Thanks Adrian and Wojtek, There is a bit called AR5K_PHY_AGCCTL_OFDM_DIV_DIS which led me to thinking it should be possible. At least now I can give up and work around it.
I still don't fully understand why fast diversity works for CCK (11b) rates with the same length preamble as the OFDM rates. i.e. The radio definitely must be "looking" at the non-default antenna for those modulations, because it receives them fine - just not OFDM. However, there is a lot I don't understand about this radio ;o) Thanks Brett -----Original Message----- From: adrian.ch...@gmail.com [mailto:adrian.ch...@gmail.com] On Behalf Of Adrian Chadd Sent: 17 July, 2014 6:13 PM To: Wojciech Dubowik Cc: ath5k-devel@lists.ath5k.org; Wright, Brett Subject: Re: [ath5k-devel] ath5k 5GHz Receive Diversity problem Hi, Right. It takes time for the hardware to figure out the signal level on each antenna - there's a coarse gain change and then a couple of fine gain changes so it can calculate RSSI. If the packet has a short preamble or it picks it up a little too late then yeah, you won't get any fast diversity. -a On 17 July 2014 01:01, Wojciech Dubowik <wojciech.dubo...@neratec.com> wrote: > On 07/17/2014 03:24 AM, brettwri...@eaton.com wrote: >> >> Hello List, >> >> I am having a problem getting Rx diversity to work with both ar5213 >> and ar5413. >> >> What I have found is that the fast diversity works great for >> receiving 802.11b (i.e. CCK) modulations. >> >> But fast diversity for OFDM just does not seem to work (and yes >> diversity is turned ON). >> >> I am using controlled conditions, i.e. cable and attenuators, and >> find that basically on the non-default antenna there is effectively >> 30dB attenuation. This means that to successfully receive an OFDM >> frame on the non-default antenna the signal has to be 30dBm above the >> noise floor! Not very useful… Basically Rx diversity for 802.11a 5GHz >> and pure 802.11g is a no-go. >> >> Does anyone know if this should work – or perhaps Rx diversity for >> OFDM is just a no-go? >> >> BTW – I’ve tried this with ath5k/mac80211 as well as madwifi and >> openhal, as well as the old closed source madwifi hal. They all >> behave the same way. >> >> Adrian – I see you have made some changes to FreeBSD related to this >> – do you have any idea if this **should** work? >> >> Anyone else? >> > AFAIK at least for OFDM the packet has to be seen first on default > antenna. Then rx diversity is switching to the other antenna during > preamble to check for a better signal. It means that first you have to > be on a proper antenna and be able to decode a packet (or at least > preamble). You can receive on the other one when you pump the signal > 30dB because it's a typical isolation for antenna RF switch on > reference designs. In our design it's typically 50dB and then for sure > you won't get rx diversity to work if you are on the wrong antenna. > That's why I know ;o) > > Rest is handled in SW i.e. if chipset has chosen alternate antenna > three times in a row then switch default to the other one. Could be > also based on tx failures. > > Br, > Wojtek >> >> Thanks >> >> Brett >> >> >> >> _______________________________________________ >> ath5k-devel mailing list >> ath5k-devel@lists.ath5k.org >> https://lists.ath5k.org/mailman/listinfo/ath5k-devel > > _______________________________________________ > ath5k-devel mailing list > ath5k-devel@lists.ath5k.org > https://lists.ath5k.org/mailman/listinfo/ath5k-devel _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
