More speed/bandwidth from an 802.11n laptop<-->WiFi Router/Modem connection?

[IainB]

I have set up someone's TP-Link TD-8950ND 150Mbps Wireless N ADSL2+ Modem Router.
It all seems to be running fine, except that the laptop I am using is only making 30-58mbps connections (rated as being "Good" to "Excellent", respectively). This is currently with just the 1 laptop connected to the network.
The complex screenshot below should explain all. (Thankyou @Mouser/Screenshot Captor.)

What I am wanting to figure out is how I can get the laptop performance to be (say) 100mbps or more.

The TP-Link router/modem seems to have the latest firmware (I checked):

• Firmware Version: 1.2.6 build 101206 Rel.68715n
• Hardware Version: W8950ND vi 00000001

The Broadcom WiFi device has the latest driver version (I checked).

I suspect the problem may be to do with a lack of draft-n standardisation:

• On the cardboard box the TP-Link device came in, it says it is "Wireless Lite N".
• The Broadcom WiFi device driver says it is "802.11 a/b/g/draft-n"

Any ideas please?    

[attachthumb=#]

[rgdot]

I think you will end up with 'advertised speeds are theoretical' no matter what you try.

[4wd]

^What he said plus you've said nothing about the environment you're trying to operate in.

ie.
How far from router?
Other close electrical apparatus, (especially stuff operating in the 2.4GHz area)?
What obstructions, (walls, etc), and construction?
Are the antennas polarised in the same direction?
Did you have your fingers crossed as a meteor flew overhead, etc, etc, etc.

You should be able to get more than that but it can come down to something simple like the devices don't like each other, which is a PITA trying to determine when they individually work OK with other devices.

[rgdot]

Of course 4wd is right. Closer to router, less walls in between, wireless chipset (latest firmware or otherwise) .... whole spectrum of things affect wireless speeds.

[40hz]

^+1 with all of the above.

One thing you might try is disabling IPv6 support on all your devices if you're not using it. (Most people aren't.) In some rare cases you'll get a slight performance boost since it's one less thing your hardware has to pay attention to. Hardly worth it IMO unless IPv6 is actually causing problems on your network - which I doubt it is - but that's about all I can suggest.

[IainB]

The rate seems to fluctuate  between 32 and 52mbps, regardless of whether you are up close to the router or 9 meters away and with two plasterboard and steel-framed walls in between. I have seen momentary peaks of 58.5 and 59.0mbps at the 9 metre point and momentary troughs of 23mbps.
Any other appliances involved? Probably not, because:

• Switching off/on a nearby TV doesn't seem to make any noticeable difference.
• Switching off/on IPv6 support for the Broadcom WiFi device doesn't seem to make any noticeable difference either.

I don't know which is the weakest link.
I could try:

• checking the performance of the ADSL2+ connection for any constraint.
• fitting a newer model WiFi device with "full" 802.11n support, in the laptop.
• installing a 150 or 300mbps router with "full" 802.11n support.

I suspect that it may be that you'll never get the 100+ mbps rates unless both the router and the WiFi device have identical or "full" 802.11n support, but that's just a guess. I don't know enough about this technology to figure out a fix.

[Carol Haynes]

I have struggled with this in the past and came to the conclusion that only devices designed to work together (whether full or draft n spec) by the same manufacturer get anything like the advertised bandwidth.

In the end I just gave up worrying about it as there doesn't seem to be a manufacturer that produces a complete range of suitable device. In particular internal mobile adaptors (eg. laptop PCI or PCIe adaptors) seem rather difficult to find and mostly don't seem to achieve full throughput.

[4wd]

The rate seems to fluctuate  between 32 and 52mbps, regardless of whether you are up close to the router or 9 meters away and with two plasterboard and steel-framed walls in between. I have seen momentary peaks of 58.5 and 59.0mbps at the 9 metre point and momentary troughs of 23mbps.

I'd be tempted to try a cheap 802.11n USB adapter in the laptop, (to eliminate the Draft-n component).

FWIW, I've used this, this and this, (all of which are Realtek 8188/8192 chipsets), to connect a WDTV Live to my ASUS RT-N16 over 12-15m through two plasterboard walls, a 50cm CRT TV, a couch and a CD rack, they all connect at 150Mb/s.

After searching a bit, it seems to me that "802.11n Lite n" just means it's 150Mb/s as opposed to 300Mb/s - 600Mb/s, (ie. it's still full 802.11n compliant, just that because it's single antenna it's speed is limited).

[IainB]

@4wd: Thanks for the suggestion, I might try one of those USB WiFi adapters, though I suspect I might just repeat @Carol Haynes' experience that:

...there doesn't seem to be a manufacturer that produces a complete range of suitable device.

Since making the OP, I have discovered that I was mistaken in thinking I had the latest firmware update to the TP-Link router. I had misunderstood the version numbering, which is apparently based on a short-form date:
The one in the router was Build 101206 = 2010-12-06 (v1.2.6).
I downloaded and installed Build 120406 = 2012-04-06 (v1.3.1).
The install eventually worked - I had to use IE to install it as the install kept failing in FFX (the install notes recommended to use IE, but I had to try).   

