Amped Wireless R20000G, SR20000G, and UA2000: Routers, Repeaters, and Adapters (Oh My!)

Wireless Networking Performance Setup

For testing the various Amped Wireless products, I don’t have an exhaustive selection of hardware that I can use for comparisons. Besides, as I mentioned in my review of the Killer-N Wireless Networking adapter: testing wireless products is hard. It’s not “hard” as in “difficult to do”, but rather it can be very time consuming. If you’ve heard the chaos theory butterfly effect, e.g. “A butterfly flaps its wings and a hurricane forms a few weeks later”, at times I’ve felt like testing wireless is even more prone to changes based on minute differences in testing parameters. What’s the weather like? What time is it? Where’s the sun (and/or moon)? Is the wind blowing? Where are my wife and children right now? Heck, where am I relative to the test equipment? I’ve tried to test each piece of equipment under similar conditions where possible, but time and weather—not to mention the movements of my children—are completely out of my control.

Neighboring wireless networks can also influence performance, but all of the tests were repeated numerous times, and I tossed out any results that deviated from the norm. Basically, I am using the highest performance results that I could consistently achieve—so if I measured 70-75Mbps in seven samples, 85Mbps in one, and 50-60 in two, I use the highest result from the 70-75 samples. I will note that in general, the 5GHz connections were much more reliable, at least in terms of throughput consistency. Once I had a laptop/adapter positioned more or less optimally for a test location, transfer rates were usually quite consistent. 2.4GHz testing was a different matter, particularly at longer distances, where it wasn’t unusual to see large fluctuations in transfer rates.

Even with my attempts to mitigate testing environment differences, I will say that there were cases where I would see one set of results (e.g. 50Mbps) and then come back later and performance would be perhaps as much as 25% faster or slower. Such is the nature of wireless testing in a real-world environment. Your performance with similar hardware could be quite a bit different than what I measured, especially if you live in an apartment complex. I’ve rounded off performance results to the nearest megabit, as anything less than that is far less than the margin of error—and in fact, I’d say the margin of error for wireless testing is probably at least 5% if not 10%.

But enough about the vagaries of wireless testing; let’s quickly go over the test equipment and locations. In order to keep things manageable, I selected three routers, three adapters, and three test locations. Couple that with the dual-band nature of some of the routers and adapters, and I have a matrix of 54 test combinations to run, and in most cases I ran each combination upwards of 20 times. (And I’m still not fully satisfied with the results, but short of restarting testing and spending another week or two there’s not much to be done now—I’m ready to throw in the towel!)

Wireless Routers

So I used three different wireless routers, as well as the single wireless repeater. Naturally we have the Amped Wireless R20000G/SR20000G, which are both 2x2:2 MIMO dual-band solutions. As a comparison point for similar hardware, I also have a Belkin N600 DB Wireless N+ Router (how’s that for a long name? The model number is F9K1102V1), which is a 2x2:2 MIMO dual-band router—basically, the same core feature set as the R20000G, but with internal antennas and obviously running different firmware and hardware. Finally, my third router is an older 2x2:2 MIMO 2.4GHz Netgear WNR3500L—it’s what I’ve been using for the past year or so at my house.

You’re probably wondering why there aren’t any 3x3:3 MIMO routers included, but the simple fact is that I don’t have any—plus as noted already, the amount of testing scales exponentially as I add more hardware to the matrix. It’s unfortunate that I lacked the necessary hardware, as I’d personally like to see how the Amped Wireless offerings compare to Apple’s Airport Extreme and the Linksys E4200 (among others), but consider this more of a “capsule review” as opposed to a full-blown roundup of the best current wireless routers. Besides the 3x3:3 routers, we’re also starting to see 802.11ac routers hit the market. We’re still waiting for 802.11ac adapters, unfortunately, and it’s worth noting that you’ll only get maximum throughput with 11ac at shorter distances—11ac is 5GHz only, and the 11ac routers use 11n for 2.4GHz support.

Wireless Adapters

For wireless adapters, I once again settled on three units. First up is the 2x2:2 dual-band UA2000, and as a comparison point I’ve included a laptop with Intel’s Advanced-N 6235 adapter (also 2x2:2 dual-band). The third adapter is a 2.4GHz only 2x2:2 USB thumbstick from Rosewill that you can pick up for $25 at Newegg—it’s what I’ve been using on my HTPC for the past several years. All three adapters were tested with the Ivy Bridge prototype Ultrabook we reviewed (though I did spot-check performance on the USB adapters with a couple other systems to verify that the Ultrabook wasn’t a bottleneck on any of the results).

Testing Locations

All testing was conducted at my house, and while I live in a residential neighborhood the surrounding houses are far enough away that they don’t usually cause too much in the way of interference. I could detect as many as eight other wireless networks, but only two of those had RSSI values better than -80 dB. All three test routers were configured to use channel 11 as the base 2.4GHz channel, with channel 7 as an optional sideband for 40MHz operation. For the 5GHz channel, I used 161 with 157 as the sideband for 40MHz. In both cases, the channels were selected to prove the least interference from neighboring networks.

Test Location A is in my home office, with the test laptop sitting about five feet away from the router and no obstructions between the two. This is the best-case scenario, as interference from other networks should be minimal at best. Generally speaking, performance throughout my ~15’ x 25’ office space is very close to optimal, dropping perhaps 10-15% at most in some areas.

Test Location B is at my home theater (really just an HDTV), approximately 30 feet away from the router in a straight line, or 40 feet walking along the wall, down the hallway, and into the living room. There are two interior sheetrock walls between the router and the test location, though there are no doors on the hallway or living room.

The final test location, C, is our “worst-case” option, and in order to get far enough away from the router I had to go outside. Thanks to the lovely Washington weather, I also decided to do my testing inside my car, as water and computer hardware don’t make good bedfellows. The straight-line distance to the router is approximately 55 feet, with two interior sheetrock walls and one exterior wall/garage door in the way. The garage door is made of relatively thin aluminum (I think), and it was closed during testing (when open, performance was substantially better).

For testing the repeater, I initially used a location closer to the router, basically giving Location B a clear view of the repeater. As location B proved to be less strenuous than I had hoped, I later shifted the repeater to location B and added the third (outdoor) location, C. The repeater is on the windowsill, with a clear view of the car in my driveway (which is not the red truck seen in the above picture from Google Maps, in case you were wondering).

Test Description

The last element of the testing is a description of the tests themselves. I turned to NTttcp for measuring bandwidth use, with the “host” system running a Core i7-975X and Gigabit Ethernet. However, I’ve noticed in the past that NTttcp doesn’t always generate the most reliable results on some hardware. As a second test I copied a 203MiB file from the same desktop system to and from the laptop. That gives us four performance results in Mbps: receiving rates for NTttcp and copy commands, and transmit rates for NTttcp and paste commands. I also measured RSSI using inSSIDer at each test location; as expected there are minor fluctuations, but I’m reporting the “average” RSSI for each adapter.