ASUS Merlin & wificall and / or facetime = drops connection ;(

[wally_tm]

Hi,

So first and foremost I would like to thank RMerlin guy for mashing up the merlin version of firmware. For most of the part it works very much awesome. I use it at home on AC87 and at work on exactly same router AC87 because this firmware gives better VPN options than vanilla firmware.

But there is a caveat - for some reason Merlin firmware does not seem to play nicely with apple portables over 5GHz :/

Over some time I've managed to narrow down the problem to the fact that when there is an merlin firmware and iphone (4, 5, 6, 6 plus, 6 plus) or ipad (mini 2, mini 3) is connected to 5GHz txrx with pre 58 firmware connection will get stuck and while viewing stations connected ... all apple products get struck at a very specific link bitrate (from mobile to station) and there is absolutely no traffic on the link - no matter what you do on hand held, data will not get through.

You can bail out of this situation by
- simply pinging hand held from any other point in the networ (this is a weirdest part !!!)
- switching wifi on hand held on/off
- rebooting station
- getting out of range and comming back
- connecting to different station


This problem usually occurs when router was up for a day or more.

Now, with 58 firmware I've not seen this issue yet BUT there is a persistent problem with data transmission between hand held and station. In normal situation when you use WIFIcall your call will get reliably dropped after 1 to 5 minutes. Facetime (audio) gets dropped as well but not as reliably. Airplay music playback works pretty good, but you can not leave it running for hours because it does drop out periodically.

Now all of the above does _NOT_ exist while having a asus firmware on the router.

About the routers:
- both bought from different sources, I bought mine somewhere in europe, company one was north of UK
- both bought with few months of time separation
- both configured with minimum configuration to get best performance (no packet inspection and other stuff). Only my home router has download manager and samba enabled. Other than that, ipv6 is disabled ( does not make any difference).

 

[RMerlin]

Apple devices have compatibility issues with the Quantenna chipset used as the RT-AC87U 5 GHz radio. Disabling beamforming usually helps, but it's not guaranteed to fully resolve it.

 

[wally_tm]

Cheers for prompt reply.

Problem is, that everything work 100% while using asus firmware, problem only occurs on merlin flavor firmware. Same settings, beam forming on etc.

On 58 firmware there are no more "stuck 5GHz" issues anymore, but there are wificall issues still occurring.

Well actually I've managed to limit a lot of connectivity issues by dropping RTS down to 2304 (and by that I mean connectivity for other applications and protocols, wifi call still sucks. )


Edit:
Is there any debug I can perform to help with resolving this issues ?

 

[RMerlin]

Cheers for prompt reply.

Problem is, that everything work 100% while using asus firmware, problem only occurs on merlin flavor firmware. Same settings, beam forming on etc.

Asus firmware uses a newer wireless driver for the Quantenna, and they haven't made it available until only two weeks ago, so 380.58 was still using the older driver. (It took nearly a year to Quantenna to resolve the iOS issue.)

You can try out the 380.59 alpha2 build, which uses the newer driver.

https://www.mediafire.com/folder/bj94sbhrh7e49/Test Builds

 

[wally_tm]

Thanks a lot

I will certainly try it.

_But_ maybe I'm not expressing my self correctly: I've had this routers in service for best part of the year now and older versions asus firmware did _not_ had the 5GHz wifi issue, NOR the wificall issue. Even firmware that routers were shipped with had no issue from our stand point.

fyi, I don't have anyone here with android device supporting wifi calling on network that actually provides wifi calling ... hence I'm only reporting it as ios related devices issue.

 

[RMerlin]

Thanks a lot

I will certainly try it.

_But_ maybe I'm not expressing my self correctly: I've had this routers in service for best part of the year now and older versions asus firmware did _not_ had the 5GHz wifi issue, NOR the wificall issue. Even firmware that routers were shipped with had no issue from our stand point.

fyi, I don't have anyone here with android device supporting wifi calling on network that actually provides wifi calling ... hence I'm only reporting it as ios related devices issue.

The iOS and Android issues started around early last year, when Asus updated the Quantenna firmware. It introduced stability issues with iOS devices, and major battery drains on Android devices. The iOS issues are well documented on various forums, and back then the main workaround had been to disable beamforming.

