2.4GHz speed differential question

[azazel1024]

My curious mind is trying to figure out an issue I am having with a router I am using outside.

The router is a TP-Link WDR841n located in my garage with the antennas mounted outside by the eves of the garage roof mounted to 3ft coax back to the RP-SMA connectors on the router.

All my devices have a nice strong signal and it effectively covers my entire backyard at good speeds, which my indoor router and AP couldn't manage. The siding is aluminum if it matters (maybe for reflection/back scatter as the antennas are mounted about 3" away from the siding).

Anyway, inside on my two Netgear 3500Lv1 I get around 170Mbps on my laptop (Intel 7260) on the one setup in 40MHz mode and I get around 80-90Mbps on the one setup in 20MHz mode. My tablet (asus T100) I get around 80Mbps on the one in 40MHz mode (tablet is 1:1, laptop 2:2) and around 45Mbps on the one setup in 20MHz mode.

Seems all completely normal and resonable.

Well, the outdoor router, I connect at around 50Mbps on my tablet with it in 20MHz mode. So...nice. However, may laptop bounces all over the place and struggles to get 35Mbps, even though both router and laptop are 2:2, so I'd expect faster than my tablet.

Disconnecting a single antenna seems to make no speed difference with the laptop...so it seems like the router and laptop are not effectively connecting through two streams, or there is some kind of interferance going on when my laptop is trying to receive/send two streams at the same time with the router.

I've play with the firmware and I've played with most of the router options and nothing seems to improve the situation.

I am just wondering if anyone might have any suggestions or thoughts. It isn't that big a deal, because I pretty much don't use my laptop outside, just my tablet and phones, which all connect to it well and fast, but it is bothering me why my, should be, much faster laptop is struggling with the connection when no other device is and when it can clearly manage much faster speeds than any of my other wifi devices can when connected to a different router/AP (but still same 2.4GHz 2:2 limitation).

 

[JackMDS]

As a first step disconnect the extended Antenna from the TP put back its original Antenna go with the Laptop into the Garage and see how the laptop performs there.

 

[JoeMcJoe]

As a first step disconnect the extended Antenna from the TP put back its original Antenna go with the Laptop into the Garage and see how the laptop performs there.

I agree.

By using the cable extenders you are most likely reducing the tx/rx powers.

 

[azazel1024]

It might be introducing noise, but it isn't noticable reducing Tx/Rx power. Without the extenders I get -25dBm at roughly 3ft from the antennas in the garage, with the extenders standing outside 3ft from the antennas I am getting -25dBm.

I do need to attach the antennas directly to the router and test in the garage to see about speed differences though...but just I don't know what could be going on here.

I did see one user on the TP-Link forums (which have not been a help) with an identical setup (minus relocated antennas) with the WDR-841n and Intel 7260 that was staring down ~90Mbps uploads and ~30-35Mbps downloads. I haven't tested uploads on this to see what results, but it is making me think that maybe something is going on between the specific card and the router that one or the other really just doesn't like.

Probably need an hour or two to test the various firmwares and all of the settings I can think up on both ends to see what I can manage.

 

[JoeMcJoe]

I have a TP-Link 1043nd, terrible wireless performance, just awful. I don't use it anymore.

As always, wireless performance depends on many things, your location, wireless devices around you, ...

Only enable 40 Mhz mode on 2.4 Ghz 802.11n if you have zero other devices around, zero. Otherwise it introduces more problems. Use 20 Mhz.

If you want a good outdoor signal, get an outdoor access point.
An example. http://www.amazon.com/Ubiquiti-PicoS.../dp/B0055PKSG6

 

[JackMDS]

TP-Link is good for gamblers.

The price is low because it seems that their QA is low too. As a result every piece of hardware is a gamble, it might be Good but chances are high that it is Not.

However, it might be that none of the Entry Level Wireless Router would do well as the installed.

So first check the Wireless under normal condition.

 

[CubanlB]

My guess would be that the intel is a lot more sensitive to frame re-transmits and is trying to reduce the dynamic rate to deal with that.

I wouldn't set the the outdoor AP to 40Mhz channel, as you are almost certainly getting some co-channel interference between your 2 APs. (another possible source of re-transmits).

How far apart are your 2 outdoor antennas? For proper diversity reception the should be a wavelength apart. At 2.4GHz that would be 4.92 inches. (I think you are ok if you keep spacing at some multiple of the wavelength, though I'm not and EE)

 

[azazel1024]

Its set to 20MHz. My indoor access point is set to 40MHz.

Antenna spacing is roughly 6-8 inches apart.

I doubt there is any cochannel interference. I've run tests and nothing steps on anything else.

