You may recall my failed attempt at using a second Wi-Fi router in repeater mode in order to overcome some signal-strength issues in the upper level of my house. I learned very quickly that compromising half (or more) of your wireless bandwidth in order to talk to a second router wirelessly isn't a very good trade-off for most clients. That's especially true in my case since our home cable modem service can reach up to 220 Mbps downstream, well beyond the delivered speeds we see out of our very nice wireless AC router, the Asus RT-AC87U.
The alternative to using a wireless router-to-router link is, obviously, some form of wired connection. Running Ethernet between the two routers and putting our older Asus RT-N66U into access point mode should allow us to have two sources of Wi-Fi signals at different spots in the house, both capable of full-speed communication with the Internet. But there's a big problem with that plan. Between the weirdness of our atrium-split floorplan and my own essential physical laziness, there was about zero chance that I'd actually run an Ethernet cable inside the walls anytime soon.
Fortunately, after my last post on this subject, some of you suggested trying a different sort of wired connection: power-line networking. A pair of power-line adapters will transfer data across your existing home electrical wiring. Although those sorts of products started out pretty poorly, they apparently have matured nicely in recent years. I immediately was intrigued by the idea and soon ordered a pair of these TP-Link adapters from Amazon for 70 bucks.
The idea was to put one adapter next to my main router and the other one next to the access-point router, with Ethernet connections going from each adapter to the adjacent router. The power-line network would then bridge between the two routers, hopefully providing a fast, reliable, low-latency connection.
Making it happen turned out to be a bit of an adventure, but not for the reasons you might expect.
When the power-line adapters arrived, I didn't mess around. I pulled them from the box, briefly glanced at the instructions and discarded them, and connected one adapter to my main router. Then I ran upstairs and plugged the other adapter into the wall socket in my bedroom and attached my laptop to it via Ethernet. I seriously didn't press any buttons or even look at any indicator lights on the little white wall-warts. Within seconds, I was pulling around 120 Mbps—maybe a little more—in a bandwidth test, with packet latency of 1-4 ms.
Man, that was easy.
Yes, the power-line adapter is rated for "up to 1200 Mbps," but I never expected to get practical speeds that fast. 120 Mbps is fast enough to outrun the Wi-Fi capabilities of most of the phones and tablets we use, and heck, I had the thing plugged into an outlet on an exterior wall that's as far away from the other adapter as possible within the house.
My next step was to connect the laptop to the RT-N66U and switch it from repeater mode into AP mode. Then I plugged the router's upstream port into the power-line adapter and fired everything up. Seemed like I was good to go, right?
What followed was a lot of disappointment, as I found that Wi-Fi clients on the RT-N66U only achieved about 60 Mbps on the 2.4GHz network and about 30 Mbps on 5GHz. What the heck? It seemed like things were no faster than before.
The process was more chaotic that I might care to admit, but my next steps involved a lot of A/B testing of various components of this network in order to track down the problem.
Moving the secondary power-line adapter to an outlet with a more central location in the house boosted Ethernet speeds to about 180 Mbps, with peaks near the 220-Mbps limit of our cable-modem service. My laptop, when directly connected to the power-line adapter, loved it. The location change also raised the speed of Wi-Fi clients on the RT-N66U to 35-38 Mbps on 5GHz and over 70 Mbps on 2.4GHz, but it wasn't exactly a breakthrough.
Was something wrong with my router, or did the combination of Wi-Fi plus power-line somehow not provide the stability needed to reach higher transfer rates?
Ultimately, I wound up sitting here in Damage Labs with the RT-N66U attached to a port on my GigE switch and configured with unique SSIDs on its 2.4GHz and 5GHz segments. Everything was as explicit as possible (maybe including my language). With my laptop five feet away, I could reach peaks of 80 Mbps on 2.4GHz and 40 Mbps on 5GHz, nothing more.
I knew the RT-N66U was capable of higher speeds, but it just wasn't delivering. Thinking there might be some bug in the latest Asus firmware update, I installed the alternative Merlin firmware to see if that would help, but speeds didn't improve.
More tweaking of Wi-Fi parameters and such was involved along the way. I'm condensing a lot of hazy frustration. But at one point while spelunking through the menus, I noticed some settings for WDS bridging that I couldn't alter while the router was in AP mode. It looked like, possibly, the router might still be configured to bridge to the AC87U over 5GHz Wi-Fi—a leftover from when I had the thing in repeater mode.
I wound up going nuclear and doing a factory reset on the router. Then, after a bit of configuration back into AP mode, a breakthrough: speeds well in excess of 80 Mbps on both the 2.4GHz and 5GHz bands.
The frigging router had been secretly stuck in WDS mode, and at least half of its 5GHz bandwidth had been reserved for wireless bridging. Ugh.
With that issue sorted, I gradually set everything back up exactly how I'd intended, testing periodically along the way. Ultimately, the RT-N66U was talking to the main network over the power-line adapters and broadcasting the same SSIDs as our main router. Clients could connect to it seamlessly. I made sure there was no overlap in Wi-Fi channel use. We now had reasonably solid 5GHz connectivity in every room of the house, with a no-doubt 2.4GHz signal as a backup.
I haven't done a ton of directed testing on the variability or reliability of the power-line link, but in regular use through the past few days, the new setup has been essentially flawless. I've offered my family the chance to complain several times each, but no one has noticed any hiccups. Periodic speed tests with phone and tablets have reached peaks around 75-80 Mbps over Wi-Fi on either router. The desktop PCs with Wi-Fi can range higher, to 180 Mbps or more. Everything more or less works like it did before, but the Wi-Fi dead spots are eliminated and, thanks to stronger signals, performance is up generally.
I do have one caveat about the power-line adapters, though. After I first plugged them in, I noticed some strange, subtle noises while sitting at my desk working. Eventually, I realized I was hearing interference caused by the power-line network doing its thing. Moving my speakers' power plug from the wall socket to the UPS resolved the problem, but it's possible we could encounter similar problems elsewhere over time.
Other than that concern and the havoc caused by the router issues, setting up the power-line networking stuff has been a huge win. Worth checking out if you need a fast, painless extension to the other side of the house.
|Gigabyte's X399 Designare-EX adds Thunderbolt to Threadripper||7|
|No, you can't enable Threadripper's extra two dice||22|
|International Talk Like a Pirate Day Shortbread||23|
|Philips 328P6AU and 328P6VU monitors make the best of USB-C||7|
|Tuesday deals: graphics cards, a mobo, storage, and a big TV||13|
|EVGA Epower V breaks the shackles of stock GPU power delivery||21|
|Reminder: iOS 11 will arrive tomorrow||35|
|In the lab: MSI's Aegis 3 gaming desktop||13|
|Rumor: Eight-core desktop Intel CPUs and Z390 chipset riding in||28|
|For some users, though, Apple's commitment to maintaining the software on its devices as they age is an even more compelling reason than hardware for...||+29|