Wireless Mini Pci Express Cards

[kiteboy]

Hi

Just about to buy a new router - possibly a cisco e4200v2 when it reaches us here in the UK - its replacing an old wrt54gs - so its about time for an upgrade

The question is which wireless client would best utilise the new router?

Ive just bought a Intel Advanced N 6200 (as it was used in the Wndr4500 retest)

Anyway thats working fine but not sure if it was a decent choice really

I have an old Dell m1530 which has three wires to connect up to the mini pci card (which the Intel 6200 card has two connections - so not sure if that will lose me some performance)

Any recommendations?
Thanks

 

[claykin]

This page from the Dell support site should help you.

http://support.dell.com/support/edocs/systems/xpsM1530/en/SM/minicard.htm#wp1180188

Your 6200 has two antenna ports, so you'd use the white and black wires to connect to their respective connectors on miniPCI card.

If you purchased a 3x3 miniPCI such as the Intel 6300 you'd be able to connect the gray wire to the third connector.

Keep in mind that many Dell laptops are BIOS locked and do not allow nonOEM miniPCI cards to work. For some models you can find BIOS hacks online to work around that issue. Hopefully your model is not BIOS locked.

 

[kiteboy]

This page from the Dell support site should help you.

http://support.dell.com/support/edocs/systems/xpsM1530/en/SM/minicard.htm#wp1180188

Your 6200 has two antenna ports, so you'd use the white and black wires to connect to their respective connectors on miniPCI card.

If you purchased a 3x3 miniPCI such as the Intel 6300 you'd be able to connect the gray wire to the third connector.

Keep in mind that many Dell laptops are BIOS locked and do not allow nonOEM miniPCI cards to work. For some models you can find BIOS hacks online to work around that issue. Hopefully your model is not BIOS locked.

The wireless card I bought does work so I assume its not bios locked

maybe the 6300 would give better performance then - will look into

thanks for the info

 

[thiggins]

You're really not going to see a big difference moving from an Intel 6200 to 6300. Save your money.

 

[rhombus]

You're really not going to see a big difference moving from an Intel 6200 to 6300. Save your money.

Are there some tests on this?

He can probably return the 6200 and buy the 6300 for £8 more here in the UK, so I'd say why not.

 

[claykin]

I agree with Tim. Right now 3x3 doesn't make much difference. Only case (sometimes) is if the client is very close to the AP/router. Even then its hit and miss.

 

[rhombus]

Right now 3x3 doesn't make much difference. Only case (sometimes) is if the client is very close to the AP/router. Even then its hit and miss.

Well, I take an empirical approach What controlled (as far as possible) testing have you performed or seen to support this?

Sure, it's possibly or even probably true (although, I think more due to RF non-line-of-sight characteristics rather than raw distance). I'm not saying otherwise, nor am I suggesting anyone rushes out to buy a 450Mbps card.

But, we are dealing with a different card (6200 vs 6300), and an extra antenna - so even at 2-stream they might perform differently, let alone with a 3-stream capable router.

Bung them in the same laptop, and test with a 3-stream Linksys E4200v2 and then a 2-stream Linksys E4200v1. Maybe add in an Intel 5300 you have lying around, and a requested BigFoot 1102 and 1103... and you've got one hell of an article I'd say!

 

[kk22]

Tim's review of the E4200V2 actually has comparisons between the 6200's 2-stream connection vs the 6300's 3-stream:

http://www.smallnetbuilder.com/wire...al-band-n900-router-reviewed?showall=&start=2 (scroll to the bottom for the 2-stream vs 3-stream summary)

The WNDR4500 Re-Test review also has the 6200 and 6300 tested.

If the price difference between the 6200 and 6300 is small and you don't mind paying the extra, plus you have 3 antennas anyway, just get the 6300 and be done with it.

 

[rhombus]

Tim's review of the E4200V2 actually has comparisons between the 6200's 2-stream connection vs the 6300's 3-stream

But they are hosted in totally different laptops...

 

[thiggins]

Are there some tests on this?

Just look at the E4200V2 review data. Yes, the Intel 6200 and 6300 are in different notebooks. But I think the data is pretty clear.

 

[rhombus]

Just look at the E4200V2 review data. Yes, the Intel 6200 and 6300 are in different notebooks. But I think the data is pretty clear.

How can it be clear?

This is poor testing protocol and poor conclusion.

As was using a 3-antenna client with one antenna not attached all this time. I don't think with even a dummy load in its place. With the best of intentions, your whole charts are null and void!

