On Fri 01 Nov 2013 11:35:06 PM GMT, Nick Kossifidis wrote: > On Fri 01 Nov 2013 11:34:35 PM GMT, Nick Kossifidis wrote: >> On Fri 01 Nov 2013 06:12:42 PM GMT, David Goodenough wrote: >>> On Wednesday 30 Oct 2013, Beat Meier wrote: >>>> Hello >>>> >>>> I have a lot of wireless repeaters which uses madwifi. Because of >>>> several problem like not anymore supported and >>>> other problems like "no xmit buf" I need to upgrade to mac80211 with ath5k. >>>> What I notice is that in the b band the ath5k driver has about 10dbm >>>> less than madwifi driver and a band too. >>>> It's not the noise it have really 10db difference. Nobody having this >>>> problem with long distance links? >>>> This is with Compex cards, CM9 ... >>>> Because of this I have bad througput (signal of -80) instead of -70 with >>>> madwifi and good througput. >>>> >>>> Any ideas what could be wrong and how to debug the issue??? >>>> >>>> Greetings >>>> _______________________________________________ >>>> ath5k-devel mailing list >>>> ath5k-devel@lists.ath5k.org >>>> https://lists.ath5k.org/mailman/listinfo/ath5k-devel >>> I seem to recall that there was a kludge in madwifi that added 10dB to >>> the real power so that high power cards could be set to their real power >>> limit rather than the FCC/ETSI limited power levels. Ath5k as I recall >>> does not do this as it is integrated into the crda data. >>> >>> Or was it the other way around, i.e. now that crda was in force with ath5k >>> you had to lie about the power level? >>> >>> Either way, I do seem to recall that the difference was 10dB. >>> >>> David >>> _______________________________________________ >>> ath5k-devel mailing list >>> ath5k-devel@lists.ath5k.org >>> https://lists.ath5k.org/mailman/listinfo/ath5k-devel >> >> a) 10dB difference is too much to be a driver bug, CM9s are tested and >> at least the one I used to have operated as it should, I don't know >> which version of MadWiFi you have or how you measure the power >> difference. Last time I checked with a R&S spectrum analyzer tx power >> setting was O.K. (well at least as good as MadWiFi let's say), most >> probably something is going on with CRDA or some layer above the driver. >> >> b) 40MHz channels are messing up the spectrum AND in the case of the >> Atheros 5k cards, when used you get even less tx power. There is a >> 20MHz turbo mode that's better. Driver supports all possible bw modes >> (5,10,20,40MHz) but there is no standard way to set them. Here is a >> patch that enables a debugfs file to do this -> >> https://dev.openwrt.org/browser/trunk/package/mac80211/patches/440-ath5k_channel_bw_debugfs.patch?rev=32595 >> it never went upstream though, there are lots of people (and some >> projects out there) that successfully use bw modes. >> >> For reference here are some screenshots from an AR5413 card I used, >> from the R&S spectrum analyzer from 2010, when bw modes were added to >> ath5k. > > Sorry forgot the link... > http://archlinux.mirrors.ovh.net/ftp.kernel.org/pub/linux/kernel/people/mickflemm/ath5k-bwmodes/
Anyway if you want better speed please don't go for the "hackish" 40MHz mode, the spectrum is something we should all care about ! Use 11n in 20MHz mode, you'll get much better results even with a single chain. _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel