Like every other website on the planet, SmallNetBuilder uses cookies. Our cookies track login status, but we only allow admins to log in anyway, so those don't apply to you. Any other cookies you pick up during your visit come from advertisers, which we don't control.
If you continue to use the site, you agree to tolerate our use of cookies. Thank you!

Wireless Features

{mospagebreak toctitle=Introduction, AC1900 vs. AC1750 Adapter}

Introduction

I made my case for not buying AC1900 routers months ago in AC1900: Innovation or 3D Wi-Fi? But according to the router makers I spoke with at CES a few weeks ago, they are selling like gangbusters. This is primarily because early adopters who have been driving most of the AC router buying so far aren't concerned about price; they just want the biggest, baddest router they can get. Manufacturers, of course, are loving this, because they can sell fewer routers and make as much, if not more money.

SmallNetBuilder has always been about making fact-based decisions. So in the interest of getting more facts to test my not-so-rosy opinion of AC1900 routers, I did some additional testing of 2.4 GHz performance for the three AC1900 routers sucking money out of wallets today: ASUS' RT-AC68U; NETGEAR's R7000 Nighthawk; and Linksys' EA6900. The ASUS and NETGEAR were originally reviewed here and the Linksys here.

AC1900 vs. AC1750 Adapter Test

My primary objection to AC1900 routers is that most clients can't achieve the 2.4 GHz 600 Mbps link rates that are the main reason for this class' existence. That's why my testing has continued to use ASUS' PCE-AC66 AC1750 class client, vs. its AC1900 class replacement, the PCE-AC68.

I also had not been able to switch to using the AC68 even if I wanted to, because ASUS had not released a driver that enabled 600 Mbps link rates. That latter obstacle changed in mid-December, when ASUS released the 2.0.7.1 version of its driver/utility for the PCE-AC68. So I went back and retested the three AC1900 routers' 2.4 GHz performance using the PCE-AC68 and the new driver.

I first upgraded each router to its latest released firmware, i.e.:

  • ASUS RT-AC68U: 3.0.0.4.374_371
  • Linksys EA6900: 1.1.42.156465
  • NETGEAR R7000: 1.0.2.194

(Yes, I know ASUS released 3.0.0.4.374.583 on 1/14/2014. But I already had done a lot of testing with 3.0.0.4.374_371, so wasn't going to change.)

I then ran 2.4 GHz up and downlink profiles with the rear of the router facing the test chamber antennas 8" away. Before starting the test run in each case, I checked that the PCE-AC68 did indeed link at 600 Mbps (or at least at rates above the 450 Mbps maximum link rate for non 256-QAM operation). The Linksys EA6900 was the only router of the three that showed a 450 Mbps link rate.

After I ran the tests with the AC68 adapter, something told me that I should run the tests again, but with the PCE-AC66 adapter. Since router firmware had been updated, I didn't want to just reuse the old test data. I then pulled each set of results into comparative plots, which revealed surprising results. Let's start with the ASUS RT-AC68U.

ASUS RT-AC68U 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

ASUS RT-AC68U 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

Theoretically, a jump from 450 to 600 Mbps link rate should provide a 33% throughput gain. If there is such a gain in these plots, I sure can't see it.

ASUS RT-AC68U 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

ASUS RT-AC68U 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

The Linksys EA6900's results were quite different, but not in a good way.

Linksys EA6900 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

Linksys EA6900 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

Since the AC68 adapter reported only a 450 Mbps link rate, the results shouldn't be any worse than the AC66's. But they clearly are much worse.

Linksys EA6900 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

Linksys EA6900 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

The NETGEAR R7000 fared a bit better than the Linksys. But it still had overall lower throughput with the AC68 adapter, even though it linked at 600 Mbps.

NETGEAR R7000 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

NETGEAR R7000 2.4 GHz downlink - PCE-AC66 vs. PCE AC68 adapters

I say a bit better, because the curves tended to come together at lower signal levels, at least for downlink.

NETGEAR R7000 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

NETGEAR R7000 2.4 GHz uplink - PCE-AC66 vs. PCE AC68 adapters

Call me suspicious, but I think it's odd that the ASUS did the best out of the three tests conducted. But, even then, it managed only to not lose throughput using a 600 Mbps link rate adapter. So if there were any benefit to using a "matched" AC1900 router and client from ASUS, it appears to be only in maintaining the same performance as you get using ASUS' PCE-AC66 adapter.


AC1900 Bridge Tests

I frequently say that if you're going to buy an AC1900 routers, you should buy two of them, since that's the surest way to get a 600 Mbps link rate in 2.4 GHz. So, in the interest of having data to back up my advice, I ran 2.4 GHz profiles with a pair of NETGEAR R7000 Nighthawks and a pair of ASUS RT-AC68Us. No offense intended to Linksys, but I ran out of time alloted to this work, so didn't run a pair of EA6900's.

