802.11n oddness (pretty picture inside)

[itsmydamnation]

running a cisco 3502 with 40mhz channel. there are no other aps except mine in the 5ghz unii 1 range. When i run one client everything stays within the 40mhz. When i bring on the second client im getting very 802.11 looking RF on a 3rd 20mhz channel, i pause the Download and it stops.

so why?

the 3502 is a 2:3 mimo radio so could it be from the 2nd client tx rather then the AP tx?

[itsmydamnation]

68 views and no comments


will do some more playing.

[beckyboo]

Is the second device capable of 40mhz communication ? I know that with some of my wireless gear it only works at 20mhz.

[itsmydamnation]

Originally Posted by beckyboo

Is the second device capable of 40mhz communication ? I know that with some of my wireless gear it only works at 20mhz.

both devices have have PHY data rate of 300mbps so yes. even if its 20mhz only it would just use channel 40.

[evilasdeath]

I just came to see the pretty pictures, I wish I had one of those devices that could show me spectrum.

I haven't looked at enough spectrals to know how it should look to comment. I also don't know if the 5ghz band compares to the 2.4 Ghz which the 20mhz channels do actually have quite a lot of spectral overlap between channels.

[itsmydamnation]

each channel is 5mhz, a single wireless stream uses 20mhz of bandwidth. if you look at the graph is goes up in groups of 4 channels (20mhz). what your looking at is an .11n extension allowing 40mhz of bandwidth for a wireless stream. what is odd is when i start up this client there is what looks to be transmission on an unexpected channel.

[Iceman]

Didn't want to be view 300~ something without any kind of input.

Does 5ghz have a fat channel intolerance mechanism?

[itsmydamnation]

Originally Posted by Iceman

Didn't want to be view 300~ something without any kind of input.

Does 5ghz have a fat channel intolerance mechanism?

what do you mean? like RF side lobes? unlike 2.4 where there is a 5mhz gap between 1,6,11 there is no gap @ 5ghz.

i showed two proper RF guys at work ( ex navy Radio tech's both amateur radio nutts, build there own RF gear etc) there quite stumped.


edit: or do you mean 40mhz vs 20mhz, then yes it does. In this case its not even needed. the main 20mhz is channel 40 and there is no other RF on 36 or 44. any 20mhz only devices (802.11a) will only connect on 40 and not see channel 36. From there its standard CTS to self/ RTS/CTS.

[Zardoz]

Hi,

This is freaky.

What is the client that spits out stuff in the wrong channel? Drivers/chipset etc? With 802.11n, it's always a brilliant idea to ensure your drivers are current.

What happens if you run only client #2? What's the performance like?

Looks like the client might be trying to do something in 40 and 44 instead of 36 as the extension channel.

Edit: also wondering wtf is going on with channel 48.

[beckyboo]

I'm stabbing in the dark here (as with the other comment), but what about ghosting of the signal ? Similar to what happens when you have your tv signal bounce back off a nearby structure and appear to be a second device.

[klightspeed]

Given that the signal isn't flat over the entire channel, and it looks like the spectrum analyzer is rather limited (averaging out the power over the channel and filtering out the centre frequency), I would say that the apparent signal on channels 44 and 48 is simply unwanted harmonics in the transmitted signal. It's about 36dB below the intentional transmission, so it's not all that strong.

[itsmydamnation]

you might be right on your theory. but just to point out 802.11 A/G/N are OFDM based signals, there not flat and the dip in the middle is expected.

can kind of see it here from another analyzer

http://www.techedbackstage.net/wp-co...A-Spectrum.png
(but that a little messy 2.4, not my nice clean 5.0 )

i cant quite remember the specs but its scan time is about 60msec for 20mhz and accuracy is within about 2db cant remember what the resolution is.

[LostBenji]

Although you are using a tool with "Fluke" on its side, there are no differences to all the other soft-tools using crappy little USB dongles. If you want real readings then you need real "spec-ans" that are a little more accurate.

Your navy (RF) mates should know this, it's basic stuff.
Your looking at one side of the center target frequency, open the view out so we can see a few channels down as well as above. What I can see is bleed-over and quite normal considering your at better than 40dBm (bloody strong signal, too close to device).

Cisco is a big-brand name too, but same again, their wireless certainly doesn't carry the same clout as their copper networking products. Harmonics, spurious noise and other taints are normal.



Stop reading into it too much....

http://www.ebay.com.au/sch/i.html?_t...4&_sop=3&_sc=1
Have a look for an idea what real tools cost when serious, they are S/H, think what new costs.

[itsmydamnation]

im aware that fluke is no better then any other product in this market space and multiple times the price (its not like i payed for it). channel 36 is the lowest in unii 1 and it doesn't let you view anything below that ( unless you go down to 4.9/2.4). could just move it up to a higher frequency 5.24ghz or something.

yeah the signal is strong because at the time i forced the AP to is max allowed power which is 20 DBm because i was also testing range/ performance and this guy only has its internal antenna.

given that its only happening with this second client ( laptop with inbuilt radio) i would have thought it was likely to be the client if anything.

[Zardoz]

Originally Posted by Zardoz

What is the client that spits out stuff in the wrong channel? Drivers/chipset etc? With 802.11n, it's always a brilliant idea to ensure your drivers are current.

What happens if you run only client #2? What's the performance like?

Still curious about the above. This is seriously worrying given that clients like this could really foul up a high density installation.