Hi @all

I bought a new WRT 3200 acm. But i have so bad 2.4 ghz wifi performance. (max 20 mbit)

i use
OpenWrt Chaos Calmer 15.05.1 r49389 / LuCI for-15.05 branch (git-16.206.66468-6f5c466)
Kernel Version    3.18.36

:~# strings /lib/modules/*.*/mwlwifi.ko | grep "^10.3"
10.3.2.0-20161011
10.3.2.0-20161011
:~# opkg list-installed | grep mwlwifi
kmod-mwlwifi - 3.18.36+10.3.2.0-20161013-2

I am now so unlucky, why the wifi performance was
s the reason that i buy him.

Please I hope somebody have a solution for me :-(

thx v much !!

(Last edited by alesi77 on 19 Aug 2017, 14:40)

The wifi driver installed is almost a year old, and there have been many, many updates since that time.

If you want something more recent then you'll ether need to download from trunk, a 3rd party build, or compile your own.

Best Regards,

@davidc502

hi now i try Lede Reboot SNAPSHOT r4651-a6f6f8d / LuCI Master (git-17.209.51293-4c9ae3f)

everything looks now good .

thank u so much.

I figured I'd piggy back on this thread rather than create a new one even though my issues/questions are a little different, the subject fits.

I installed the LEDE stable build 17.01.2 on a WRT3200ACM and I'm not sure if what I'm seeing with WiFi is normal.  2.4GHz seems rock solid to me, so no issues there.  5GHz, however, there's the two 5GHz WiFi radios.  By default LEDE configures the same SSID/channel/bandwidth for both, however, regardless of settings the radio0 defaults to full power, while radio2 seems to use only 15mW max.  But the odd issue is, if I set different SSID's for both, I never, ever see the SSID for radio0, I only see the SSID for radio2.   Worse, since that's the one with 15mW power, that means range is REALLY bad.  If I leave the room and have one single wall between me and it, speeds drop through the floor (speeds are great if I'm physically close to it), and reception is very low.  I can get signal 1-2 rooms away but at 3-8mbps out of 50mbps type performance.

Does everyone have similar experience with the WRT3200ACM?  And is it specific to open source or is it that way on stock as well? (I haven't actually tried it stock, I bought it specifically for OpenWRT to use the OpenVPN client (I did find a bug in OpenVPN's luci panel that configures it wrong if setting a TLS client or server I'll have to figure out how to report to them.)

The 5GHz also seems to drop sometimes, but since I can't get radio0's SSID to show, that's radio2.  And if I change radio-'s SSID it goes in a bad state and won't come up again until I reboot.  I have heard that the dropped 5ghz has been a problem even with stock firmware.

I've heard that dd-WRT has a newer build this week using a newer driver, but the forum thread over there shows someone with the same problems as before.  I'm not sure what LEDE snapshots are up to atm in terms of drivers. I've never used dd-WRT so that's just a reference point for me, I'm more familiar with OpenWRT.

@tux-o-matic

Good news is that we don't have control freaks for moderators here, and we don't have the dreaded "thread" gestapo    So no worries.

Okay, so there are probably a few things you may be unaware of......

1st. the FCC mandated vendors lock down the power table for Wifi. This means power is already programmed in and is not adjustable on any channel.  So, you are stuck with the power given. The work around would be to use range extenders, and others have had success using high-gain antenna that focus the RF horizontally giving more range.

2nd. On 5Ghz we have DFS channels available to us. However, when we choose a DFS channel, the router has to sit back and listen for a signal before bringing the wifi interface up, and this process may take a couple of minutes.   You may wiki 5Ghz and find out what channels are DFS, and see if that's what is delaying the SSID from being seen.

3rd. Third party builds are likely to have the latest wifi drivers. Also, daily trunk snapshots found on the LEDE site may or may not have the latest drivers depending on if someone commits those changes. Today, I believe LEDE trunk has the latest drivers.

4th. OpenWrt will be merging with LEDE, and LEDE will be changing its name to OpenWrt by the time the next release comes out, so feel free to try LEDE as it was build on the OpenWrt foundation to begin with.

Let me know if that helps to clarify or if you have further questions.

@tux-o-matic, If using LEDE stable you can load the latest mwlwifi, but check issue #198. LEDE master is still one version back, with PR1288 waiting to be pushed, but there is that issue. Try turning radio2 off.

