Does An AC Router Improve N Device Performance?

sfx2000 said:

Huh???

None of this makes sense man...

Click to expand...

In what way? The original question "Does AC improve 5GHz N performance."

My answer Yes, sort of. My DD-WRT AC routers are connected. Router2 is a client bridge. Router 2 is mostly for media. DD-WRT is 5GHz N-only. The bridge link is at ac speeds but I can connect with N clients to the main router. Netgear bridge default is AC but it works only as a bridge. DD-WRT allows more flexibility.

With DD-WRT, 5GHz N-Only improves performance because I have both N and AC available. Bridging works at AC speeds but N clients can still attach. If I had an AC wireless card in my laptop, it could probably run at ac speeds.

Thanks to DD-WRT, Router 2 still has a 2.4GHz SSID available if needed. 5GHz carries the signal to Router 1.

The original question really wanted to know if higher quality electronics improved overall N-5GHz speeds. It probably does, but I'm looking to improve my entire relationship with my home network. The AC bridge makes it happen and DD-WRT allows me to still use 5GHz N devices.

Sorry man, you tend to be really terse, and sometimes it's hard to follow your comments in the context they're given in...

Consider the audience - we've got a range from very experienced engineers to advanced hobbyists to folks that have just joined up looking for technical information.

You're obviously a very knowledgeable person, and your contributions are appreciated...

sfx2000 said:

Sorry man, you tend to be really terse, and sometimes it's hard to follow your comments in the context they're given in...

Consider the audience - we've got a range from very experienced engineers to advanced hobbyists to folks that have just joined up looking for technical information.

You're obviously a very knowledgeable person, and your contributions are appreciated...

Click to expand...

i have the same problem with writers who take 1000 words to say what can be said in 100. i usually never finish what they write because it's hard to draw a straight line from start to finish. TMI makes me go to the next article. I keep that in mind when I write. Some sentences can be dense .... guilty.

Moved from WNDR4000 to R8000

First post, here. I'm sure you guys have forgotten more about wireless networking than I'll ever know, but here's my experience.

The wifi devices we have are: two HTC One M8 phones, a Lenovo X131e Chromebook, a Nexus 10 tablet, an Intel-based iMac, a Brother laser printer and a HP Pavilion dv7-4295us laptop. All are 802.11n, except for the phones, which are 1x1 802.11ac. I'm not sure what wifi the printer uses.

I just replaced my WNDR4000 with a R8000, and the difference could not be more stark! We have a long, rectangular ranch-style home with the internet entering the home at the home office, which is at one end of the house.

Our master bedroom is at the other end of the house, about seventy feet away, and the signal has to travel through a fireplace, two walls and heavy oak furniture to get to my side of the bed. Our garage is twenty-five feet beyond that point, through another wall and a steel storage shed.

With the WNDR4000 I could barely get 2.4ghz into the bedroom, and 5ghz would drop if I walked into the bedroom with my HTC One M8. The signal was very weak on our front porch (with 5ghz being unavailable) and data transmission was very slow. Forget getting any wifi in the garage.

After replacing the WNDR4000 with the R8000, the increase in signal strength is dramatic, allowing useful connections all the way out to the garage. I'm typing this on the Chromebook, which is currently attached to the R8000 at a channel width of 80mhz; it would never do that with the WNDR4000. My phone gets good reception in the garage, loading websites quickly, even graphics-heavy sites like ESPN and CNET. Every one of our wifi devices is getting a much better signal than it did before.

It really does make a difference using a 802.11ac router with non-802.11ac machines.

I have the R8000 set up to automatically assign devices to the appropriate radio, with both 5ghz radios configured identically, except for the channels they operate on. It was very easy to configure, and so far there have been no glitches. No slowing down the faster devices when the slower devices are operating. Videos load and play much faster and better on all machines now than they did with the WNDR4000.

I'm very happy with the R8000, so far.

Stumbled upon the SNB article from google and got into this discussion from it.

First off, I love real world anecdotes that help make my case for upgrading. However, what I don't understand, is why the official charts don't seem to match the results posted on this article?

With a reference N66U router, there's apparently a roughly 4x performance increase in the mid-to-long distance performance. However if I look at the router charts (for example http://www.smallnetbuilder.com/tools/charts/router/bar/117-2_4-ghz-profile-dn?see=P_60) I see no indication that at the same level of attenuation, the ac radios are at all better than the old 802.11n kings. In fact in particular on the linked profile, the N66U is actually twice as fast as the AC68U instead of 3x slower.

