21 USB wireless AC adapters tested on WRT1900AC and RT-AC68U

[bigben]

Hardware.Info Magazine tested 21 wireless AC adapters from Asus, AVM, Belkin, D-Link, Edimax, EnGenius, Linksys, Netgear, Sitecom, TP-Link, TRENDnet and ZyXEL.

Test results matrix (image)
Will become available later at this comparison web page as well.

Tests were done on Linksys WRT1900AC and Asus RT-AC68U.

Gold Award for Edimax EW-7822UAC, EnGenius EUB1200AC and Linksys WUSB6300.
Silver Award for TRENDnet TEW-805UB and TP-Link Archer T4U.

 

[thiggins]

I see no test results. That link goes to a product comparison table with no test results. Please provide a link to the review.

 

[RogerSC]

No Asus USB-AC56 tested? Really like mine since Asus put out a new driver for Windows 8.1. I hear that it's a lot like the Edimax EW-7822UAC, though.

 

[bigben]

I see no test results. That link goes to a product comparison table with no test results. Please provide a link to the review.

Corrected.

 

[RMerlin]

One information that would have been interesting is the chipset used by each device. Tho I know some manufacturers love changing chipsets without advanced warning. Linksys were pretty bad for this, DLink did it quite a bit as well.

 

[sfx2000]

One information that would have been interesting is the chipset used by each device. Tho I know some manufacturers love changing chipsets without advanced warning. Linksys were pretty bad for this, DLink did it quite a bit as well.

I would add Netgear to the mix - depending on the HW revision, could be great for linux or a non-starter...

Asus clients (outside of their laptops) tend to be fairly stable - for USB client adapters, they're fairly consistent... I don't have any experience with their PCI adapters, but I'm guessing here it's similar...

Buffalo/Melco is similar to ASUS in this respect...

 

[Mordred]

The maximum test distance of 10m for 2.4G is way to short. I mean it gets interesting once you have one or two walls in between or at least a distance of 50 in case of free sight.

 

[bigben]

One information that would have been interesting is the chipset used by each device.

Here you go.

Broadcom BCM43526

• ASUS USB-AC53

Ralink-MediaTek MT7610U

• AVM Fritz!WLAN USB Stick AC 430
• Linksys AE6000
• Sitecom WLA-3100
• ZyXEL NWD6505

Ralink-MediaTek MT76xx 5GHz-only

• Edimax EW-7711MAC
• Sitecom WLA-3001

Realtek RTL8811AU

• Edimax EW-7811DAC
• Edimax EW-7811UAC
• Netgear A6100

Realtek RTL8812AU

• Edimax EW-7822UAC (Gold Award)
• EnGenius EUB1200AC (Gold Award)
• Linksys WUSB6300 (Gold Award)
• Sitecom WLA-7100
• TP-Link Archer T4U (Silver Award)
• Trendnet TEW-805UB (Silver Award)
• ZyXEL NWD6605

The above adapters do currently not have different h/w revisions.

The Belkin F9L1106az, D-Link DWA-171, D-Link DWA-182 and Netgear A6200 do already have several h/w revisions revisions. Unfortunately I don't know which revision was tested, so I did not include them in the above list.

 

[theveterans]

Intel 7260 still blows all of them away at 500 - 550 Mbps throughput. It's mini-pcie though but USB 3.0 shouldn't be a bottleneck. Maybe the USB 3.0 drivers aren't updated on the machine that those USB adapters were tested.

 

[RMerlin]

Intel 7260 still blows all of them away at 500 - 550 Mbps throughput. It's mini-pcie though but USB 3.0 shouldn't be a bottleneck. Maybe the USB 3.0 drivers aren't updated on the machine that those USB adapters were tested.

One potential reason I can think of is those USB adapters often have a very small antenna, and it's not always ideally positioned (being alongside the laptop base), while the mini PCI-E antennas are in the standing screen, above the bulk of the laptop, and also away from the EMI interference of the laptop itself.

 

[sfx2000]

One potential reason I can think of is those USB adapters often have a very small antenna, and it's not always ideally positioned (being alongside the laptop base), while the mini PCI-E antennas are in the standing screen, above the bulk of the laptop, and also away from the EMI interference of the laptop itself.

1) Power budget - USB 3.0 does allow for a bit more, but to maintain compatibility with USB 2.0, 5V/500mA is a challenge for more than 2 RF chains with the associated front end modules (LNA for Rx/PA for Tx) - barring major changes in physics, 2-stream is about as far as one can go and be reliable...

2) RF design - as you suggest, there are better opportunities for antenna design when built in as opposed to being a USB dongle

3) USB Overhead - USB has much more overhead on the signalling compared to PCIe - both for 2.0 and 3.0 - PCIe is much more efficient here, and much higher base bandwidth, even with a single lane interface, which most consumer NIC's are...

4) USB 3.0 and 2.4GHz self-jamming - will always be an issue here for USB3.0 based adapters when operating in 2.4GHz - it's about a 5dB hit, sometimes more...

 

[htismaqe]

One potential reason I can think of is those USB adapters often have a very small antenna, and it's not always ideally positioned (being alongside the laptop base), while the mini PCI-E antennas are in the standing screen, above the bulk of the laptop, and also away from the EMI interference of the laptop itself.

I have found this to be the biggest contributor in my experience.

 

[azazel1024]

1) Power budget - USB 3.0 does allow for a bit more, but to maintain compatibility with USB 2.0, 5V/500mA is a challenge for more than 2 RF chains with the associated front end modules (LNA for Rx/PA for Tx) - barring major changes in physics, 2-stream is about as far as one can go and be reliable...

2) RF design - as you suggest, there are better opportunities for antenna design when built in as opposed to being a USB dongle

3) USB Overhead - USB has much more overhead on the signalling compared to PCIe - both for 2.0 and 3.0 - PCIe is much more efficient here, and much higher base bandwidth, even with a single lane interface, which most consumer NIC's are...

4) USB 3.0 and 2.4GHz self-jamming - will always be an issue here for USB3.0 based adapters when operating in 2.4GHz - it's about a 5dB hit, sometimes more...

That is primarily a design issue though with #4. It is possible to have well shielded ports as well as USB3 controller/tracing and cut down significantly on the interference. In the white paper Intel put out, it looked like the difference (granted, they were testing with USB3 storage, not USB3 MACs) was around 12-20dB of interference depending on location with fully unshielded everything and 2-5dB for well shielded everything.

I do wonder/hope that with USB3.1 they'll shift away from 2.4GHz signalling and move to something like 2.8GHz or so. That shouldn't greatly reduce the maximum cable length, ~20% reduction in max cable length, but that should also hopefully move the EMI far enough from the 2.4GHz spectrum slice to prevent interference.