As is the idea that wireless almost inevitably drops to zero for over half a second many times within a 60 second test period. If this is happening, you have a huge fluctuating signal problem or worse or probably due to the former mentioned issue (did you check the corresponding RSSI plots on IxChariot?)

In my opinion, a lot of your wireless conclusions are highly flawed.

 

[claykin]

:

In my opinion, a lot of your wireless conclusions are highly flawed.

Please post the URL to your test website so we can read your professional comparisons.

 

[rhombus]

Please post the URL to your test website so we can read your professional comparisons.

You can make fun, and yes I am forthright, but I actually make some serious points. Going forward, it's up to Smallnetbuilder whether they are considered or dismissed.

It's one thing to say "these are the results with x and y". Everybody can be happy with that. But it is quite another to then draw and repeat firm generalised conclusions about underlying technology and products, and apply consumer advice based on that, when the test comparison is fundamentally and potentially significantly unfair. I actually am a fan of the tempered sceptical SNB attitude to the latest and greatest, I have been around the block myself, be in no doubt. But unknowns are unknown until properly disproved.

So. That said - we both know that I don't need to be a professional reviewer running my own review website in order to perform high quality, thorough, controlled, wireless testing and to use it in practice on a daily basis. Nor do I need that to spot poor testing procedures and conclusions.

I could if I so wanted provide 24 hour wireless to wireless IxChariot plots, one end of a property to another, with not a single so-called "dropout". Conversely, I could also recreate bandwidth fluctuations on demand by manipulating the RF environment. But this is entirely avoidable, and should be controlled in a wireless test procedure - they should not be a regular feature with the products and environment being used here.

I also know from practice that "dropouts" do not occur almost inevitably. If they did, my highly-specialised use of wireless every day for work (which demands an ultra-reliable and medium bandwidth connection) would simply not function. Nor would products like low-latency consumer wireless video streamers (one 1080p HDMI "extender" is reviewed even recently in Smallnetbuilder), or wireless cameras and surveillance (which form the bulk of many specialist wireless companies' portfolios), even get off the ground.

Smallnetbuilder's results using different laptops show little to marginal increases using the laptop with the 3 stream adapter. It has now been dismissed that the laptop needs to be controlled, despite the obvious impact of the laptop and its antennas, and despite a recent article from SNB itself showing the importance of the laptop. In that article, you will see test results from Netgear showing 3-stream throughput significantly higher than is even theoretically possible with a 2-stream adapter. Therefore, I do not think firm universal conclusions should be drawn comparing SNB's "2-stream vs 3-stream" results. Who knows how such a comparison will fare in a real-world and demanding RF environment, I have not performed this testing myself yet or seen anything controlled or comprehensive. But I may do it myself one day, yes...

Thanks
Best of luck everyone
Rhombus

 

[sfx2000]

rhombus raises a good point in that apples should be compared to apples, and not oranges.

I don't have a problem with using different CPE as long as both are characterized from an RF perspective. As Tim has noted in other articles, different laptops have different RF characteristics - antenna placement, RF loss, antenna gain - these all effect the link quality.

Ideally, Tim would have something like the Azimuth test bench, where the AP's and STA's can be isolated, and the RF environment controlled - Tim has had access to platforms like this in the past... and it did give objective results.

I've appreciated his commentary, the device teardowns, and the subjective testing that gives a rough indication of what to expect the performance of a given device to be. It likely is not the final word, but a good indication...

3 by 3 is fairly new, and the chipset drivers, implementations, and RF work has many chances to improve over time. We've seen this with the 2 by 2 devices, where updating the STA driver has dramatically changed test results - go see the WRT160NL for example.

 

[thiggins]

Ideally, Tim would have something like the Azimuth test bench, where the AP's and STA's can be isolated, and the RF environment controlled - Tim has had access to platforms like this in the past... and it did give objective results.

I did use Azimuth for about a year. The results were interesting to device and equipment manufacturers who understood the throughput vs. loss plots. But non-techies just wanted to know how far APs/routers would reach and the throughput they would get.

Yes, rhombus raises a valid point about comparisons. I think I am more careful with my testing than he gave me credit for, but he is entitled to his opinion.

I try to keep my conclusions and commentary fact-based and backed by data. And I'll agree that I am taking somewhat of a leap by drawing my conclusion about three stream vs. two stream performance from two different client platforms.

