ASUS GT-AXE11000 5 vs. 6 GHz ... a preview

[thiggins]

Got my bought-on-my-own-nickel AXE11000 fired up and paired with an Intel AX210 on Windows 10, with the borrowed-from-Killer 22.30.0.11 driver.
I ran some quick downlink rate_vs_range tests, comparing 5 and 6 GHz using 160 MHz bandwidth. I let the router auto-select channels and it chose 48 for 5 GHz and 85 for 6 GHz.
Both bands linked at the maximum 2402 Mbps at the start of the run. Here's the plot.

That dip around 36 dB is consistent run-to-run, indicating work needed in rate adaptation. At any rate, the takeaways are:

• 5 GHz has higher throughput than 6 GHz
• The curves track pretty well, i.e. 6 GHz falloff does not appear to be steeper than 5 GHz
• I need to expand the run to add more attenuation range. But it doesn't appear that there would be that great a difference in range.

My RvR tests will now also measure latency. I run a separate ping traffic pair while the TCP/IP traffic is running. This is the basic technique used in bufferbloat testing. Since it's ping, it's measuring round-trip delay. But that's fine for our purposes. Here's the plot:

Again, both track pretty well, with 6 GHz latency rising a little faster than 5 GHz as signal levels drop.

Looks like I'll be doing this guy as my first review.

 

[JagoUK]

Looking forward to review.

One thing to consider when comparing 5ghz to 6ghz.
5ghz can be problematic to some. Weather radar on ch 124 in my case. (Although rules about DFS and weather radar could be changing soon)

 

[RMerlin]

I'd like to see the difference in attenuation when using a low versus a high 6 GHz channel. I'm curious of the difference between these two channels, since there is close to a 1 GHz difference between them.

 

[thiggins]

@JagoUK In my normal testing, I won't be testing 5 GHz with 160 MHz. As you said, it's not practical for many users. I'll be doing 80 MHz, which I'll check next.

@RMerlin Will do.

 

[thiggins]

Well, 80 vs. 160 MHz bandwidth sure makes a difference! Here's a 5 vs. 6 downlink run with 80 MHz bandwidth for both. This time it's channel 36 for 5 GHz, channel 213 for 6 GHz. Looks like for range, you'll want to run 160 MHz in 6 GHz. Unfortunately, that just perpetuates the channel-hogging nature of each new Wi-Fi standard.

@RMerlin Yes, I know it's not apples2apples from before. I forgot to force the 6 GHz channel. Will do that next time.

 

[avtella]

Your testing is pretty much inline with what I'm seeing with an AX210. Seems power levels are possibly lower on 6Ghz.

 

[thiggins]

Your testing is pretty much inline with what I'm seeing on a different AXE router paired with an AX210. Seems power levels are possibly lower on 6Ghz.

Well, they are supposed to be since 6E TX power limites are based on power spectral density (PSD) vs. EIRP. The main question has been how much range reduction will there be?

When you say "different AXE router" you mean something other than a GT-AXE11000?

 

[avtella]

@thiggins it seems the following two articles indicate that by HT80 - HT160, the 6Ghz band should be at similar power levels and coverage should be similar. I wonder if spectrum sharing with incumbents is a cause for lower performance? I suppose Intel drivers not being fully ready or AP side wireless firmware not being fully tweaked yet could also be a cause?

Solutions - Wi-Fi 6E: The Next Great Chapter in Wi-Fi White Paper

Wi-Fi again moves into a new spectrum, and along with the 6-GHz band come new rules and new radios capable of operating in and implementing the new access categories. This paper details the new Wi-Fi 6E band and the rules associated with it.

www.cisco.com

Power Spectral Density | Official Juniper Networks Blogs

Power Spectral Density is the amount of power over a given bandwidth. Read the blog to find out what this means for Wi-Fi 6E.

www.mist.com

 

[thiggins]

@avtella I don't know what you mean by "spectrum sharing with incumbents", Do you mean other devices on the channel?

My tests were done in RF tight chambers connected via programmable attenuators; it's a clean environment.

I didn't/don't have an RF power meter available to verify power levels. At this point anything is possible. The "only 3dB" EIRP difference between 80 and 160 MHz channel bandwidth both articles reference is 2X power. And while path loss difference isn't huge, it's still another factor.

The more interesting question to me is why in both 80 and 160 MHz cases is peak throughput always higher for 5 GHz?

