Hi there,

I installed Kamikaze 8.09 RC 1 (openwrt-brcm47xx-squashfs.trx) on my WL 500gP (v 1) with an Atheros mini-pci WLAN board (TP-Link TL-WN660g).

I installed kmod-madwifi and hostapd-mini and set up WLAN as described here http://wiki.openwrt.org/OpenWrtDocs/Har … us/WL500GP and found that I have bad reception. 5m away from the router I hardly see the network.

I use psk2 encryption and have the setup working in ap mode.

I tried out both antenna sockets with each of them I tried the txantenna / rxantenna settings 1 and 2. I checked the other options from http://sr.uz/index.php?p=223&more=1 … 1&pb=1 like turning diversity off explicitely, turning on bursting, setting distance, and increasing txpower. None of this helped.
The following is my output from uci show wireless:
wireless.wifi0=wifi-device
wireless.wifi0.type=atheros
wireless.wifi0.disabled=0
wireless.wifi0.mode=11bg
wireless.wifi0.distance=600
wireless.wifi0.channel=3
wireless.wifi0.txantenna=1
wireless.wifi0.rxantenna=1
wireless.cfg03d42f=wifi-iface
wireless.cfg03d42f.device=wifi0
wireless.cfg03d42f.network=lan
wireless.cfg03d42f.mode=ap
wireless.cfg03d42f.ssid=network
wireless.cfg03d42f.encryption=psk2
wireless.cfg03d42f.key=xxxxxxxx
wireless.cfg03d42f.txpower=85
wireless.cfg03d42f.bursting=1

Unfortunately I can't tell if this is an issue of 8.09/RC1 only, since beforehand I had the original ASUS system on the router (also using the broadcom WIFI - there reception was alright).

Does anybody have suggestions on how to go on here? Is it worth to replace the standard madwifi with the ath5k kmods and retry? Could it be a hardware problem of the mini-pci card?

Cheers,
hol9000

Hi.

I see you have a txpower of "85". The txpower is given in dBm so 85 is a bit high, maybe madwifi reverts to minimum signal strength if the value is out-of-range. Try to lower the txpower value to something like 15 or 20.

Edit:
Are you sure that the antenna selection is right?

~ JoW

(Last edited by jow on 3 Dec 2008, 14:42)

Thanks for the suggestion.

In the beginning I started with no explicit txpower-settings, then I increased it to 85, since that value was mentioned somewhere in the forums, then I reduced it to 18 like mentioned in the sz.ru link. Unfortunately none of these settings increased the reception quality.

Cheers,
hol9000

Hi.

Okay, run "athstats" on the shell and look which antenna has the "higher numbers". Make sure it's the same as configured in rxantenna, txantenna.
For me it looks like this on a Fonera:

Antenna profile:
[1] tx 32414990 rx 62224404
[2] tx   212600 rx      774

So #1 is clearly the primary antenna.

On my wl500gp for example, the pigtail was too short to be connected to the main antenna plug, so I had it to connect to the "aux" port and change antennas in configuration.

~ JoW

This is the output of athstats:

5227 tx management frames
2812 tx failed due to too many retries
90115 long on-chip tx retries
10481 tx frames with no ack marked
69169 tx frames with short preamble
11407 tx frames with an alternate rate
130556 rx failed due to bad CRC
19 PHY errors
    19 CCK restart
1411540 beacons transmitted
71298 periodic calibrations
1445517 rate control checks
rssi of last ack: 8
136 switched default/rx antenna
Antenna profile:
[1] tx     2341 rx   526378
[2] tx    78700 rx    52498

Situation in the Antenna profile is a bit confusing here. It might be that the numbers are messed up because I changed the antenna socket and the tx/rx settings a number of times and didn't reboot in between.

hol9000

Hi.

Maybe reboot to get clear stats. Also explicitly disable the diversity before:

uci set wireless.wifi0.diversity=0
uci commit wireless

~ JoW

Ok, diversity is set to off again and I did a reboot.

