Faster wifi with 1043nd V1

Report wireless and/or network connectivity problems in this forum.

Moderator: Moderators

kabi
Posts: 29
Joined: Thu Aug 28, 2014 5:33 am

Faster wifi with 1043nd V1

Post by kabi »

Hello everyone

This seems to be ever-green topic here for any Atheros based router - including my 1043nd v1.

I'm playing with latest Gargoyle (kernel 3.18.75, Gargoyle version 1.10.0, Gargoyle revision ab16f958).

After big disappointment over the speed of WiFi with 'default' GUI wifi setup - I've started to experiment with direct 'vi /etc/config/wireless' and tried various 'advices' found eithet here or on LEDE or openwrt forums. However I've still been finding this just *NOT* as fast enough as it ought to be - since my net-provider gives rather rock solid 150mbps ATM and my 'best' wifi speed was usually somewhere between 30-40mbs - even with "option nosyns '1'".

Of course I've tried different channels - using channel #1 seemed to be best even thought channels > #1 can use slight more txpower (24 over 'just' 23 for #1).

Then I've started to wonder myself how 'antenna chaining' is configured - and noticed in 'iw phy phy0 info' output this:

Available Antennas: TX 0x7 RX 0x7
Configured Antennas: TX 0x7 RX 0x7

This is according to some 'dd-wrt' post wrong for 1043nd V1 - since antennas for TX are only 1+3 while RX is 1+2+3 so correct number here is supposedly: Configured Antennas: TX 0x5 RX 0x7

Doing some 'research' how to set this up on openwrt - using these 2 lines:
option txantenna '5'
option rxantenna '7'
does the trick (Note: BOTH lines must be provided, otherwise option setting for tx is ignored if rx even in it's default setting 7 is not provided)

And to my surprise now laptop can take over 70-80mbps when not to far away from router - so much much better then anything stock.

What I've not managed to use is to setup anything HT40 mode - in this mode the Windows machines seems to cannot connect ??

dd-wrt seems to provide some HT20/HT40 dual mode ??
How that can be made in openwrt/gargoyle world ??

I'm also seeing there messages in log:
[49680.530000] ath: phy0: DMA failed to stop in 10 ms AR_CR=0x00000024 AR_DIAG_SW=0x42100020 DMADBG_7=0x000084c0
[52554.240000] ath: phy0: DMA failed to stop in 10 ms AR_CR=0x00000024 AR_DIAG_SW=0x42100020 DMADBG_7=0x000286c0
[57840.580000] ath: phy0: DMA failed to stop in 10 ms AR_CR=0x00000024 AR_DIAG_SW=0x42100020 DMADBG_7=0x000084c0

Considering this is even on completely UNLOADED wifi, where there is just one sleeping Nexus7 doing nothing - it's probably some common feature of Atheros chipset - but so far it seems WiFi keeps running and not disconnecting.

Since I like bwmon info present in Gargoyle - I'm not using tha lastest LEDE yet - but since I've now figured big impact of 'txantenne' option I could probably give it a try once again to see if the kernel error message was fixed, since before the speed of any new router firmware was simply too slow compared with some oldish dd-wrt or even stock TP-Link firmware.

The primary point of this post is - user of TP-LINK 1043nd V1 need to manually configure /etc/config/wireless.

Maybe Gargoyle could provide some more GUI options for WiFi settings - that can improve speed in major way ?
(like dd-wrt has them)
Last edited by kabi on Tue Apr 17, 2018 5:08 am, edited 1 time in total.

Lantis
Moderator
Posts: 5525
Joined: Mon Jan 05, 2015 5:33 am
Location: Australia

Re: Faster wifi with 1043nd V1

Post by Lantis »

Gargoyle is designed to provide powerful, yet simple configuration of your hardware.
The typical use case does not include selecting individual antennas for transmit or receive, and therefore this option is not exposed to the GUI. It is of course, retained in the backend for you to configure as you wish.