In each case, I set the main router to Channel 6. I selected the "Up to 600 Mbps" wireless mode for the NETGEAR and 20/40 mode with the default of Turbo QAM enabled for the ASUS. I configured the second router in non-WDS wireless bridge mode in both cases to ensure that I had a WPA2-AES encrypted link fully capable of maximum link rate.

I checked the status display in both R7000's, and both reported a 600 Mbps link. The ASUS reported only a 486 Mbps link rate, which is the lowest of three 256-QAM link rates with a 40 MHz channel (the three rates are 486, 540 and 600 Mbps).

The R7000's downlink plot finally shows a notable throughput boost for strong to medium signals. But as signal levels move into the weak range, the throughput advantage disappears as the 20 and 40 MHz bandwidth plots track for a little while. As signal levels drop further, we see lower throughput for the 40 MHz B/W plot, indicating lower usable range. This confirms what I used to see when running 40 MHz mode tests, i.e. lower range with 40 MHz bandwidth in use.

NETGEAR R7000 2.4 GHz bridge 20 / 40 MHz compare - downlink

NETGEAR R7000 2.4 GHz bridge 20 / 40 MHz compare - downlink

The uplink plot shows the same pattern as downlink, at slightly reduced maximum throughput.

NETGEAR R7000 2.4 GHz bridge 20 / 40 MHz compare - uplink

NETGEAR R7000 2.4 GHz bridge 20 / 40 MHz compare - uplink

I calculated the difference in average throughput between 20 and 40 MHz in both directions and got a 29% difference in downlink and 27% in uplink. But again, most of the gain comes at strong to medium level signals and at the sacrifice of reduced range.

Despite its lower reported link rate, the ASUS started out with 331 Mbps vs. 117 Mbps for the R7000. But throughput fell sharply with only 21 dB of attenuation and kept going down from there. Once again, we see 40 MHz bandwidth throughput fall below 20 MHz mode at around 51 dB of attenuation.

ASUS RT-AC68U 2.4 GHz bridge 20 / 40 MHz compare - downlink

ASUS RT-AC68U 2.4 GHz bridge 20 / 40 MHz compare - downlink

The shape of the AC68U's uplink profile looks more like the R7000's, but starting again with higher throughput (~190 Mbps vs. 160 Mbps).

ASUS RT-AC68U 2.4 GHz bridge 20 / 40 MHz compare - uplink

ASUS RT-AC68U 2.4 GHz bridge 20 / 40 MHz compare - uplink

Difference in average throughput for the ASUS between 20 and 40 MHz modes came in at a 64% difference in downlink and 26% in uplink. But again, most of the gain comes at strong to medium level signals (particularly for downlink!) and at the sacrifice of reduced range.

Closing Thoughts

I can't say I'm totally surprised by the poor results from the PCE-AC68 adapter. It's a unique beast and has had a long birthing process. Perhaps another driver revision will improve its performance or perhaps not. It's not that it matters that much, since the audience for the adapter is rather limited. I won't be changing to it as my reference adapter anytime soon, if ever.

The bigger story is how AC1900 bridges performed. More people are likely to try to take advantage of AC1900's potentially higher 2.4 GHz link rates this way, even though it ain't cheap. But with AC2300 on the June/July horizon, AC1900 prices should start to drop and make an AC1900 bridged pair more affordable.

If you decide to give it a shot, remember that you can achieve > 450 Mbps link rates only using 40 MHz bandwidth. AC1900 routers are supposed to obey 20/40 MHz coexistence rules and fall back to 20 MHz bandwidth if there are interfering networks. So you might find yourself on the lower curves in the plots above, which is about where you'd be with a pair of top-performing (and less expensive) AC1750 or N900 routers like those shown below.

AC1900 router alternatives

AC1900 router alternatives

So are AC1900 routers worth their higher price tags? Maybe. Since the three on the market all use the same Broadcom BCM4360 chipset used in AC1750 routers, their improved 5 GHz performance is more due to better drivers than advanced silicon. And the ASUS and NETGEAR AC1900s can also deliver improved 2.4 GHz range vs. their AC1750 counterparts.

But if you're looking for AC1900 to deliver higher 2.4 GHz throughput, your best chance—as I've said from the start—is to buy a pair and have clear enough spectrum to use the 40 MHz bandwidth mode you'll need to get it.

P.S. All three products have also been retested with our normal 2.4 and 5 GHz benchmark tests.

Discuss this in the Forums