I can see a few possible reasons for this, and was wondering if anyone has any insight as to where my logic is failing.

1) Either the different routers have different levels of effective attenuation at those different locations (seems unlikely as attenuation should be a function of the environment and not of the router).

2) The testing procedure between when the N66U was run vs. the newer AC routers has changed so dramatically that the original results would most likely change significantly (drop to 1/6th?). Looking at the testing procedure changes this doesn't seem to be the case but I could be missing a key piece of information.

3) Changes in firmware have been so great over time that between the time the new ac routers were tested and writing of this article the performance shifted by a factor of multiple hundreds of percent.

Am I missing something?

It is a combination of everything.

N Class routers were tested at the time with N Class clients. Today, that would make no sense to test AC Class routers with only N Class clients.

Firmware has changed and seemingly (officially), towards lower throughput. Pesky Netgear whining and FCC standards and all.

Circuit and antennae design optimization and topology is always a moving target. As is the newer SoC's with faster cpu's and increasingly more ram.

New software tools and here, I'm including drivers for both the routers and our clients. These too are always progressing and comparing to a 3 year old device with probably more than 3 year old tech at the time of introduction is showing what the real progress has been, given the RF power reductions that have taken place for everyone in the last few months.


The charts were made with a fixed in time firmware and similarly outdated hardware and software testing setup. No one would expect each router to be tracked through firmware upgrades, let alone the changing testing procedures as they make sense to implement as time goes by.


The best use of the charts is to see the relative performance of products, not the absolute, nor real world results that should be expected.

In this regard, they continue to serve as the standard for the whole world.

Triggerhappy,

There were three big changes between the Rev 7 and 8 test processes: client, test chamber and DUT rotation during testing. So results can't be fairly compared between the processes.

New routers are better at transferring data via software and hardware. Simple, imho.

Steve40th said:

New routers are better at transferring data via software and hardware. Simple, imho.

Click to expand...

That may be true in general, but not true specifically.

Otherwise, SNB would not be needed nor would the charts be useful either.

But in a general sense processors are faster so new routers should be better.

coxhaus said:

But in a general sense processors are faster so new routers should be better.

Click to expand...

Yes, they should be. Your results may vary.

L&LD said:

That may be true in general, but not true specifically.

Otherwise, SNB would not be needed nor would the charts be useful either.

Click to expand...

SNB helps people weed out the numbers games, ie routers with higher prices and sequential numbers. Price and part numbers dont always mean better performance.

Steve40th said:

SNB helps people weed out the numbers games, ie routers with higher prices and sequential numbers. Price and part numbers dont always mean better performance.

Click to expand...

You're kind of contradicting yourself now.

I agree with your statement quoted above, but to meld your two posts together with what I have already stated; 'newer part numbers do not always give you the best performance'.

Steve40th said:

SNB helps people weed out the numbers games, ie routers with higher prices and sequential numbers. Price and part numbers dont always mean better performance.

Click to expand...

Interesting that you make this point...

Use Case - Alice has a 3-stream capable laptop that is 802.11n compliant - (N900 class), or N450/N450 - Bob has a two-stream AC1200 class laptop that is 2-stream, but it's N300/AC867 if you split the band classes up, and stay standards compliant (folks, VHT40 mode in 2.4GHz is non-standard)...

The numbers game is getting very much out of hand... I've tried to make this point, I think others have ranted about this as well - smart folks that have made huge contributions to the collective here - but it seems to fall on market driven dead ears...

sfx2000 said:

Better and newer chipsets, updated drivers, more optimization on the networking stack. Much more recent Board Support Packages (BSP's) with later linux kernels.

It's the trickle down effect of making 802.11ac work in 5Ghz, and as such, it's a selective reduction of certain 802.11n options - it's a much cleaner stack.

Good stuff!

Click to expand...

@sfx your are right on, chipset for 802.11ac better horse power to process 802.11ac signals, better EVM in both Tx and RX sensitivity. To decode 256QAM vs 64QAM, for sure 64QAM will benefit the extra processing.
I found in most 802.11ac design, the choice of more powerful PA(power amplifier) really push the signal to the regulation limit, you really need the extra SNR to decode 256QAM. PA much more linear so cleaner signal for receiver so for sure 802.11n is improve, I saw the big difference from AR9390 to QCA9890 when using 802.11n.

zerodegrekelvin said:

I found in most 802.11ac design, the choice of more powerful PA(power amplifier) really push the signal to the regulation limit, you really need the extra SNR to decode 256QAM. PA much more linear so cleaner signal for receiver so for sure 802.11n is improve, I saw the big difference from AR9390 to QCA9890 when using 802.11n.