 

[avtella]

I meant incumbents like TV broadcasters etc that apparently also share the 6Ghz spectrum, going by the third slide in the link. My apologies, I overlooked the fact that your testing was done in a clean environment.

https://www.wirelessinnovation.org/assets/Webinar_Slides/Spectrum_Sharing_Deep_Dive/Gibson%20-%206%20GHz%20Unlicensed%20Update%20-%20WInnForum%20Spectrum%20Sharing%20Workshop.pdf

 

[RMerlin]

The more interesting question to me is why in both 80 and 160 MHz cases is peak throughput always higher for 5 GHz?

Could be missing/poor optimization on Asus's or Broadcom's part for that band.

 

[Fatawan]

@avtella I don't know what you mean by "spectrum sharing with incumbents", Do you mean other devices on the channel?

My tests were done in RF tight chambers connected via programmable attenuators; it's a clean environment.

I didn't/don't have an RF power meter available to verify power levels. At this point anything is possible. The "only 3dB" EIRP difference between 80 and 160 MHz channel bandwidth both articles reference is 2X power. And while path loss difference isn't huge, it's still another factor.

The more interesting question to me is why in both 80 and 160 MHz cases is peak throughput always higher for 5 GHz?

Are you sure those 6E channels are correct? If I am reading specs correctly, channels 85 and 213 are both 20 MHz wide. Are you being throttled by firmware problems? Also, I would be very curious if the router allows you to use channel 15 on 6E. This router came out before proper filters were available that would separate the upper part of the band centered at 5.67GHz and the lower band of 6E at 5.925-ish GHz. I was told that manufacturers were just going to block out that lower section of U-NII-5 because of interference. I have attached a pic of a proper BAW filter(which is now available) overlaid on the DR cavity filter that is most commonly used. Also, if you crack it open, would love to see pics of the innards!

 

[Razor512]

For 6GHz WiFi the router effectively just picks the closest 160MHz chunk of spectrum to the channel you picked where the one in the list will be the primary channel, and probably 7 others will be selected along side it depending on which channel width you selected. I guess the only major issue would be spanning 2 secments of the spectrum, thus causing you to get a combination of the worst aspects of both, similar to when you use 160MHz on the 5GHz band, it always creeps into the DFS range which means the entire signal gets crippled down to 24 dBm transmit power.

 

[Fatawan]

@thiggins Never mind on internal pics. The FCC pics are now available. As suspected, this router uses DR filters made by Xiamen Sunyear for the 6GHz band which are very large and cut off the first 80-160 MHz of UNII5. Shame. The newer models will have BAW filters that will allow full use of the 5.925-7.125 GHz bandwidth of WiFi 6E, or I should say they SHOULD have them if manufacturers are doing the right thing.

 

[gfondeur]

I should say they SHOULD have them if manufacturers are doing the right thing.

Lets hope for that

 

[Dave Fey]

Sort of a question/area of concern. Am I right that AX210 hooked up to a normal AX201 set of antennas is using antenna's not properly optimized for 6GHz? Is that what you did? A lot of users that I read grabbing AX210's have been using them in basic AX 5GHz wired notebooks. If someone got a set of 6e antennas on a desktop install that would play out differently. But proper amplifiers might be hampered by improper antennas.

 

[thiggins]

Sort of a question/area of concern. Am I right that AX210 hooked up to a normal AX201 set of antennas is using antenna's not properly optimized for 6GHz? Is that what you did? A lot of users that I read grabbing AX210's have been using them in basic AX 5GHz wired notebooks. If someone got a set of 6e antennas on a desktop install that would play out differently. But proper amplifiers might be hampered by improper antennas.

Excellent question. The AX201 was connected to octoScope's "5 GHz" high-gain antenna. octoScope has reviewed its programmable attenuators and antennas and they are good through the 6E frequencies, albeit at somewhat higher losses.

AX210's will work with current dual-band antennas. They just won't get optimum signal in 6 GHz.

 

[Dave Fey]

So that's one area that the difference between 5 and 6 would get a bit tighter. Thanks, good for me to keep in mind. Means I can still upgrade older "cards" with some loss if operating in the e bands, but waiting for actual 6e optimized antenna's in to come equipped in devices has some merit. Mind you I'm keenly aware that the law of diminishing returns means real world we aren't going to move the speed needle by so much, but I am happy to start to see testing just to see what we see.

 

[thiggins]

The biggest change 6E will bring is the ability to use 160 MHz channels without the DFS hassles. Unfortunately, they'll be used by many people who don't really need the high bandwidth, so the 7 available 160 MHz channels will get crowded pretty quickly.

 

[gfondeur]

True