Settings look like this now:
wireless.wifi0=wifi-device
wireless.wifi0.type=atheros
wireless.wifi0.disabled=0
wireless.wifi0.mode=11bg
wireless.wifi0.distance=600
wireless.wifi0.channel=3
wireless.wifi0.txantenna=1
wireless.wifi0.rxantenna=1
wireless.wifi0.diversity=0
wireless.wifi0.txpower=18
wireless.cfg0392cb=wifi-iface
wireless.cfg0392cb.device=wifi0
wireless.cfg0392cb.network=lan
wireless.cfg0392cb.mode=ap
wireless.cfg0392cb.ssid=network
wireless.cfg0392cb.encryption=psk2
wireless.cfg0392cb.key=xxxxxxxx
wireless.cfg0392cb.txpower=18
wireless.cfg0392cb.bursting=1

Also I switched txpower to 18 again.

Output of athstats is the following:

107 tx management frames
419 tx failed due to too many retries
5482 long on-chip tx retries
528 tx frames with no ack marked
3837 tx frames with short preamble
1725 tx frames with an alternate rate
4181 rx failed due to bad CRC
67833 beacons transmitted
269 periodic calibrations
69460 rate control checks
rssi of last ack: 17
rssi of last rcv: 18
1 switched default/rx antenna
Antenna profile:
[1] tx     3952 rx     9250
[2] tx        0 rx    13336

The rx on [2] was at high numbers for some time without wifi activity from my notebook. The numbers for [1] went up only after I started to use wlan myself (before there were at 0 or 1).
Seems that tx/rxantenna is set to the right slot. Can the high number for rx on antenna [2] be explained by traffic in the same channel but from other networks? Or does this point towards some hardware problem?

Cheers,
hol9000

Hmm, I would read that as the second antenna could receive/"hear" more packets, which would indicate that this should be the right antenna to use. You should check once more with tx/rxantenna set to 2.

Looking at my stats:

I have much much fewer received packets at the unused one (this was collected in about 10 days of traffic), and there are some networks around here (although I don't know if they are on the same channel).

KM

Hi.

Yes, definitely, change txantenna and rxantenna to 2 and try again.

~ JoW

Ok, I changed rx/txantenna to 2, and after some traffic athstats looks like this:

807 tx management frames
1021 tx failed due to too many retries
16536 long on-chip tx retries
1120 tx frames with no ack marked
7876 tx frames with short preamble
2434 tx frames with an alternate rate
1455 rx failed due to bad CRC
3 PHY errors
    3 CCK restart
91780 beacons transmitted
269 periodic calibrations
93984 rate control checks
rssi of last ack: 16
rssi of last rcv: 16
1 switched default/rx antenna
Antenna profile:
[1] tx        0 rx     4579
[2] tx     8213 rx    43211

Reception is still bad.

But I checked again and the antenna is connected to main, which should be antenna slot 1, right?
Anyways, then I detached the antenna from the card and tried to see what was going on. I still have (bad) network a few meters away, at the moment it's quite hard to compare if the link quality became worse or if that is within the range of fluctuations of what I get _with_ antenna.

At that point do you think it makes sense to reinstall everything and try with the original broadcom wifi?

Cheers,
hol9000

To add to the last post:
Over night I connected the antenna to aux (which should be slot [2]) and set rx/txantenna to 2.
The antenna profile shows the normal traffic on [2] and receives a lot of other stuff on [1]:
Antenna profile:
[1] tx        0 rx   170882
[2] tx    16964 rx    53781

This is the contrary situation to my post (Yesterday 18:16:01), where I had the antenna connected to main and rx/txantenna set to [1].

Hm, does anybody else have some idea what might go wrong?

Are there other Wl500-gp users with atheros wlan and 8.09 RC1 installed who have similar problems or who can confirm that their setup is working?

