Hi, I've had the chance to connect two WRT54GLs to a 50GHz Agilent spectrum analyzer. The devices where set in "ap" mode on channel 1. Here there are the pics of the screen:

What do you think? The analyzer was set in the "max hold" mode of course. The power reading was approximately 8.8dBm. I understand a spectrum analyzer is not a perfect power meter, but shoudn't I expect something near 19dBm from a Linksys? Am I missing something?

This is the output of an Ubiquiti Litestation2 in client mode scanning all the channels. It eventually found the local wifi network at -93dBm while connected to the spectrum analyzer:

Finally, we conected our 24dBi grid antenna to an HP network analyzer, mainly to read the VSWR:

Any comment from the experienced people is very welcome!
Thanks
Alessio Sangalli

Cool!

I assume you were using OpenWrt in these tests? Do you have wl0_txpwr set? (normally OpenWrt doesn't set this and uses the 'default' value of the driver instead). I don't know jack about spectrum analyzers, but this makes me wish I had one .

Well, we also tried to reduce the txpwr and the analyzer reported a lower power level, but the image was taken at the "standard" value of 19dBm, as "iwconfig eth1" reported it!

The energy is dispersed over the 802.11g bandwidth and scanning through that with an 1 MHz aperture and comparing two different scan situations (100 MHz and 300 MHz) and different transmission circumstances is not going to give reliable and comparable results; already the background noise level shows to be 10 dB different between the two.

I.m.h.o. to have an idea of power it would be better to have the analyzer look at the entire emission spectrum (e.g. ABW =30 MHz). It doesn't even really need to scan but can be tuned to the centre frequency of the chosen channel. The emission power, apart from being not always there but only in the transmission period of the simplex connection and thus depending on duty cycle receive/transmit, is actually a momentary power that gets distributed over a bandwidth because of the modulation used. If you look only into a slice out of that bandwidth the reported power inside that slice must of course be less.