[ath9k-devel] weak signal strength using ar928x

[Luis R. Rodriguez <lrodriguez@atheros.com>]

On Sun, Jan 17, 2010 at 02:29:20PM -0800, Lorenzo Bianconi wrote:
> >> On Fri, Jan 15, 2010 at 06:54:23AM -0800, Porsch, Marco wrote:
> >>> >>> Hello list,
> >>> >>
> >>> >>> i'm experiencing a very weak signal strength using ath9k and ar928x wireless in a wndr3700 router. I placed two of these approximately 3 meters line of sight of each other, and see signal strength of about -90dbm in mesh mode [1].
> >>> >>> The problem counts for sending and receiving, also in bss-mode. I posed my laptop approximately one meter away from the router as ap and measured the signal strength of the netgear firmware (openwrt 7.x with propietary driver) and openwrt (compat-wireless-2010-01-11, compat-wireless-2009-12-05), which gives me 20dbm difference [2].
> >>> >>> Using 'iwlist scan', original firmware shows 14 APs in my neighbourhood, openwrt shows 4.
> >>> >>
> >>> >> [...]
> >>> >>
> >>> >> I have taken a further look into the problem and realized, that 'iwconfig txpower' has no affect to the signal strength. I also tried up-down or scanning after change.
> >>> >> A test with hacked 'regd.c' (all set to 30dBm, no further limitations) also showed no changes in signal quality. Power management is off.
> >>> >>
> >>> >> A interesting detail is, that 5GHz (channel 52) also has quiet limited range, but far better than 2,4GHz. 5GHz txpower is set to 30dBm by default.
> >>> >>
> >>> >> Kernel is Linux OpenWrt 2.6.30.10 #10 Fri Jan 15 11:12:04 CET 2010 mips, wireless now from compat-wireless-2010-01-13.
> >>> >>
> >>> >>
> >>> >> Some debug outputs after 10min of iperf traffic over mesh-link between two identical wndr3700. I have seen no other 5GHz wifi in range.
> >>> >> 1) 3 meters line of sight
> >>> >> 2) 10 meters with thin glass/metal walls
> >>> >>   (in both cases 2,4GHz wont communicate at all)
> >>> >>
> >>> >>
> >>> >> 1)
> >>> >> [  3]  0.0-600.0 sec  1.27 GBytes  18.2 Mbits/sec
> >>> >>
> >>> >> cat /sys/kernel/debug/ath9k/phy1/rcstat
> >>> >>    HT    MCS   Rate    Success    Retries   XRetries        PER
> >>> >>                6.0:          0          0          0          0
> >>> >>                9.0:          0          0          0          0
> >>> >>               12.0:          0          0          0          0
> >>> >>               18.0:          0          0          0          0
> >>> >>               24.0:          1          0          0          0
> >>> >>               36.0:          1          0          0          0
> >>> >>               48.0:        286         19          0          0
> >>> >>               54.0:    1160839     113918         26          7
> >>> >>
> >>> >> Station 00:24:b2:58:fb:54 (on imesh5g)
> >>> >>        inactive time:  570 ms
> >>> >>        rx bytes:       2907897221
> >>> >>        rx packets:     2001025
> >>> >>        tx bytes:       422706850
> >>> >>        tx packets:     1161637
> >>> >>        signal:         -67 dBm
> >>> >>        tx bitrate:     54.0 MBit/s
> >>> >>        mesh llid:      4935
> >>> >>        mesh plid:      19635
> >>> >>        mesh plink:     ESTAB
> >>> >>
> >>> >>
> >>> >> 2)
> >>> >> [  3]  0.0-600.0 sec    294 MBytes  4.11 Mbits/sec
> >>> >>
> >>> >> cat /sys/kernel/debug/ath9k/phy1/rcstat
> >>> >>    HT    MCS   Rate    Success    Retries   XRetries        PER
> >>> >>                6.0:       6936       7587       2100          0
> >>> >>                9.0:      17804      18453       5797          0
> >>> >>               12.0:     153370     112773      21194         13
> >>> >>               18.0:      40331     104799      45526         49
> >>> >>               24.0:         94       4822       7450         54
> >>> >>               36.0:          0         15         21         82
> >>> >>               48.0:          0          0          0          0
> >>> >>               54.0:          0          0          0          0
> >>> >>
> >>> >> Station 00:24:b2:58:fb:54 (on imesh5g)