Cheers,
hol9000

I tried the latest trunk, but all i get is an kernel oops...

Hi,

Wouldn't you want to install from here instead since your radio/chipset is Atheros and not Broadcom?

http://downloads.openwrt.org/kamikaze/8.09_RC1/atheros/

I am NEW so do NOT rely on what I just wrote but the more experienced should confirm it or not.

Rich

Rich, thanks for the suggestion. The chipset of the Asus WL500g Premium is broadcom. Just the original minipci wifi card was replaced with an atheros card. I don't think that the atheros platform builds would start at all (if one would manage to even get them onto the router).

Michu, you didn't try the RC 1 build at all, right?

Cheers,
hol9000

hol9000,

I am watching this thread because I also have an Atheros card to put into my WL-500gP prior to 8.09 RC1 being released.

Rich

nope, i just build from trunk... but it used to work, perhaps its because of gcc 4.24...

If you are using gcc 4.2.4, you should try 4.1.2. It seems 4.2.4 is a bit too eager with optimizing or so (see this ticket).

KM

I had the same problems and saw that settings reported by " grep -r . /proc/sys/dev/wifi0/ /proc/sys/dev/ath/ 2>/dev/null" are different from config file (especially diversity). Additionally I saw a lot of discennections from ap. Maybe it is a problem of newer ath drivers?
Could you check it?

You're right, that is the case here as well:
grepping in /proc/sys/... with your command gives this:
[...]
/proc/sys/dev/wifi0/txantenna:2
/proc/sys/dev/wifi0/rxantenna:1
/proc/sys/dev/wifi0/diversity:1
[...]

while in the output of
uci show wireless
the following is set:
[...]
wireless.wifi0.txantenna=2
wireless.wifi0.rxantenna=2
wireless.wifi0.txpower=18
wireless.wifi0.diversity=0
[...]

That would explain the numbers in the antenna profile.

I tried the following:
echo 0 >  /proc/sys/dev/wifi0/diversity

Afterwards the respective entries in /proc/sys/ look like this:
/proc/sys/dev/wifi0/txantenna:2
/proc/sys/dev/wifi0/rxantenna:2
/proc/sys/dev/wifi0/diversity:0

Reception gets better, but quality is still far from being as good as with the original broadcom card and the original Asus firmware.
But at least this is a starting point for further experiments

Ok, it gets weirder. I rebooted, and now
uci show wireless
and the entries in /proc/sys/...
correspond:
/proc/sys/dev/wifi0/txantenna:2
/proc/sys/dev/wifi0/rxantenna:2
/proc/sys/dev/wifi0/diversity:0

wireless.wifi0.txantenna=2
wireless.wifi0.rxantenna=2
wireless.wifi0.txpower=18
wireless.wifi0.diversity=0

but the antenna profile still shows the usage of different antennas:
Antenna profile:
[1] tx        0 rx     1195
[2] tx      720 rx        0

kanjimonster: you dont have troubles compiling bcm47xx with gcc 4.1.2?

check my ticket here: https://dev.openwrt.org/ticket/4287
any suggestions?

(Last edited by michu on 7 Dec 2008, 18:01)

Compiling fine here with 4.1.2. Did you already try a distclean (remember to backup your old .config), and recompiling with 2.6 as the only change?

KM

Hi,

just wanted to add that I also have very bad reception with my Asus WL-500W / Gigabyte WIAG02 Atheros MiniPC using Madwifi and
Kamikaze 8.09 RC 1 (openwrt-brcm47xx-squashfs.trx).

I still need to dig further, but wanted to let you know that you are not alone with your trouble ;-)

Cheers,
XOR

Thanks for the feedback. Today a new cable and another antenna arrived, so on the weekend I'll be able to try out if there's a hardware problem and if two antennas are a temporary solution to my kind of "forced" antenna diversity setup.

Cheers,
hol9000