I also DISABLED IPv6 (using MicrosoftFixit50409 - DISABLE IPv6.msi) and then ENABLED IPv6 (using MicrosoftFixit50440 - ENABLE IPv6.msi) just to be sure get everything back to "normal".

However, the router still seems to fluctuate mostly between 39 and 58.5mbps.

[kyrathaba]

However, the router still seems to fluctuate mostly between 39 and 58.5mbps.

Mine does the same after much the same fiddling. I'd say Carol's conclusion above is pretty spot-on.

[Carol Haynes]

Also I don't really find TP-Link hardware all that consistent. I have seen a number of identical models that seem to perform differently in the same location - wonder if it is a quality control issue?

The most consistent results I got were from Netgear Rangemax hardware where all machines used compatible Netgear devices from the same range.

What are you actually trying to achieve - fast network file transfer?

By the way looking at your initial post again the WiFi quality isn't 100% which can affect speeds.

There are some chipsets that don't play well together - so there may be a conflict between the Broadcom device and the router? Not sure what chipsets TP-Link use.

I used to have some problems with Intel chipsets and that was a conflict with a router chipset.

[40hz]

Many times you can get around snarky problems, and get significantly better performance and features, going over to a third-party firmware replacement such as Polarcloud's Tomato or DD-WRT.

Right now I'm going through a Linksys WRT150N that was previously sitting in a client's junk bin because it was unreliable. It required frequent reboots, bogged under more than one user, and was dog slow (28-37Mbs) as shipped from the factory.  When subsequently upgraded with Linksys's new firmware (on the recommendation of Linksys's own tech support) - that finally broke the thing once and for all. However, with DD-WRT (v24 build 13575) installed, it's rock solid and gives me a very steady 117Mbs with 55% signal strength after going through a floor and three walls. Not too shabby for some freeware installed on a piece of equipment that was slated for the scrapheap.

That said, this solution is not for the faint of heart even if it isn't a technical challenge to install either of the above. It's only when an installation goes wrong that it can get extremely...um...educational. You haven't really lived until you've needed to learn the differences between tftp, tftpd, ftp, ncftp, ftpd, and pure-ftpd to do a reload - or better yet - had to fire up a soldering iron to create a JTAG cable and then attach it to the circuit board of a cheapo bricked router!  <*GRIN*>

Doing a DD-WRT or Tomato load isn't an option here unfortunately. The TP-Link TD-8950ND isn't listed as a supported device for either package.

[4wd]

Not sure what chipsets TP-Link use.

Broadcom in the TD-8950ND.

Agree with 40hz about the firmware, ASUS RT-N16 WiFi was reasonable but still had the occasional drop-out, reflashed with Tomato and it's been rock solid....now if only I could do the same with the power company 

[IainB]

...What are you actually trying to achieve - fast network file transfer?...

...Polarcloud's Tomato or DD-WRT...

...Broadcom in the TD-8950ND...

All I am trying to do is get more of the supposed spec. performance (150mbps) out of the TP-Link TD-8950ND. The highest I have seen it momentarily go to was 72mbps, so far. The signal quality seems to vary from Good to Excellent, regardless of how close to the xmitter the PC is or what the mbps rate is.    

Yes, it would have been nice if I could have used something like the Tomato approach. Nothing would daunt me in that regard, and, having been an electronics hobbyist since age 10 and having built various bits of kit (e.g., including things like an amplifier, preamplifier, PLL VHF tuner, two-valve superhet SW tuner, calculator, multimeter) and having modded the PCBs of various bits of kit -  including laptops - I am quite happy wielding a soldering iron.

I hadn't known it was a Broadcom chip in the TP-Link TD-8950ND. I couldn't find it mentioned in any of the documentation that I have (Data Sheet, User Guide, Installation notes). The laptop's WiFi is a Broadcom device also, and I have triple-checked that the drivers for that are up-to-date. Shouldn't they work in harmony?    
Yeah, right.

[40hz]

Yes, it would have been nice if I could have used something like the Tomato approach. Nothing would daunt me in that regard, and, having been an electronics hobbyist since age 10 and having built various bits of kit (e.g., including things like an amplifier, preamplifier, PLL VHF tuner, two-valve superhet SW tuner, calculator, multimeter) and having modded the PCBs of various bits of kit -  including laptops - I am quite happy wielding a soldering iron.

"Hail and well met" fellow builder and circuit bender!

[4wd]

I hadn't known it was a Broadcom chip in the TP-Link TD-8950ND.

Oops! I only got half, it's Broadcom, (for the wired), and Atheros, (for the wireless) - I'll see if I can narrow it down a bit.

The laptop's WiFi is a Broadcom device also, and I have triple-checked that the drivers for that are up-to-date. Shouldn't they work in harmony?    

Apart from the obvious that the laptops device is pre-802.11n spec  

Another reasonably cost effective thing to try would be replacing the antenna on the router with a higher gain omni-directional.

Or try a beer can.
[NOTE: There are some semi-naked ads on the page.]