Perhaps a better way to put it is that I don't yet see the 50% throughput improvement that the switch to 450 vs. 300 Mbps maximum link rates would imply. Part of this could be the difference in laptops. Guess I'll have to slap the ol' 6200 card in the X220i and go get the data.

 

[rhombus]

Guess I'll have to slap the ol' 6200 card in the X220i and go get the data.

OK, but you'll just be introducing another variable. Like I said, I recommend the Intel 6300 over the 6200 not just because it is 3-spatial stream, but because it is a very different and superior card in design. The 6300 is not just a 6200 with an extra radio.

Muddling cards about can only help answer the question as to the better performing card. Yes, it is a very practical question, and a great test to carry out. But there is a much better method to truly test spatial streams.

 

[Andrew LB]

Keep in mind that many Dell laptops are BIOS locked and do not allow nonOEM miniPCI cards to work. For some models you can find BIOS hacks online to work around that issue. Hopefully your model is not BIOS locked.

I used to have a Dell XPS m1530 before I commented to a friend about how I had to have been 'one of the lucky ones' because the G92 core nVidia graphics processors Dell used had a defect and were prone to frying the motherboard. Well... about a week afterward, the laptop no longer worked.

That laptop isn't bios locked and has a total of three mini pci-express slots. Two of them can be fitted with wifi cards and the third only works with a cellular network card unless you want to swap out the antennas for that slot since they're the wrong frequency.

My XPS m1530 came with an Atheros based 4965AGN wireless-n card and I ended up adding a second card, a 1503bgn broadcom chipset which allowed me to use either both cards at once while running windows (2x300mbps) and the broadcom card by itself while running Mac OS X. Broadcom chips tend to get recognized as Apple Airport Extreme by the OS. (lookup iAtk0s for instructions)

 

[AndCycle]

okay ... I just made a personal test with

ThinkPad X201s,
Intel 6200/6300,
Buffalo WZR-HP-G300NH/WZR-HP-G450H

and my conclusion is
yes, 6300 do works better in all condition, and better with 3-stream AP,
although there is not much gain

here is the chart


my test done with iperf-2.0.5,
I put two AP side by side, but only power on one of them during the test,
laptop is only 2 meters away from them,

the result might not accurate at all, I got lots of interference during the test,
G300NH always crashed half way of 40mhz test,
iperf do crashed sometimes, and it's not really reliable as I thought,
missed few result but I think that's ok,

for the interference problem by neighborhood AP,
I keep each transfer for only 10 secs as default, trying to keep test window small,
do 100 iteration and choose only top 5 value for average,

the attachment is my raw data and script for reference,
hope this helpful

 

[StratmanX]

iperf do crashed sometimes, and it's not really reliable as I thought,
missed few result but I think that's ok,

What makes you think it was iperf that crashed and not a signal dropout?

for the interference problem by neighborhood AP,
I keep each transfer for only 10 secs as default, trying to keep test window small,
do 100 iteration and choose only top 5 value for average,

the attachment is my raw data and script for reference,
hope this helpful

10 seconds seems a short time. Netgear recommended the following settings when I did some testing for them with iperf:

Server Computer: iperf -s

Client Computer: iperf -c "Server IP Address" -w 100M -t 30

Like you, I was testing 2 different AP's (in my case their 3800 and 4500 routers) and switched back and forth between the routers at each of five different locations tested.

Can you please explain why you chose the variables for iperf in your script?

 

[AndCycle]

What makes you think it was iperf that crashed and not a signal dropout?

because there are some option like "--tradeoff" I wanna use for test,
but this always leads to a crash,
after some googling it reveals part of the function is broken and doesn't fix for a long time,
and daemon mode crashed when I execute dualtest command,

I am not really looking close for signal dropout, my laptop mainly use wired network most of the time,
wireless is for alternative backup, my smartphone and ipad

10 seconds seems a short time. Netgear recommended the following settings when I did some testing for them with iperf:

I tried 1 min at the beginning, and find out that my neighbor is using his wireless network all the time,
and one of them got a broken d-link online that occupy entire channel 11,
that's why I turn into multiple small test for consist result,
by the way, I lived in a apartment

Like you, I was testing 2 different AP's (in my case their 3800 and 4500 routers) and switched back and forth between the routers at each of five different locations tested.

Can you please explain why you chose the variables for iperf in your script?

-y CDMSV is for csv report, easier for processing data,

-w 64k?
that's base on some previous usage experience,
win xp has a 8k window as default, and it's too small for wired giga network,
64k seems to be a right value, and it's the default for win7 so I pick it.

anymore I can help?