Quantenna (and Asus) resolved the iOS-specific issue with an update released last autumn. The Android power drain issue remained, until (from what I've been told) the new MU-MIMO capable driver that was released two weeks ago.

The wifi driver code is closed source, so I use the exact same driver as used in the stock firmware, therefore wifi stability (or instability) will always be identical to whichever Asus original firmware also used the same driver. That's one part of the firmware over which I have no control, therefore performance will always remain unchanged.

Try the alpha build - it will at least allow you to compare against the latest driver as used by Asus in 3.0.0.4.380_2695. I can't remember how old the 380.58 driver is, but it's certainly many months old, probably from one of the completely broken Quantenna releases.

 

[wally_tm]

Just uploaded in on my work router.

What to I need to do to make sure that new driver is being used (I think in past it required a hard rest with the button)

Lets try that path for time being and see where it gets us.

As a side note I've noticed that android had a bug with wifi driver that was somehow never resolved that if you had dtim interval bigger than 1 driver would just ignore sleep mode.

I've read it somewhere in thread for a bit advanced distributed networks that this problem did touch them a lot earlier because they had cutting edge routers and were able to actually do something with this values (their default was changed from 1 to 5, and first ios devices went hay wire than after few months androids started to crop up, possibly due to slow upgrade process, but with ios 9 most problem went away for them). It was predominantly in some schools ... can't remember name of the system / hardware but it was a managed distributed access point system with centralized management.

From my experience if you set dtim = 1 there would not be a problem and you could set time gaps to insane numbers like 5 seconds and still everything would work ok and battery life would be dramatically increased (samsung note 3 was my test rig, but my colleagues xperia z2 has battery life of 5 - 7 days regardless of setting)

 

[RMerlin]

Just uploaded in on my work router.

What to I need to do to make sure that new driver is being used (I think in past it required a hard rest with the button)

It's generally recommended to do a factory default reset, however there's a chance it might not be necessary in this case (since the 5 GHz radio is mostly managed by a separate firmware). I would advise however doing a full power cycle, not just rebooting.

Only do a factory default reset if you experience any weird issue.

As a side note I've noticed that android had a bug with wifi driver that was somehow never resolved that if you had dtim interval bigger than 1 driver would just ignore sleep mode.

It could be related. I vaguely remember a mention of the issue being indeed tied to DTIM, but we never got any clear technical explanation as to the root cause. It's possible that something Quantenna changed last year triggered the issue. Why it took them so long to fix it however is beyond me.

I've written Quantenna off as a bad idea for home devices. They lack the expertise (IMHO) to deal with a diverse home environment, compared to their usual business-oriented experiences where they have better control over the whole environment. Even the gamble to get MU-MIMO out earlier failed - BCM's beta support came out at about the same time as Quantenna's own beta support, give or take a few weeks.

 

[wally_tm]

You are preaching to converted I've always been a fan of skull trail system, and bought into it with first asus implementation for dual xeon ... it worked absolutely awesome ... when it worked ... year or two back I upgraded my self to dual lga2011 ... again it is just a monster power horse ... but some things are not so great:
- power managment - 2 years to make it work with going to low power states (not sleep, I mean actual cpu C states ... you had to manually dissable C4 C5 C6 C7 on some cpu's)
- marvell sata controller (extends what north bridge gives by default) - 3 years and not fixed ... just splats your raid with errors ... had to use separate intel SAS card to cope with all the disks
- VTd works ... just barelly, but if you want to enable any fun stuff like page translation of faster IO access you will corrupt everything :/

So yeah, my opinion about Asus is : "great hardware, underpaid firmware"

going back to original:
@RMerlin
Is there a way I could check a version of wifi module loaded to check whenever is loading a propper one ?

 

[RMerlin]

Is there a way I could check a version of wifi module loaded to check whenever is loading a propper one ?

You will always have the "proper version" loaded. The factory default reset has to do with low-level settings that might change between driver revisions (such as power output levels per streams). In short, you can't have a firmware running the incorrect driver - driver is part of the firmware itself.

 

[wally_tm]

@RMerlin

Thanks a lot mate !

I did as you suggested 380.59_alpha2 + hard power cycle and everything so far seems to be a OK ! Didn't wanted to reply to early to no jinx it but so far there are no dropped calls what so ever !