Click to expand...

Er, some mixed metaphors here. The regulatory limit in the US anyway is many watts (to oversimplify the rules on antenna gain vs. EIRP). Taking a 20 or 40MHz (!) HPA (amp) up by 6dB or more is quite costly and hurts the market price badly. Which (pricing) is why in OFDM with /without MIMO) the practice is to back-off the PA power as the modulation order increases (peak to average ratio increases the RMS power). The regulatory limits in EIRP in most domains is far higher than an affordable product with more than about 0.1 to 0.2W (100-200mW) in the highest modulation mode - (excluding use of a 20-30dB antenna as the regulatory constraint).

stevech said:

Er, some mixed metaphors here. The regulatory limit in the US anyway is many watts (to oversimplify the rules on antenna gain vs. EIRP). Taking a 20 or 40MHz (!) HPA (amp) up by 6dB or more is quite costly and hurts the market price badly. Which (pricing) is why in OFDM with /without MIMO) the practice is to back-off the PA power as the modulation order increases (peak to average ratio increases the RMS power). The regulatory limits in EIRP in most domains is far higher than an affordable product with more than about 0.1 to 0.2W (100-200mW) in the highest modulation mode - (excluding use of a 20-30dB antenna as the regulatory constraint).

Click to expand...

When working with a 10 watt or 10,000 watt PA, it's actually easier than working the the PA/LNA's in handsets and wifi gear... We're not dealing with Watts, we're dealing with mWatts, at most, perhaps 200 to 600 milliwatts of power.

11n was a challenge for many, but they worked thru it - 11ac is more a challenge, as one needs to be fairly linear across an 80MHz spread in 5GHz... and trust me, it ain't easy - heck, it was hard enough at 1.25 MHz channels at 1900Mhz for CDMA/PCS.

@ zerodegrekelvin - stevech is one of those really smart guys I mentioned in the PM, I have a tremendous amount of respect for his knowledge and insight.

sfx

sfx2000 said:

When working with a 10 watt or 10,000 watt PA, it's actually easier than working the the PA/LNA's in handsets and wifi gear... We're not dealing with Watts, we're dealing with mWatts, at most, perhaps 200 to 600 milliwatts of power.

11n was a challenge for many, but they worked thru it - 11ac is more a challenge, as one needs to be fairly linear across an 80MHz spread in 5GHz... and trust me, it ain't easy - heck, it was hard enough at 1.25 MHz channels at 1900Mhz for CDMA/PCS.

@ zerodegrekelvin - stevech is one of those really smart guys I mentioned in the PM, I have a tremendous amount of respect for his knowledge and insight.

sfx

Click to expand...

I have no doubt you guys are smart , I never question that, we are talking about wifi txpower here.
As per FCC for 2.4ghz 30dbm/1W conducted power max + 6dbi ant =36dbm max EIRP.
in 5Ghz FCC just allow lower band to 300mW.
In the enterprise radio from both Atheros and Broadcom, I saw an increase in Tx conducted power from the reference design 802.11n to the 802.11ac, this is coming from the DVT test reports and txpower can also be obtain from FCC report. Let pick the ASUS RT-AC87U, the txpower is massive when you compare to counterpart in 802.11n AP, this is what my Agilent PSA says as I measure.
I also did measurements for QCA9990 wave2 radio and BRCM43465 wave2 radio, PA is using so much power in the overall power budget, this is why your 802.11ac router is much hotter and need massive heatsink.
You can also see the PA being used from FCC internal photos and derive the txpower max from the datasheet if you don't have access to the schematic.

So by having the government regulate the power output of a radio wave, how can someone modify it to be more powerful??
Its like a car, some people hot rod them..
Am I over thinking this?

Steve40th said:

So by having the government regulate the power output of a radio wave, how can someone modify it to be more powerful??
Its like a car, some people hot rod them..
Am I over thinking this?

Click to expand...

No you are not over thinking
There are always ways to "overclock" the txpower if you know how to solder, but more power does not translate into better wifi. Overclocking the power amplifier you run in non-linear zone, more errors from the receiver point of view.
You have to be extremely good a soldering to remove the old PA and put a more powerful one.
FYI, Netgear has been fine $240000 by the FCC for going over the limit