HT40 mode is enabled by selecting b+g+n or a+n mode and selecting Channel Width of 40Mhz. Note however that at least in the 2.4GHz band, the use of 40MHz may not be allowed depending on your local wireless environment, and will self disable. It will fallback to HT20 in this instance.
I hope this clears up that configuration option.

I don't have any information about the error message you are experiencing, but it is an upstream issue, not Gargoyle specific.
http://lantisproject.com/downloads/gargoyle_ispyisail.php for the latest releases
Please be respectful when posting. I do this in my free time on a volunteer basis.

kabi
Posts: 29
Joined: Thu Aug 28, 2014 5:33 am

Re: Faster wifi with 1043nd V1

Post by kabi »

Lantis wrote:Gargoyle is designed to provide powerful, yet simple configuration of your hardware.
The typical use case does not include selecting individual antennas for transmit or receive, and therefore this option is not exposed to the GUI. It is of course, retained in the backend for you to configure as you wish.


My primary point would be - is there some way, how to simplify this 'relatively' complex way "how to configure 1043ndV1" ?

i.e. at least maybe provide 'input edit text box' with /etc/config/wireless content for easy edit for those users which are not experience Linux users ?
(Although maybe Gargoyle is only installed by Linux experts ??:)

Also it could be helpful if the 'build' for 1043ndV1 would go by default with mentioned 'option txantenna '5' in wireles config ?
(though some compiled option)

I'm pretty sure this option IS mandatory for this box to get decent WiFi performance.


Lantis wrote:HT40 mode is enabled by selecting b+g+n or a+n mode and selecting Channel Width of 40Mhz. Note however that at least in the 2.4GHz band, the use of 40MHz may not be allowed depending on your local wireless environment, and will self disable. It will fallback to HT20 in this instance.


Using 'US' country - assuming this should have no limits ?
Or do I have to use different country ?
The main point was - it's been not working with Gargoyle,
while HT40 has worked with dd-wrt and seems to be supported by 'iw' command (output added bellow my post)

Lantis wrote:I don't have any information about the error message you are experiencing, but it is an upstream issue, not Gargoyle specific.


Yep - lot of that even in bugzilla.redhat.com :)
That's the reason I'll probably try to experiment with recent LEDE that runs on 4.9 kernel to see if there is any progress on Atheros chipset driver.

Code: Select all

root@Gargoyle:/etc# iw phy phy0  info
Wiphy phy0
   max # scan SSIDs: 4
   max scan IEs length: 2257 bytes
   max # sched scan SSIDs: 0
   max # match sets: 0
   Retry short limit: 7
   Retry long limit: 4
   Coverage class: 0 (up to 0m)
   Device supports AP-side u-APSD.
   Device supports T-DLS.
   Available Antennas: TX 0x7 RX 0x7
   Configured Antennas: TX 0x5 RX 0x7
   Supported interface modes:
       * IBSS
       * managed
       * AP
       * AP/VLAN
       * WDS
       * monitor
       * mesh point
       * P2P-client
       * P2P-GO
       * outside context of a BSS
   Band 1:
      Capabilities: 0x104e
         HT20/HT40
         SM Power Save disabled
         RX HT40 SGI
         No RX STBC
         Max AMSDU length: 3839 bytes
         DSSS/CCK HT40
      Maximum RX AMPDU length 65535 bytes (exponent: 0x003)
      Minimum RX AMPDU time spacing: 8 usec (0x06)
      HT TX/RX MCS rate indexes supported: 0-15
      Frequencies:
         * 2412 MHz [1] (23.0 dBm)
         * 2417 MHz [2] (24.0 dBm)
         * 2422 MHz [3] (24.0 dBm)
         * 2427 MHz [4] (24.0 dBm)
         * 2432 MHz [5] (24.0 dBm)
         * 2437 MHz [6] (24.0 dBm)
         * 2442 MHz [7] (24.0 dBm)
         * 2447 MHz [8] (24.0 dBm)
         * 2452 MHz [9] (24.0 dBm)
         * 2457 MHz [10] (24.0 dBm)
         * 2462 MHz [11] (23.0 dBm)
         * 2467 MHz [12] (disabled)
         * 2472 MHz [13] (disabled)
         * 2484 MHz [14] (disabled)
   valid interface combinations:
       * #{ managed } <= 2048, #{ AP, mesh point } <= 8, #{ P2P-client, P2P-GO } <= 1, #{ IBSS } <= 1,
         total <= 2048, #channels <= 1, STA/AP BI must match, radar detect widths: { 20 MHz (no HT), 20 MHz, 40 MHz }

