio frequency interference (RFI), either where the train's pantograph forms an imperfect connection with the line (causing arcing), or because equipment on the train (the inverter mentioned, or the motor) itself emits RFI.

But the problem may not be the WiFi but the DSL connection feeding the WiFi. The DSL connection is likely to be provided over a long copper line. It's meant to be twisted pair. However, often the DSL connection can work with impairments in that arrangement: it often will even work with one half of the pair disconnected, though your landline probably won't. But those impairments will cause it to be even more vulnerable to RFI - for instance, if only one half of the pair is there, you effectively have a long aerial. If half of the pair is slightly shorted to ground, you have something slightly like a long aerial. The connection will be unreliable, and possibly has nothing to do with the train's RFI at all (@TurboJ suggested a physical explanation).

Look at the logs of the router (or look at the DSL carrier light on it - often called something unhelpful like "internet") and see whether that goes off during these periods. Alternatively use a wired connection and see if that survives. If a wired connection survives (or the DSL carrier light doesn't go off), then it might well be the WiFi. Otherwise, you're hunting in the wrong place.

You might also ask neighbours whether they have problems. If they don't, again that points to your DSL.

| improve this answer | |
\$\endgroup\$
0
\$\begingroup\$

My suggestion would be this.

  1. On a computer or smartphone that has a WiFi radio, install a WiFi signal analyzer program that measures and shows the power level (signal strength) for each WiFi channel at the computer's location.

  2. When a train IS NOT passing by, use the WiFi analyzer software to measure and record the power level of each WiFi channel.

  3. When a train IS passing by, use the WiFi analyzer software to measure and record the power level of each WiFi channel.

  4. Compare the measured WiFi channel power levels for the "no train present" version "train present" cases.

If a train provides WiFi access points for its passengers to use, chances those access points are using specific channels. The WiFi analyzer program on your computer will see a spike in the power levels for those specific channels when a train passes by. Once you know which WiFi channels the train is using, then hopefully your friends can reconfigure their Linksys router to use different WiFi channels from those the trains use.

See also the figure titled "Non-Overlapping Channels for 2.4 GHz WLAN: 802.11b (DSSS) channel width 22 MHz" in the Wikipedia article List of WLAN Channels. The goal is to select channels that minimize interference with the WiFi channels that are used on the trains.

Of course it could be that your friend's Linksys router is malfunctioning. I had a "name brand" WiFi router that was less than two years old that exhibited exactly the same failure mode that you described: it worked fine most of the time, but every so often the WiFi would drop out for a few minutes for no apparent reason, and this happened multiple times throughout the day. I used a WiFi signal analyzer program on my laptop computer to continuously monitor and log the router's signal strength, and when I analyzed the logged data I noticed that the router's signal strength occasionally dropped to zero for a few minutes and then it would pop back up to full strength. I ended up replacing that router with a new one which solved the problem for me. (NB: The recommended way to verify this would be to take the router to a location away from the train tracks, plug it in, and monitor the router's signal strength with the WiFi signal analyzer software on your computer or smartphone.)

If these suggestions don't work, try this: buy a truckload of industrial-grade aluminum foil and a palette of aluminum foil HVAC tape. Wrap the house's entire exterior in aluminum foil and use the aluminum tape to seal closed the seams along the edges of the foil sheets to reduce RF leaks. Bond the aluminum foil wrapper to a copper grounding rod that is driven about 2 meters into the Earth. The house is now (mostly) enclosed within a Faraday cage. That ought to fix it. /s ;-)

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Yeah, it's rare for me to get more than 2 years out of routers. \$\endgroup\$ – Harper - Reinstate Monica Mar 23 at 20:53
0
\$\begingroup\$

If your friend has problems 5-20 times per hour, and only 6 trains pass by in that period, that seems to rule out interference caused by the trains themselves, i.e. wifi on the train, interference caused by the power electronics on the train.

So the interference is caused by railside equipment, or by trains further away broadcasting interference via the overhead power lines (e.g. power spikes when trains cross a boundary into the line segment that passes by the house).

Still, it'd be good to check if there's a correlation between the times of service interruption and trains passing by.

| improve this answer | |
\$\endgroup\$
-7
\$\begingroup\$

The answer is YES.

But to cause such an issue, you must be very close to the electric grid which is running very high current. Maybe 10 KVA or more. As you state its 40 Meters, (~120 ft, it is a bit difficult, but possible) it is quite possible if the current passed is much higher than 10KVA.

Even during thunder storms, when a thunder falls close to your home, the first thing which takes a hit is your router and network switches.

The best i can suggest is to go to your router settings and reduce the signal strength from strong to the least possible setting. that should help to some extent. If you are using a very powerful router, it might not help. Reason being, these routers are powerful because they emit a strong signal. despite your changes, their signal reaches the area where there is some electrical field interference.

Remember, the signal should be strong enough to just cover the home and shouldn't reach the area, where there is a presence of any electrical field.

Answering Comments

In short, The router signal gets short-circuited by the electrical field and this causes the issue. It's very similar to the short-circuiting happening when you drop a plugged-in electrical device into water. The board gets burned and the device becomes useless.

However, the router is transmitting wireless signals, but the wireless signals are getting short-circuited. This puts pressure on the router and hence the outage.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Interesting... but are you sure weakening the victim(router) further helps? This seems counterintuitive... \$\endgroup\$ – beccaboo Mar 21 at 9:08
  • \$\begingroup\$ I didn't downvote! \$\endgroup\$ – beccaboo Mar 21 at 9:09
  • 8
    \$\begingroup\$ no, that's not how the physics works at all. \$\endgroup\$ – Marcus Müller Mar 21 at 9:19
  • 1
    \$\begingroup\$ It's also not how communication networks like Wifi work; the fact that somewhere else, a packet is overlayed with noise (not "short-circuited", that's not even close to what happens) has no influence whatsoever on the router. If the router could know what happens somewhere else, when we wouldn't need to have the router, because we could transfer information without using energy. \$\endgroup\$ – Marcus Müller Mar 21 at 9:22
  • 1
    \$\begingroup\$ "... the wireless signals are getting short-circuited. This puts pressure on the router and hence the outage." This is not correct. The WiFi signals will be present and will propagate through the region of interference. The problem is that any WiFi receiver's input stage may be swamped by harmonic noise radiated by the overhead lines. \$\endgroup\$ – Transistor Mar 21 at 9:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.