[kyrathaba]

I sometimes get green with envy when I hear about the performance improvements with Tomato or DD-WRT, but I'm afraid to mess with my currently working setup. I have a Linksys RangePlus WRT110 router, and I can watch its properties and see speed fluctuate between 18 Mbps and 54 Mbps within sixty seconds:

[attach]

[lanux128]

i use a TP-LINK device as well, and i find it helpful to set the bandwidth mode to "11n-only", instead of the default "mixed b/g/n mode". of course this is assuming that you don't have any old devices that are still using 802.11 a/b/g.

[IainB]

...Apart from the obvious that the laptops device is pre-802.11n spec...

Yes, this had already been covered (draft-n), but there was a suggestion that - had they both been Broadcom - then they might have been able to play together nicely.

...and see speed fluctuate between 18 Mbps and 54 Mbps within sixty seconds...

I've hunted around for a little Systray WiFi monitor app that can bring up a floating window and show just the collected basic status details about the current WiFi AP connection - per the image in the OP - such as SSID, standard (e.g., IEEE 802.11n), the value for mbps (say) every ½-second or so, and the bars for signal quality/strength(?). However, the closest that I could find was an app that only worked in Win XP. There were other apps (Nirsoft has some good ones) that monitor all the WiFi APs within range, but nothing else quite like what I have described above.

...Another reasonably cost effective thing to try would be replacing the antenna on the router with a higher gain omni-directional.
Or try a beer can.

Yes, I already have a design in mind using some cardboard and ally foil. It will certainly affect the signal quality/strength and probably therefore the mbps value too. The beer can idea looked dead simple, but I'm not sure it will fit nicely on the wall-mounted router I am using here. I shall try it anyway.    

...i use a TP-LINK device as well, and i find it helpful to set the bandwidth mode to "11n-only", instead of the default "mixed b/g/n mode". of course this is assuming that you don't have any old devices that are still using 802.11 a/b/g...

Yes, I have set it to "11n only", and the router stats show that it can see the MACs for all the devices it needs to support - including a Kindle (I hadn't known the Kindle supported the 802.11n standard). All the devices (including the Kindle) seem to work just fine.

Oh, and I also tried changing the Wireless Country/Region (which is a bit naughty), in case the transmission strength might vary, but that didn't seem to make any significant difference, so I restored it to my region.

[40hz]

I sometimes get green with envy when I hear about the performance improvements with Tomato or DD-WRT, but I'm afraid to mess with my currently working setup. I

@K - If it's working for you, and you're happy with the performance you're getting, I'd skip ever doing a firmware swapout. The only time I'll use Tomato or DD-WRT is if I'm either working with a throwaway - or I if bought a specific box (like a WRT-54GL) to put it on. Same goes for the manufacturer's firmware upgrades. If your box is working - don't. Only upgrade to attempt to fix a problem you're actually experiencing. (Especially if it's Linksys firmware.)

FWIW, most out of the box gains are modest once you switch over. You really do need to tweak things up to get the level of performance many claim to be getting out of DD-WRT. (I also suspect that many of these claims - other than for stability and up time - are just a tiny bit exaggerated. YMMV)

[40hz]

@IainB - did some looking around and talked to a few people yesterday. Two of my small business clients use the exact same router. One client loves the thing. Called it the best WAP they ever owned. Fast, reliable, set & forget etc... The other has three of them. Said two worked like a charm. The third was nothing but trouble, Had to get it replaced twice before they got one as good as the other two.

Sounds like maybe there might be some QC issues with the manufacturer. Considering the price/spec ratio (it is a bargain) I could see where the maker might figure it's just cheaper to replace defectives than it is to properly QC them before shipment.

Looking at the distribution curve of customer ratings, it looks like they're most heavily clustered on the very high and low ratings which says to me "jellybean" manufacturing.

Any chance of getting a replacement unit before you pull too many hairs out?

[Carol Haynes]

As I said in a previous post that has been my experience of their products too.

[40hz]

As I said in a previous post that has been my experience of their products too.

Ok IainB - Looks like there's consensus on both sides of the pond for that!

---
@C - sorry. Credit where due.

[kyrathaba]

Same goes for the manufacturer's firmware upgrades. If your box is working - don't. Only upgrade to attempt to fix a problem you're actually experiencing. (Especially if it's Linksys firmware.)

@40Hz: Yep, good advice. Agrees with what my gut tells me.

[40hz]

Yes, I already have a design in mind using some cardboard and ally foil. It will certainly affect the signal quality/strength and probably therefore the mbps value too. The beer can idea looked dead simple, but I'm not sure it will fit nicely on the wall-mounted router I am using here. I shall try it anyway.

Let me know how you make out with the foil and cardboard approach? I tried a few published versions of those (and one or two of my own design) on a Linksys WRT-54G. For the most part, I didn't see all that much performance difference with any of them in place.

The cantenna projects discussed here were a different story however even though they're your basic waveguide/directional designs. I've done up a few of the Hunt's 26.5 pasta sauce version - mostly because I was able to locate a couple of cans with the exact same dimensions. I was very happy with the results.

Cantenna how-to and design information here if anybody else is interested. (I say anybody else because I'm sure IainB has already found this - and at least fifty-seven even more interesting and informative pages on the subject - if I know IainB. )