(Last edited by Villeneuve on 25 Aug 2017, 16:36)

@davidc502 Thanks for the great info!   Yeah, I tend to get paranoid when entering a new forum in terms of posting threads. Lots of sites are pretty scrutinizing of new threads!

1. I was aware the power max was locked down (which actually makes sense considering it's licensed and tested at a certain power output, despite all the hate for the decision), but I figured you had all the room from 0 to max to play with to reduce it, not that it was fixed at whatever default it was set to. I can see the upper limit being unalterable, but not fixing it permanently AT that limit!  I  was playing with a Tomato build with an older router and found the power settings still worked, though.   On the 3200,  Radio0 and Radio1 both are 22mW.  Radio2 is apparently capped at 15mW (or that's what it's using...I assume it's capped, but I don't know.)

Range extenders, I was considering using one of these as an extender.   I wanted to take advantage of the MU-MIMO functionality that none of the dedicated extenders have.  Though I keep an Almond + around since the built in UI makes it easy to shuffle between my router projects

2. Interesting about DFS.  I'll have to do some research on that.  I suspect that's not the issue here as it was (A) the default channel set by LEDE, I think 36, and more importantly (B) both radio0 and radio2 were set to the same band and I wasn't seeing the delays on radio 2, just radio 1 (which never became visible) (though radio 2's 15mw signal is so low, it's almost useless....it's clear radio 0 is the 5GHz I SHOULD be having access to at several times the power output.  Unless they're supposed to operate in tandem? There's a LEDE thread about the third radio but it's a bit over my head.  Something about a radar/bluetooth/nfc radio that's not even used in the official firmware yet.  Maybe I should try totally disabling radio2!  I hadn't thought of that before. It might be interfering with radio0 somehow. I'll give that a try later on.

I did notice that if I map the 2.4Ghz LED light to radio1 (2.4ghz), which technically seems hardware-mapped already, radio2 goes inactive and unable to start until reboot.  So radio2 definitely has something odd going on at a hardware level.

3. I might have to give the trunk a try next on my next unit.  I gave stable a spin since it's stable, and since it was my first flash on a WRT3200 so I wanted it to go well.  I ordered two units, right from Belkin, and they sent me two units that were WRT3200ACM's on the OUTSIDE.....but one of them is apparently actually a WRT1900AC (not even ACS!) on the inside!  Ugh!   I might have to rethink buying Linksys again, but I got tired of Asus' physically fragile hardware.  Especially the amplifiers.

4. I'm glad to hear the two projects are merging.  It's a confusing mess right now.  Like Libre Office and OpenOffice, these projects really don't benefit from these political schisms that nobody but the dedicated community actually follows. It makes everything look like a fly-by-night, and right now dd-WRT is the one that looks like a large scale, steady project despite the opposite being true.  I've been using OpenWRT for years, went to start on this project and found all the confusing mess of LEDE etc.  Though I have to say, I like the cleanup that's happened under LEDE's watch. A little Tomatoification of OpenWRT made it, IMO, jump from being the "bloated confusing Linux hacker distro for routers" i.e. Debian, and turned it into a nice, clean, usable system.  DD is looking more like the disorganized one now.  (Some incomplete Luci panels notwithstanding )

@Villeneuve

Thanks for the info!  I didn't know mwlwifi could be updated independently!  That's pretty useful!  But....#198 sounds pretty awful.  I may hold off on updating until that gets resolved.  It might make matters worse than they already are by causing problems for the wired & 2.4 GHz connections that work fine currently.

I'm definitely going to try turning radio 2 off.  I was hesitant to do that since I wasn't sure if it was somehow part of the AC handling, but it makes sense that a third radio fixed at low power, intended for BT/NFC features not even used in OEM now might not play nice with the real radio0. I suppose there could be an actual hardware problem with radio0 as well, though the OS is pretty convinced it's up and running, so if it's not running, it's doing a convincing job of returning status.

@tux-o-matic

Yeah, Radio 2 is a bit of a mystery... I ended up just disabling it, and there has been no negative effect. There have been a couple times where I've tried to actually use the Radio2, but after about an hour it goes completely to hell, so gave up.

As to the Power rule, Linksys put the power table "on chip or EPROM", so that's how it is locked down. It doesn't surprise me any older router wouldn't be affected by the new FCC rule. I also have a 1900acs Version 1 where the power is completely configurable. However owners of the 1900acs V2 would be affected including the 1200ac V2.

