WiFi vs. Powerline Ethernet Bridges/Adapters


#1

The discussion comes up, relatively often, here on the forums, about the relative advantages of various network connectivity technologies. Often, in these discussions, WiFi gets a bad rap. I've maintained that's because most home WiFi systems are... not optimal [note 1].

One of the options often discussed is powerline Ethernet adapters (or "bridges"). WiFi has been working well for us [note 2], but I was curious as to how well powerline Ethernet bridging would. Did my research [note 3] and chose for my test Comtrend G.hn 1200 Mbps Powerline Ethernet Bridge Adapter 2-Unit Kit PG-9172KIT [note 4].

As most people purchasing such devices have found: The manufacturer's promises, vis-a-vis throughput, are exceedingly optimistic. I think the best I ever saw, with the two of them on relatively benign branch circuits on the same side of our home's split-phase service, was a couple hundred Mb/s. The good news is the worst I ever saw, using the most hostile setups I could find [note 5], was 60Mb/s to 80Mb/s. And, even in those hostile hookups connectivity seemed pretty stable.

But how did these things perform feeding a streaming device, as opposed to the WiFi connection we've been using? The following two screen shots tell the story:

That's a Channels speed test from a computer to the Fire TV Gen. 2 on the TV in the family room, using the Comtrend G.hn powerline Ethernet bridges [note 6].

Compare to the same thing done with the Fire TV on our wireless LAN:

Over three times the bandwidth, nearly identical latency and nearly half the jitter. The point here being WiFi isn't necessarily the worst choice, as it's sometimes made out to be.

Conversely: Sometimes powerline Ethernet bridges will win.

I did two tests of the powerline Ethernet bridges where they definitely out-performed WiFi in our environment. The first was to a corner room in the basement, with the laptop sitting on my lap. That's about as far from the AP as I can get and still be in the house. The laptop would get only ±17Mb/s on wireless, but averaged ±75Mb/s on powerline. The powerline tests were over several days, at various times of the day and evening.

Out in our shed, about 60 feet (18m) from the back of the house and about 85 feet (26m) from the AP [note 7], WiFi did only about 3Mb/s [note 8]. The Comtrend powerline bridges performed at a solid 70Mb/s.

Admittedly, none of this is very scientific. It's just some tests/demonstrations under just some particular sets of circumstances.

Notes:

  1. In fact most home WiFi networks use poor hardware and their access points (more often WiFi routers) are often poorly-sited.
  2. Our WiFi access point is centrally-located and is business-class hardware.
  3. "Research" was mostly Amazon reviews and a bit of light web searching
  4. The Comtrend G.hn product was chosen more for its ratings for reliable connections than the fastest ones.
  5. Worst-case setups were with each node at the far ends of branch circuits, putting as much house wiring between each node as possible. One was across the phase split, the other had the slave node on the busiest, noisiest branch circuit in the house.
  6. Computer and powerline bridge (at one end) both connected directly to home LAN's "backbone" Etherswitch. (So is the WiFi AP.)
  7. About 120 wire-feet (37m) from the distribution panel.
  8. Shed WiFi test was on 2.4GHz. 5GHz wasn't strong enough to establish an iperf3 connection back to the server. Test device was an iPad, rather than the laptop. (It was an afterthought.)

#2

I have had good results using Google WiFi. The pods repeat the wifi signal. Each pod has two ethernet jacks, so devices that can't perform well on wifi get plugged into a pod.


#3

Perhaps you have, but many who've posted here have found mesh WiFi networks and high-bandwidth video streaming a bad mix.


#4

I use the Google WiFi pucks, too, and I really like them... but like @jseymour said the farther from the main puck the lower your wifi speeds get. I only use them hardwired or else the farthest access point away gets about 30Mbps vs 1Gbps on the main puck.


#5

I have been using Eero WiFi mesh network and have had great performance. One tweak for our network was turning off tuner sharing. But even with that feature on we only had a few dropouts. It only really put a strain when we had about three TVs going or two tv and the kids were gaming or watching YouTube.

Also our DVR was wireless as well. That is our next upgrade is to get a dedicated DVR computer and hardwire it. Then we could go back to turner sharing.


#6

WiFi isn’t necessarily the worst...

But using WiFi extenders (device’s that Connect to your WiFi router, with intention to amplify it somewhere else in your house), is the worst idea anybody could think of.

Wireless communications are half-duplex (send OR receive) vs full-duplex (send AND receive simultaneously). So when you connect to a WiFi extender, which then connects to your WiFi router through wireless communications, you are connecting to a network node via half-duplex, which connects to your router via half-duplex. When looking at it like this, hopefully you understand why this is not in best practice.

Nothing beats the Ethernet cable....nothing.

MoCA 2.0 is okay, but in order for it to run optimally, you want to NOT use splitters.


#7

I use Google WiFi (x5) here but they are ALL connected to Ethernet Cable. So the backhaul is not over WiFi. However, I do not have Real Ethernet to each one. I use MoCA 2.0 to a couple of them which works well. Some will use a splitter for the ethernet / TV. But since I do not use TV over COAX I do not use them for TV out.

The reason I have so many WiFi Access Points is because of outside cameras. I have them at the 4 corners of my house plus the main router. The problem I see is when the client does not connect to the closest AP.

However, I do try to use Ethernet (even if it comes from MoCA) instead of WiFi whenever possible.


#8

MoCA is also full-duplex, so you are getting most of the benefits with that. Splitters are like ethernet hubs (not switches...hubs), in that they allow for more than 1 client connection within the same collision domain....this means there are chances for packet collisions and noise in the signal. But, with your application, that is pretty much brought down to a minimum. I give your setup an A+.....especially for using wired connections for the backhaul.

STAY AWAY from those power line adapters. I've seen them work, don't get me wrong, but they are not in best practice. Ethernet lines are prone to EMI (electromagnetic interference). With an ethernet signal being carried through a power circuit, it SHOULDN'T work very well, yet it does....not sure how or why, but you also never get the speeds as advertised on the box of the power line kits (not to mention, power line kits are also half-duplex, not full-duplex, and all devices connected through power lines are on the same collision domain)


#9

As with many other things: "It depends." If wired Ethernet is off the table, there's no coax running to the desired spot, so no MoCA, and you can't get acceptable WiFi performance to a particular location: Then powerline Ethernet bridges may be the solution.

I ran the pair with which I was experimenting between two worst-case locations for three days, with a laptop set up at the far end, running iperf3. I ran bandwidth tests against it repeatedly. Almost never got less than 60Mb/s. Hit peaks of 85Mb/s. Averaged somewhere north of 70Mb/s.

There was one five-minute test I started before bedtime. When I checked the next morning it'd had two packets that had dropped to 6Mb/s. The rest were normal, as described above.

If that room was not going to have wired Ethernet, I'd consider powerline Ethternet bridges for it. However, I might be inclined to try an 802.11ac wireless bridge, first. The laptop only got 17Mb/s sitting a couple feet off the floor, but an 802.11ac bridge, located at or above the dropped ceiling would probably do much better.


#10

It is not often enough that I run into a fellow user of iperf. THANK YOU!!


#11

I updated my testing doc with the results of testing WiFi and powerline out to our shed in the back yard.

I'm pretty impressed with these Comtrend G.hn powerline Ethernet bridges. They aren't blazingly fast, but they're remarkably stable and consistent.


#12

I use NetGear power line adapters in my music studio.They work great but they do throw out rfi / emf which affects my LP playback. Thus I unplug them when spinning vinyl. Otherwise for streaming UHD 4k video and surround they work great.