I have the main indoor AP (Netgear 3500l) set to 2+6 (my tablet has issues if it is set to 1+5...for some odd reason), my basement router (Netgear 3500l) set to 11 and the outdoor 841nd to 11.

I've run my laptop sitting on the indoor AP at max throughput and my tablet sitting outside and seperately on the basement router at max throughput and neither device slowed down compared to only one operating.

The basement and outdoor APs pretty much have no physical overlap with their coverage so, safe to use the same channel. The closest either gets is roughly 20dB of each other, and generally, especially if close to either one, the signal strength difference is closer to a 60dB difference.

I would switch the outdoor AP to 40MHz, but I am mildly worried about the small amount of physical overlap in coverage between the two. Its a very small slice, but it is there.

You may be on to something with the Intel. Not sure if there is much I can do about it though. Maybe DD-WRT on the router? See if it is firmware related?

 

[CubanlB]

Is this all with one SSID? Notebook associating with the inside AP sometimes?

What tests are you running to verify coverage?

When you say at worst there is a 20db difference between the to AP on ch 11, is this from the client perspective about half way between?

The basement AP could very well be raising the noise floor for the TP link AP. You would want to see how much the TP link can hear the basement AP on the same channel, though, this may not be possible.

Are the dBm outputs of the client computer and the TP link AP matched? (Usually 20dBm for home stuff)

This may seem counter intuitive but you could try forcing the outside AP into b/g mode and setting the max data rate to say 36mbps and see if it is more stable (though throughput will be poor it might be better than it constantly changing) You can probably ignore this part as you may not be able to set available rates, though forcing g mode (ERP-OFDM) may give better results.

 

[azazel1024]

I am running inSSIDer to ensure it is associated with the outdoor AP, no chance it is reassociating there. Signal strength is ~-35dBm on the outdoor AP and ~-75dBm on the indoor AP during testing.

Single SSID for the whole WLAN.

For the ~20dB difference between the basement and the outdoor AP, it is when standing near the corner of my house near the basement AP. Outdoor AP reads as ~-68dB and the basement AP is around -85dB. Rounding the corner of the house and the basement AP bumps up to around -70dB and the outdoor AP drops off to around -80dB. So in some cases its a bit less than 20dB difference, but only in about a 5-10ft space outside. Move a couple more feet and less of the house is between the basement AP and the client and more is between the outdoor AP and the client and its up to low -60dB range on the basement and high -80dB range on the outdoor AP.

I cannot attempt to bridge the basement and outdoor APs. From their perspective locations the signal strength ranges from -89dBm to dissappears completely from inSSIDer. So at least from the routers perspective, the other is completely hidden. From a client perspective, there could be noise floor issues, but only in a narrow slice outside. Where I am testing the basement AP is showing up as high -80dB range to hidden.

I can force modes and rates on the TP-Link. I briefly tried 11n only from mixed, but that was on the latest firmware, which seems to have LOTS of problems. I didn't try on the current (one step back) firmware.

It seems like my options are, tweak settings on my client (not a lot left to tweak), change from mixed mode, to 11n only, try a different channel (though channel 1 and 2 seem to not work at all on the AP. I'll get speeds on all devices from 1-10Mbps, 11 seems to work great on my phone and tablet, but weird on the laptop) and maybe try a slightly lower max rate to see if it'll do anything.

11g probably won't be a good option, as even on the laptop, my lowest speeds are roughly the max 11g can put out. It ranges from 2.4-4.5MB/sec...and the best 11g can do is around 2.7-2.8MB/sec. Average currently is probably about 3.8MB/sec I'd guess. there are no drop outs or hangs, just very irregular speeds and not nearly as high as I'd expect (my tablet is pretty rock steady in comparison at 5.5-6.2MB/sec with no real dips above/below this).

The weather this weekend looks nice, so I'll hopefully be able to grab an hour or two to test out a variety of firmware and setting configurations.

I can also try out the previous Intel driver version. I am up-to-date on it, but interestingly, when I did update, it was frozen on 150mbps (indoor AP) until I changed the driver to 20MHz only and then back to auto on the client and then it would select 300Mbps. Disabling/reenabling the adapter and rebooting didn't fix it and its stable since toggling that...but I am wondering maybe if there could be an issue in the latest Intel driver (speeds are no better or worse on the indoor APs since driver updates, but I never tested the laptop outside on the prior driver version).

 

[JoeMcJoe]

Only use channels 1 or 6 or 11, nothing else as they're the only ones that don't overlap.

A correctly operating wireless device will revert back to a single channel 20 Mhz if it detects interference in the 40 Mhz double channel configuration.

Use 20 Mhz for 802.11n on the 2.4 Ghz band for improved reliability.
If you want speed, use the 5Ghz band for short ranges and much less interference.