MU-MIMO --- Currently doesn't work on the 3200acm. It has been recognized as a issue, and this feature will be added, but no time table.

DFS -- When a 5Ghz DFS channel is selected, whilst the router is listening to the channel before bringing the wifi interface up, 2.4Ghz will be available as normal, and I believe you suspected that they are not in tandem. So, I agree this seems to be another separate issue. Find a build with the latest wifi driver "10.3.4.0-20170810"  << Notice "2017" "08" "10"  Now you know the latest commit will have the date in the driver name. 
opkg list-installed | grep mwlwifi

You may have to completely forget the SSID with all of your clients before rejoining, so give that a try.

*EDIT*
Kaloz mwlwifi >>  https://github.com/kaloz/mwlwifi/commits/master

(Last edited by davidc502 on 25 Aug 2017, 17:25)

@davidc502

Hah, well at least I'm not the only one at odds with radio2!   I'll have to see what happens to radio0 when I take radio2 down. With luck I'll start seeing the SSID.   I checked the wiki, and anything between 50 and 144 are DFS.  I think I'm on the default 36, so while crowded, DFS shouldn't be involved (unless there's a bug.)

I want to hate the FCC for the rule, but I can't get past the part where it actually makes sense to lock the transmit power on a consumer radio to the certified max. I just think the weird implementation of not being able to REDUCE it, either is idiotic.  Fixed max shouldn't mean fixed AT max.  We'll always have ham for shoving 100 watts into a coat hanger when we want to......

Thanks for the MU-MIMO info.  Is that in the Marvell driver or an OpenWRT/LEDE specific issue?  That's kind of a bummer...might as well go back to my (kind of low power also) Asus N88U AC as an AP if I'm not getting MU out of it.

I think I'll hold off on updating the driver due to that lockup/reboot issue worse than my current problems (unless I discover it happens anyway with the current driver.)

@tux-o-matic

From my understanding MU-MIMO is purely a wifi driver issue which needs to be worked on. However, give the 3200acm a try with the latest driver as you might be pleasantly surprised by its performance.

Not sure about the lockup/reboot issue?? 1900ac v1 owners have this issue running the latest kernel, but believe that's the only known reboot issue right now that I remember....  There might be someone out there with a 3200, but is probably a isolated issue.

@davidc502  Ahh, so if Marvell's involved that'll take ages

I'll maybe try the new driver on the next build.  Whenever I get it.  Linksys is rather annoying, they want to set up a call now for discussing the fact they sent me a WRT1900AC inside a WRT3200ACM shell....how to take a simple problem (that should never ever happen) and make it a difficult one.  Fun fact: The FCC IDs on the sticker aren't even valid for the actual radios.  These guys have their acts together!

I'm looking forward to playing with it more though.  Might try those high gain antennas at some point too.  I've never been happy with my AC range even on my old Asus (it's better than 15dbm radio2.....but not by much.)

Sadly, the lockup/reboot issue, the link @Villeneuve posted for the issue shows multiple people that managed to duplicate it, and there's a few mentions in the dd-wrt forums about it as well, so it seems like a legit error, but seems to be created specifically by android devices doing...something...other devices don't.  Since I have Android devices on the net, I can see wired locking up with it from time to time.   Still, on build 2, I'l give it a shot.

Success!  Turning off radio 2 brought radio0 back to life!  So apparently either radio2 interferes with radio0, or radio2 just cant be ysed in ap mode with an ssid broadcast. maybe it can be used hidden, or in client/ad hoc modes.

That said signal strength is only marginally better.   I have 7db antennas on the way.

IIIRC wrt3200 supports band steering.   Does anyone know if that functions in LEDE?

(Last edited by tux-o-matic on 26 Aug 2017, 00:18)

So I think I bricked it.  I installed stable. I installed the new precompiled driver.  Rebooted....and now I can't connect at all to it.  SSH never displays the login prompt, Luci, sits at "LuCI - Lua Configuration Interface" forever then times out.  I tried holding reset during a reboot, and it seems to not have reverted to the Linksys partition.

Edit: Weird, on lan port 1 it thought it had internet access even with wan disabled.  I switched to lan 4 and it still wasn't connecting.  I connected the wan back in and suddenly it worked.  It seems like stable needs an internet connection to load enable login? Both Luci and ssh couldn't get a login without it.