       * #{ WDS } <= 2048,
         total <= 2048, #channels <= 1, STA/AP BI must match
   HT Capability overrides:
       * MCS: ff ff ff ff ff ff ff ff ff ff
       * maximum A-MSDU length
       * supported channel width
       * short GI for 40 MHz
       * max A-MPDU length exponent
       * min MPDU start
Last edited by kabi on Thu Apr 19, 2018 4:22 am, edited 1 time in total.

Lantis
Moderator
Posts: 5525
Joined: Mon Jan 05, 2015 5:33 am
Location: Australia

Re: Faster wifi with 1043nd V1

Post by Lantis »

I do not think that option is necessary for your device. Plenty of users have sufficient speed without it.
It is an advanced, non-standard option, so there is no need for the GUI to expose it. The average user has no need to see the back end configuration.
For advanced user, it is there to use.

You should only use the country code that you are in. Using any other country code would be a violation of the relevant regulations for your country.
The 40MHz fallback is part of the 802.11 standard, you can't avoid it (legally).
Lantis wrote:It’s more than likely your router is only using 20mhz width to avoid co-channel interference as per the standard.
It’s been discussed a lot before, but here’s the link again:
https://www.smallnetbuilder.com/wireles ... ghz-part-1

“The good neighbour policy”


It works fine on Gargoyle. If you review the logs, it will probably even tell you that it has fallen back to HT20 mode.

Code: Select all

20/40 mhz operation not permitted
http://lantisproject.com/downloads/gargoyle_ispyisail.php for the latest releases
Please be respectful when posting. I do this in my free time on a volunteer basis.

NGowner
Posts: 46
Joined: Tue Mar 09, 2010 4:12 pm

Re: Faster wifi with 1043nd V1

Post by NGowner »

kabi wrote:After big disappointment over the speed of WiFi with 'default' GUI wifi setup - my 'best' wifi speed was usually somewhere between 30-40mbs - even with "option nosyns '1'".


kabi wrote:Doing some 'research' how to set this up on openwrt - using these 2 lines:
option txantenna '5'
option rxantenna '7'
does the trick (supposedly rx line is already default)

And to my surprise now laptop can take over 70-80mbps when not to far away from router - so much much better then anything stock.


Lantis wrote:Gargoyle is designed to provide powerful, yet simple configuration of your hardware.
The typical use case does not include selecting individual antennas for transmit or receive


Based on my reading of hundreds of your posts over the years, this was really not the answer I was expecting to hear from you Lantis.

There are posts on this forum from those who have seen, and taken the time to comment, on the 1043ND's comparatively poor WiFi performance in Gargoyle. I, too, had a 1043ND anchoring my network for years and experienced the relatively low WiFi throughput numbers. Never commented on it, because at the time, most of my wireless use was to phones. 30-40 Mb/s to a phone is generally not an issue. Laptops and bridges, well that's another matter, and drove me to replace my 1043ND (and other devices) with Archer C7s over the past 4 months.

Here Kabi has gone and found out why the wireless performance on 1043NDs is suboptimal, and provided a solution. Add an additional line in the code for the 1043ND's unique antenna deployment to effectively double WiFi throughput.

I agree that adding something so esoteric as assigning various antennas to specific functions is well outside the accepted use case of Gargoyle. And you said as much in your reply.

But not having rejoiced at the solution having been found to the 1043NDs' lackluster WiFi performance and immediately committing to add that single line of code to the specific gargoyle builds for 1043ND's seems a bit out of character for you. Perhaps there may be things that I/we don't understand about how individual builds are made that makes this code addition for a single unit a non-starter?