 

[azazel1024]

No 5GHz basestations, so not an option.

On 2.4GHz, my one AP in 40MHz mode and 2+6 doesn't interfer with the others on channel 11 and 20MHz.

There is no overlap with 2+6 and 11 with the way 11n OFDM does channel bonding for 40MHz. There is POTENTIAL interference with that scheme, but the likelihood is low, because the transmit power has to be so much lower on both ends when they are edging in to each other with 11n OFDM (20dBm lower is the MINIMUM, IIRC).

In testing I've been able to produce no interference, which lines up nicely with what the wifi specs say. Nearest neighbor networks internal to my house are <<50dBm below the strength of either of my indoor routers.

I do not use my basement or outdoor APs in 40MHz mode because of the overlap. Outside I could probably get away with it as the basestations have such a huge difference in signal strength between them and most outdoor locations, the outdoor AP is much, much stronger than the indoor AP, except a small slice of the outside and all indoor locations are much stronger on the indoor AP than the outdoor AP (by 30+dBm). Heck, even nearest neighbor networks are <20dBm of the signal strength of my outdoor AP short of hoping the fence in to their yards, and is generally 30-40 dBm higher.

Eventually, I will use 5GHz. Not today though and probably not next month or in 6 months. The plan is to replace the two indoor units with C7 Archers. The outdoor is going to be more of a wait and see. I'd like to do an 11ac product too, but outside I care a lot more about "just working", not about file transfers.

Currently, the TP-Link 11n router I am using does work okay for what I want, its just one of those unexplained mysteries to me why it isn't working better with my laptop and I'd like to resolve it if possible. If I can, it still works fine.

Eventually I'll replace it with something with concurrent dual band 11n at least, so long as the router has dual band external replacable antennas. I'd go with a C7 archer outside too, but the 2.4GHz being internal only is a sticking point (and I do NOT want to have 6 antennas sticking through the wall, 3 max). So it might be an N600 unit (WDR3600?) for a couple of years till things settle down more with 11ac and there are more options (an AC1200 unit with external dual band antennas would be perfect).

 

[CubanlB]

Yeah, you don't seem to be doing anything obviously wrong with the setup. Really the only difference is the client radio at this point. Tablet and phone may not be doing antenna diversity, but the fact that the chipset and driver are going to be very different from the intel in the notebook makes it really hard to narrow it down to one feature on the intel.

Capping the actual 802.11 frames may give you some indication about what negotiation is actually going on between the TP link and the intel, especially if you can get the phy headers of the frames. but that might be more work than might be warranted.

http://wiki.wireshark.org/CaptureSetup/WLAN

 

[azazel1024]

I may run wireshark. I did reveal some interesting things about a USB3 gigabit ethernet adapter I was attempting to use and figuring out why the receive rate was so abymally low (dropped packet central! Like 25+% were being dropped and having to be resent to the adapter).

Yeah, I can't even point to it and say everything is Broadcom except the notebook. The tablet and iPhones are Broadcom based, and my netgear 3500l are broadcom based...but the TP-Link has an Atheros chipset (just like most/all of TP-Links routers)...and the Intel plays VERY nice with the broadcom routers.

I guess Saturday's plan will be to setup on my deck, crack a beer and start playing with everything.

 

[CubanlB]

I guess Saturday's plan will be to setup on my deck, crack a beer and start playing with everything.

That sounds horrible, feel bad for ya...

Wireshark has some specific requirements to be able to collect certain types of information. a Nix or BSD environment might be beneficial.

I just realized that I picked up the same TP-Link for my aunt, now I'm going to have to play with it for a while before I give it to her...

 

[azazel1024]

Yeah, I could have a defective unit. I am not beyond thinking that, but Channel 1 and 2 would just not work at all. I have no bloody idea why. I didn't try in 40MHz mode selecting 1 or 2 as the start, but in 20MHz, nothing. Nice strong signal, but I'd get a few seconds of 400-1,200KB/sec, and then it would drop out for 10-15s before I'd get another burst, then drop. That was on all of my devices. Channel 11 was/is okay, other then the odd laptop issue. I may try 6, but not ideal since that is sharing spectrum with my familyroom AP, where there IS some overlap in client coverage, not a lot, but some.

Even if I can't get it working "right", it is working okay. I just don't like odd issues, so until I either solve it or decide it is insolvable, I am going to plug away at it. I mean, I got it for $20 on sale, so I don't have much heart burn about it.

My only "heartburn" is it does make me question getting the WDR3600 to replace it in a year or so...but that generally seems to have very positive reviews. I'd also go with a no hassle return place and test the heck out of it early and just return it if I have any odd issues, but still makes me nervous (though I've had very good experiences with all of the other TP-Link gear I have/have had).