> >>> >>        inactive time:  800 ms
> >>> >>        rx bytes:       670383000
> >>> >>        rx packets:     442114
> >>> >>        tx bytes:       18396981
> >>> >>        tx packets:     218811
> >>> >>        signal:         -86 dBm
> >>> >>        tx bitrate:     12.0 MBit/s
> >>> >>        mesh llid:      57085
> >>> >>        mesh plid:      19775
> >>> >>        mesh plink:     ESTAB
> >>>
> >>> >Hi Marco,
> >>> >
> >>> >I faced the same problem and I solved this issue with this simple patch:
> >>> >
> >>> >Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
> >>> >---
> >>> >--- a/drivers/net/wireless/ath/ath9k/main.c
> >>> >+++ b/drivers/net/wireless/ath/ath9k/main.c
> >>> >@@ -2817,8 +2817,10 @@
> >>> >        }
> >>> >
> >>> > skip_chan_change:
> >>> >-       if (changed & IEEE80211_CONF_CHANGE_POWER)
> >>> >+       if (changed & IEEE80211_CONF_CHANGE_POWER) {
> >>> >                sc->config.txpowlimit = 2 * conf->power_level;
> >>> >+               ath_update_txpow(sc);
> >>> >+       }
> >>> >
> >>> >        spin_lock_bh(&sc->wiphy_lock);
> >>> >        disable_radio = ath9k_all_wiphys_idle(sc);
> >>> >
> >>> >
> >>> >Now "iwconfig wlan0 txpower" works fine.
> >>> >
> >>> >Regards.
> >>> >
> >>> >Lorenzo
> >>>
> >>> Thanks for your reply,
> >>>
> >>> I applied your patch, but unfortunately i still cannot change my txpower. Even after setting txpower to 0 or off the AP shows the same signal level in 'iwlist scan'.
> >>
> >> Please see the patch
> >>
> >> http://bugzilla.kernel.org/attachment.cgi?id=23981
> >> http://bugzilla.kernel.org/show_bug.cgi?id=14642
> >>
> >> Power change is reported there to change after a scan, this patch
> >> is supposed to fix that upon immediate tx power change but I never
> >> got a report of its results.
> >>
> >> Feel free to follow up on that bug report. If you have something
> >> that can actually measure the output power that would be best.
> >>
> >>  Luis
> >> _______________________________________________
> >> ath9k-devel mailing list
> >> ath9k-devel at lists.ath9k.org
> >> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
> >>
> >
> > Hi Luis,
> >
> > I tested this patch on two Ubiquiti Rocket M devices (AR9280 chipset)
> > in this simple network configuration:
> >
> > Rocket1 <--wireless link--> Rocket2
> >
> > I injected packets from kernel space using a VAP in monitor mode on Rocket1
> > and misured the Rx power through another VAP in monitor mode on
> > Rocket2 analyzing the RSSI information reported from the driver. I
> > obtained the relationship between the Rx power and the Tx one is quite
> > linear when the Tx power is in the range from 3dbm to 20dbm whereas I
> > got some unexpected results when I set lower Tx power values
> > (0dbm-2dbm)
> > Tomorrow I am going to test this patch in AP/STA mode.
> >
> > Regards.
> >
> > Lorenzo
> >
> 
> Hi Luis,
> 
> I carried out some tests in order to verify if this patch works fine
> also in AP/STA mode.
> Here is my topology:
> 
> AP = PC Engines Wrap with the latest version of OpenWRT trunk
> (compat-wireless-2010-01-15)
> and MikroTik Router Board R2n (Atheros AR9223 chipset) as miniPCI card
> 
> STA = Dell Studio XPS 13 (Intel Core 2 Duo P8700, 4096MB SDRAM DDR3)
> with Debian Lenny (kernel 2.6.32.3) and wireless Dell 1515 (Atheros
> AR9280 chipset)
> 
> STA <--Wireless link--> AP
> 
> I gradually decreased the Tx power on the AP and I measured the Rx one
> through the information reported by Wicd on my Dell laptop.
> I obtained these results:
> 
> 1) iwconfig wlan0 txpower 20 ---> Rx power -37dbm
> 2) iwconfig wlan0 txpower 15 ---> Rx power -41dbm
> 3) iwconfig wlan0 txpower 10 ---> Rx power -45dbm
> 4) iwconfig wlan0 txpower 5 ---> Rx power -51dbm
> 5) iwconfig wlan0 txpower 0 ---> Rx power -37dbm ????
> 
> Ignoring the unexpected behavior I obtained in the latest test,
> in my opinion this patch works correctly.

Thanks Lorenzo I'll submit this upstream now, the 0 txpower anomoly can
be addressed separately.

  Luis