On a side note, I gave MU-MIMO a go and it delivers an astonishing results !!! It's not like I've got peta bytes / sec transmission but at least now everybody seem to be nicely negotiated, there are far less hickups ... signal reaches to a remote conference room ... and even thou there is supposed "1 bar" on laptop or on mobile every thing sill work absolutely ok ! So far now it works so great that it seems like seamles hand of between wifi and carrier - if you get to edge of wifi - rather than getting usual crappy performance you get ample to the point that it simply switches off an caries on on 4g with data transmission. I even got tempted and tested new wifi call & volte hand off and did not get a dropped call yet !!!!!

I'm gob smacked !

Anyway, thanks for great firmware and pointing me in right direction - I was nearly loosing hope !


Mod - ready for closure, should I add anything in the title ? [soved] etc ?

 

[RMerlin]

On a side note, I gave MU-MIMO a go and it delivers an astonishing results !!! It's not like I've got peta bytes / sec transmission but at least now everybody seem to be nicely negotiated, there are far less hickups ... signal reaches to a remote conference room ... and even thou there is supposed "1 bar" on laptop or on mobile every thing sill work absolutely ok ! So far now it works so great that it seems like seamles hand of between wifi and carrier - if you get to edge of wifi - rather than getting usual crappy performance you get ample to the point that it simply switches off an caries on on 4g with data transmission. I even got tempted and tested new wifi call & volte hand off and did not get a dropped call yet !!!!!

MU-MIMO has no impact at all on range. All it does is better allocate the available bandwidth if you have MU-MIMO compatible clients.

Any range improvement you see is probably from other changes in this new driver.

 

[wally_tm]

Hang on, mu-mimo allows MUltiple users to have Multiple Input (RX) and Multiple Output (TX) paths ... right ?

If so than if there are more than one station on the network talking at the same time they can both benefit from effect of diversity unit (which is actually part of RX for combining multi path signals).


So this is equipment that I used to develop in the past:
http://broadcastrf.com/gigawave-mvl-hd3-receive-system/

Idea is that if you combine signal from multiple inputs (preferably if it was transmitted on different paths / different polarity antenas etc), yes you might get a slight signal increase BUT you get an out standing result of dramatically decreasing noise floor due to noise self canceling. Also any transient interference always have a lot lover chance to affect multiple paths.

We used a 4 way diversity units to pick up a signal from 100mW 2.4 Ghz camera system transmitting full hd over 64QAM (54Mbps) with line of sight but no ground clearance from distance of 2 miles ... or from Olympic stadium in Beijing without even a direct line of sight at distances of 200 yards with tons of concrete around us.

Trust me, mimo rescues low signal levels a lot through noise canceling effect !!!

 

[sfx2000]

MU-MIMO has no impact at all on range. All it does is better allocate the available bandwidth if you have MU-MIMO compatible clients.

Any range improvement you see is probably from other changes in this new driver.

Maybe some clean up or fixes in the TxBF side... and that would help SU clients as well as MU in 11ac/VHT mode.

 

[RMerlin]

Routers have been supporting MIMO for years already. MU-MIMO adds on top of that by allowing devices to do it simultaneously. Any range improvement would be from MIMO and Beamforming itself, not from MU-MIMO specifically (which still requires clients to specifically support it).

Basic info on MU-MIMO: http://www.smallnetbuilder.com/wireless/wireless-features/32787-how-mu-mimo-makes-your-wifi-better

 

[wally_tm]

Ok guys, maybe I've not stated it clear enough - at the border of the network ... exactly the same place as before (this is before mentioned conference room) , signal strength (silly bar indicator) is minimal BUT wifi performs _good_. Signal was always available there, but people were ending up disabling wifi on their phones to use 4G because data over wifi was choppy ... and some of clients were holding on to 5ghz like crazy ... not trying to fall back on 2.4. So my results for mu-mimo vs no mu-mimo with this router and this specific firmware is is that devices benefit at the border of the network in terms of reliability of connection and real data throughput.

I'm not stating that there is some magic stuff going on there ... I know what improvements multi diversity units should bring to he table.
Our setup is not perfect, over 30 wireless devices hanging on single router with 2 more tiny companies using their wifi in a very close proximity (same building).
I'm reporting that becuase if somebody else is experiency an issue with enabling mu-mimo on their ac87 maybe we could cross reference some data and squash another bug at the same time ?