Ethernet-Over-Power (aka powerline ethernet)

Imagine you have a piece of electronics equipment that you want to install somewhere. The device will need power and perhaps networking. Power can be supplied from a nearby outlet or batteries. Networking can be provided wirelessly, or by running wires to the location. Chances are most areas of your home already have outlets nearby, available networking might be a different story. Running ethernet to a location can be expensive and messy, especially if you want to "do it right" by running it inside the walls, using faceplates, and making everything look nice. Wireless sounds great in theory, but is often tricky to get working reliably in practice. Ethernet-over-power (aka powerline ethernet) is a way of using the existing power lines to also carry the networking data; removing the need to run ethernet to, or setup a wireless connection at, locations that already have power but no networking.

Not to be confused with power-over-ethernet (PoE) which is the exact opposite scenario in which you have a location that has ethernet but no power outlet. PoE allows you to transmit power (upwards of 71 watts) as well as network data over the ethernet cable; removing the need to run power to locations that already have ethernet but not power.

Our house was built in the early 1970's and doesn't have wired ethernet. Most people can solve this problem with WiFi. WiFi works somewhat for us, but not very well. The original house was built with 16" thick granite stone walls, then in 2001 an addition was built. So now the house is quite long, with a 16" stone wall in the middle. Including the garage, our entire house is about 100' long by 30' wide.

When we first moved in, we setup WiFi. Currently WiFi comes in 2 different bands: 2.4GHz and 5GHz. 2.4GHz is slower, but travels further and through more things; 5GHz is faster, but has a shorter range and doesn't go through walls as easily (especially walls made with 16" of granite). Setting up the WiFi router in the original house at 2.4GHz works okay for some areas of the original house and barely works in the addition. Setting up the WiFi router in the original house at 5GHz barely works in the original house, and not at all in the addition. Conclusion: WiFi isn't that great. At a minimum, in order to get it working halfway decent, I would need to install a bunch of repeaters throughout the house. Getting WiFi repeaters to work well and seamlessly isn't easy (although the situation is improving with the move to mesh networking).

Incorporating users' reviews into research on any topic can be a dicey proposition, and the user reviews of ethernet-over-power did not disappoint in that regard! Some people claimed that it worked perfectly, others said it didn't work at all. Some people said that they could "hear" the devices from their neighbours' ethernet-over-power setups, others said they couldn't communicate with their own devices from one side of a given room to the other. Some people claimed that ethernet-over-power signals could not cross an electrical panel, others said it crossed just fine. Some people claimed that all the devices you wanted to connect together had to be connected to the same side of a given electrical panel, others said no such restriction existed. Reading, and interpreting, customer reviews is an art onto itself, so it's hard to know if some of those bad reviews are actual limitations of the technology, or bad products, or faulty devices, or customers doing something wrong, or perhaps customers not understanding. Perhaps it was a case of older technology versus newer?

When we first moved in I had discovered ethernet-over-power and hoped it was a cheap solution to all my problems. But all these contradictory reports caused me to drop it from consideration back then.

More recently I was chatting with a co-worker about such things and he mentioned ethernet-over-power as a great solution that he himself was using. I mentioned that I had investigated it a while back and dismissed it, but he assured me that it worked great. I asked him about "same side of the panel" issues, I asked him about "crossing a panel" issues, I asked about "hearing the neighbours", and everything else I could remember. He said none of those issues existed and that it worked fine for him. In any case, based on his recommendation I bought a couple units and gave it a try.

Today as I write this entry I have been using ethernet-over-power for just over a year. I'm currently in the process of running ethernet wires around the entire property so I can get rid of ethernet-over-power. It doesn't work well for me, but that doesn't mean it won't work well for you.

I've purposefully avoided saying which exact brand and products I'm using. My guess is they all work somewhat the same. Short-comings that I find in one are probably present in the others, and things that work well in one are probably similar to features in others. It's probably also worth noting that all the devices that I bought come from the same company. In fact I bought all the same specific products, except for one device which still came from the same company, but was a 2-jack version of the same device. Therefore I don't think that the reason why it doesn't work well for me is because I chose one company over another, or one product over another, or because I mixed products from different manufacturers. There are limitations to the technology itself which I believe are the root of my issues. Perhaps a competitor's product would have worked slightly better, or maybe slightly worse, but in the end I don't think ethernet-over-power was ever going to solve my issues.

Ethernet-Over-Power Basics

Starting with a house that has an ethernet network, and a network of power wires:

 

The ethernet network can only be "injected" once into the power wires at one point; loops cause the network to go haywire.

 

 

Any number of leaf devices can be attached to the power network, but the "only be one bridge" rule must be maintained.

It is even possible to have networking equipment "after" the ethernet-over-power network. Just make sure the rule requiring only one bridge is maintained.

 

 

Many of the ethernet-over-power products come in a 2-pack. The two devices are identical, either one of them can be used as the bridge device or for a leaf device.