It seems like it did reset, but it reset back to openwrt defaults rather than linksys.  My settings are cleared.  I'll try installing the driver again.

(Last edited by tux-o-matic on 28 Aug 2017, 16:12)

Power switch 4 times. And this may be of interest.

Thanks.   Somehow weirdness is going on.  That first reset DID clear the settings (but didn't revert to linksys, which is weird considering the only two flashes on the router are linksys' own upgrade image, then openwrt).  I'm not sure why it's requiring internet to connect, but it actually is,  Seems it's unrelated to the driver.  After I cleared it, I installed wget again, grabbed the file again, then rebooted before remembering to actually install it...and the same thing happened.  It took FOREVER to come back up.   A lot longer than my other unit and ridiculously long compared to OEM.  Weird.

For future reference though: When you say power switch 4 times....does that mean in rapid succession or with a specific delay between?

Some time ~2-6 seconds iirc, one LED will light, power???? maybe.

For a moment there was going to ask if it went into a boot loop...   but appears the problem has been worked out?

@davidc502

Yeah, honestly I'm not sure what was going on.  It came up normally, flashes power and internet lights, then 5ghz and lan lights come up and pulse normally, but I couldn't connect to administration including sitting at the "LuCI - Lua Configuration Interface" message before timing out.   SSH times out.  The web server obviously responds but can't do much else.  It displays that text but never loaded Luci.

But it's apparently just a long slow boot, it eventually comes up normally.   It's pretty weird. ran and worked on it over a few hours and it was fine.

The first time when I reset and it didn't revert to linksys I thought it was bricked, but when oWRT came up, it had cleared all settings and reverted clean, so it must have worked, but somehow a single oWRT flash overwrite both partitions? Odd.  I don't mind as long as a partition is salvageable!   I thought it was related to the new driver, but it's not, because the first reboot after that without installing the new driver mod had the same effect, so it's just weirdness of booting this particular unit with oWRT.  Now I have the new driver loaded.  I'm getting signal meters on radio0 now, so that's good. I'll have to do some strength testing with the new driver later on.   I put my 7db antennas on the other unit over the weekend and saw pretty much no change, even laying them horizontal to push signal straight down to the floor floor below.  Still getting 4/5 bars signal on 5ghz where my piddly Asus gets 5, and getting a slower speed out through a few walls where an Almond+ doesn't.   Which is weird given this thing has a decent radio power, and those are some big antennae. The antennas were a disappointment. Broadcast power of 23dBm isn't so bad, so it should be performing better than that.  Here's hoping the driver improves that a bit!  Right now I'm connecting without any external antennae at all.  Never tried that on a router before.  Works great up close. The fun things we discover when geeking out....

It seems like the new driver might be providing a somewhat stronger signal.  Still seeing lower speeds and low strength in certain locations but the bigger antennas might be putting those spots in the "donut" as well.

I'll do some more testing at longer range, today or tomorrow, but so far it might be closing the gap a bit.   Still not the best wifi signal I've seen from a router.....some very consumerey routers seem to have better 5ghz, but it's better than it was I think.

With 5Ghz, put it on channel 161, as you will get the strongest signal there.

Interesting.  That just shot a broadside through my understanding of wireless radio signals.  I thought the higher the frequency the LESS distance and penetration it was capable of.  But 161 is higher frequency than 36.   I'm going to have to do some research on that one!

I made the change, but haven't been able to do much range testing yet.  I think it made a positive impact but I haven't checked the limits of the range yet. Thanks for that tip!

I noticed something between the two builds.  The two hardware units seem to be supplying different power tables!  The one build seems to top out at a 30dBm TX-power, 199mW.   The OTHER unit tops out at 30dBm, 1000mW!!  One unit has 5x the xmit power!   I would assume that's a difference in OpenWRT but since you said the power tables are locked and can't be changed, there's a hardware revision where supposedly there are no hardware revisions.

The 1200ac V2, 1900ac V2 and 3200acm all have the power table on chip, and  hence power can not be changed.

Are you saying these two units are identical?

Yes, both are WRT3200ACMs. AFAIK there's only one hardware version for the 3200, but the power levels aren't going beyond 23dBm on that one either in the (useless) drop down or the actual reported tx power. Might be a defect.  Might explain my mediocre range even on the highest channel and with 7dBm antennas.  199mW and a a full watt are a pretty different tier of output.