[NG]Owner

ispyisail
Moderator
Posts: 4809
Joined: Mon Apr 06, 2009 3:15 am
Location: New Zealand

Re: Faster wifi with 1043nd V1

Post by ispyisail »

But not having rejoiced at the solution having been found to the 1043NDs' lackluster WiFi performance and immediately committing to add that single line of code to the specific gargoyle builds for 1043ND's seems a bit out of character for you. Perhaps there may be things that I/we don't understand about how individual builds are made that makes this code addition for a single unit a non-starter?


This is a bit harsh. (A lot really).

There are only so many hours in the day that volunteers can provide.

The V1 is now at the "end of life" in respect to Gargoyle. (This will also be a factor).

Many years ago Eric stated if you send him enough money he will do what ever you want.

NGowner
Posts: 46
Joined: Tue Mar 09, 2010 4:12 pm

Re: Faster wifi with 1043nd V1

Post by NGowner »

Wasn't my intention, but I can see how it reads that way.

I also dug into the downloads section and learned that there are 285 (yes, with the help of Excel, I counted) different builds that have to be maintained just for the Atheros AR71XX chip on the 1.10.x branch alone. Ay caramba!

Perhaps the supported routers should be looked at and pared down to a more manageable level. Which would also, based on the age of the router in quesion, likely render the subject of this thread moot.

[NG]Owner

kabi
Posts: 29
Joined: Thu Aug 28, 2014 5:33 am

Re: Faster wifi with 1043nd V1

Post by kabi »

Lantis wrote:I do not think that option is necessary for your device. Plenty of users have sufficient speed without it.


There is no way to go beyond 50mbps without this option in my testings. Do you mean that by 'sufficient' speed ?

1043nd V1 is device which has different wiring with TX/RX - see here: http://svn.dd-wrt.com/ticket/2786

BrainSlayer ddwrt maintainer wrote:for 1043. the 1043 is no 3x3 mimo router. it has 3 antennas. correct. but it has only 2 transmission paths, but 3 receive paths. read the AR9132 CPU spec to get more details.


Seriously IT NEEDS TO BE FIXED.

Lantis wrote:It is an advanced, non-standard option, so there is no need for the GUI to expose it. The average user has no need to see the back end configuration.


Average user is supposed to believe defaults are OK - but that's not that case.

Lantis wrote:You should only use the country code that you are in. Using any other country code would be a violation of the relevant regulations for your country.


My house my rules ;) I don't need to worry about neighbours - they are too far away...

Lantis wrote:It works fine on Gargoyle. If you review the logs, it will probably even tell you that it has fallen back to HT20 mode.


I'll need to dig deeper into figure out where the problem is burried here - probably by reflashing dd-wrt and comparing why Windows work there and not for Gargoyle.

ispyisail
Moderator
Posts: 4809
Joined: Mon Apr 06, 2009 3:15 am
Location: New Zealand

Re: Faster wifi with 1043nd V1

Post by ispyisail »

Just ignore this post Lantis. (You do a fantastic job by the way)

I know where I would like your energy spent and its not on an old V1

@kabi your welcome to push any code edits to the repository
https://github.com/ericpaulbishop/gargoyle

or

Add a well structured request in "Issues"
https://github.com/ericpaulbishop/gargoyle/issues

kabi
Posts: 29
Joined: Thu Aug 28, 2014 5:33 am

Re: Faster wifi with 1043nd V1

Post by kabi »

ispyisail wrote:Just ignore this post Lantis. (You do a fantastic job by the way)

I know where I would like your energy spent and its not on an old V1

@kabi your welcome to push any code edits to the repository
https://github.com/ericpaulbishop/gargoyle


I'll need to likely check LEDE first - since looking at some existing opened issues and they way they are resolved isn't inspiring (i.e. Allow us to disable non-important things like bandwidth monitoring from GUI is particularly interesting for me)

ispyisail wrote:Add a well structured request in "Issues"
https://github.com/ericpaulbishop/gargoyle/issues


Issue: https://github.com/ericpaulbishop/gargoyle/issues/751

Post Reply