From what I can tell, ethernet-over-power has no problem "going through" electrical panels, and has no problem connecting devices which are found on either side of the panel. Maybe that's related to the specific panels I have, but I actually have 2 panels involved in my setup, from 2 different manufacturers, and neither panel seems to be an issue.

Distance and Length

We have a house that is significantly longer in the east-west orientation, and narrower north-south. Since our house is long east-to-west, I decided to try placing the bridge using an outlet that is somewhere around the middle of the house (lengthwise) and then test this setup by placing leaf devices at various points around the house (including at both extreme ends). Unsurprisingly, the devices placed closer to the bridge work fine, but surprisingly devices at the extreme east and west side of the house don't get any connectivity. Length/distance is definitely an issue for this technology. Moving the bridge to different outlets around the middle of the house did change which leaf devices could be reached, but I wasn't able to find one place for the bridge such that all leaf devices could be seen.

It's important to note that distance is a measure of the length of the power wires, not the distance between two outlets "as the bird flies". Just because two outlets might be located relatively close to each other in a room, doesn't mean they they have a short run of electrical wire between them. Both of them could be on the same circuit, or they could each be at the end of two separate and long runs. However, if two outlets are located close to each other in the same room, it's not unreasonable to assume there isn't much wire between them, but that might not always be the case. 

House length, hence power wire length, is definitely a factor in whether this technology will work for you. Manufacturers of these devices will often try to put some kind of number on what sort of distances over which they expect their technology to work, but these numbers might as well be pulled from thin air. Several manufacturers claim their devices are good for up to 1000'. My house is only 100' long. And yet placing the bridge in the middle of my house, I could not communicate reliably with devices at both ends (i.e. 50' each way). Therefore, either whoever wired up my house went crazy and used 10x the amount of wire required, or the specs on the ethernet-over-power devices is bogus.

If the values were just a little off then I could dismiss it as being the difference between theory and practice, or the difference between a lab environment and a setup in the field. But being off by a factor of roughly 10 is getting into "bogus" territory.

The Other Factor

The second factor took me a little longer to figure out. It was so elusive that I don't even know what it is or what to call it.

For the past year my hybrid ethernet and ethernet-over-power network was working relatively well. However, perhaps once every 2 weeks or so, I would have these periods where certain leaf nodes would drop off the network, then re-appear again at a later time. Sometimes only the furthest device would be gone, sometimes 2 or 3 devices would be gone, and on rare occasions, all of the devices on the ethernet-over-power network would disappear. Once, for about 4 days in a row,  the entire powerline network would completely disappear at exactly 7pm every day for roughly an hour. Eventually all the network devices would re-appear again and the network continued mostly working. Sometimes they would re-appear after a couple minutes, sometimes it would take several hours.

Due to the unpredictability of these events, it was hard to understand what was happening. I searched and searched online trying to find answers but nothing definitive came up. After a while I developed a theory that these "brown outs" tended to occur when large loads were running (hot water tanks, ovens, or the cloths dryer) so I started paying attention to those things. Usually those did explain some of the issues, but not entirely. For example: my cloths dryer (a 2-phase, 240V, 30A device) had no effect on the powerline network that I could tell. My dishwasher, however, which is just a regular single-phase 15A system, definitely would cause issues. I'm also pretty sure that my seasonal winter lighting (strings of LEDs) were causing issues too.

So it seems as though the best way to explain it is to say that some devices will cause issues with the working of a powerline system and others won't. Usually the devices that are a larger load on the electrical system will be the ones that cause issues, but that's not always the case. Sometimes a "regular" device can cause havoc. There's a general correlation between higher load devices causing larger disruptions to the powerline network, but it's not absolutely true in all cases.

One person I was speaking with suggested that perhaps these issues could be the result of some combination of load, phases, and frequency, or perhaps the dirtiness, of some electronic devices. Another person claimed that some utility companies might be using some form of ethernet-over-power themselves to read power meters for billing purposes.

I don't know the cause, I only know that for me ethernet-over-power is not entirely reliable and determining the exact causes will need much more investigation. But even if I found exact answers to all my questions, it still wouldn't fix the problem.

My Conclusion

The conclusion I have reached is that ethernet-over-power is affected primarily by 2 factors: distance, and something else. The something else is hard to define and could be combination of higher loads, or something to do with phases and/or frequencies, or poorly designed electrical devices which cause noise in the power network, or the result of interference by utility companies. The point is, you can't simply make a list of all your electrical devices, then look through the list and know beforehand which ones are going to interfere with the ethernet-over-power network or to what extent (if at all).

Ethernet-over-power does not work reliably enough for me and will not provide the coverage and 24/7 reliability that I want. I tried it for a year and am removing it in favour of running more ethernet. However, although it doesn't work for me, my co-worker claims it works perfectly for him.

Hard-wired networking provides the best experience and the most reliable connections. Alternatives to hard-wired networking will probably be cheaper, but less reliable. Will ethernet-over-power work for you? Unfortunately there's no way to know in advance. The only way to know for sure is to spend the money and try it. If it works, great! Otherwise...