I wrote a how-to on my blog that describes in detail the performance differences between:
* using your existing WiFi network (“Standard” mode),
* using SonosNet 2.0 (“Boost” mode) with an Ethernet-connected Sonos speaker,
* using SonosNet 2.0 (“Boost” mode) with an dedicated BOOST device, and
* adding a second and then a third dedicated BOOST device
Initially, I had only two Sonos One smart speakers, each in different rooms. I used Standard mode, connecting them to my Wireless AC WiFi home network. As I added more Sonos speakers, I encountered occasional audio drop outs, particularly noticeable with paired speakers and particularly with lossless audio. Now, I have purchased three Sonos BOOST hardware devices to feed over a dozen Sonos One smart speakers.
In the article, I walk through using the Sonos Network Matrix to optimize your network (and therefore audio) performance. Beyond the red/yellow/green color codes, I describe how to understand signal strength, noise floor, and OFDM ANI level. I also provide tips for coverage, avoiding wireless interference, network management, and enabling STP on your router.
The full article is here:
http://freetime.mikeconnelly.com/archives/6050
If you have any feedback, please let me know!
Mike
P.S. Also be sure to read my articles on:
Sonos One Smart Speaker (Part 1): The State of Smart Speakers
http://freetime.mikeconnelly.com/archives/5648
Sonos One Smart Speaker (Part 2): Review of Its Voice Services Including Amazon Alexa & Google Assistant
http://freetime.mikeconnelly.com/archives/6004
Page 1 / 2
Really, really great stuff here, Mike. I have read these posts with interest. Great content. I do hope to see you around the community more! Let us know if you need anything. Many thanks.
Thanks Edward R. I am new to the Sonos family with the new Sonos One. So likely more feedback as I use Sonos more.
One question: I am still trying to diagnose audio dropouts on one of two paired speakers with lossless (ALAC) audio played from my '17 iMac. As best as I can tell, it works well with 2-3 rooms but starts to falter in Party Mode. This leads me to believe it is a network bandwidth issue since I don't experience the dropouts with (lossy) Internet audio streams, which are much lower bandwidth. This happens even with three BOOST devices. Diagnostic code: 8345731
For a given audio source, is Sonos streaming audio to the Sonos speakers as multiple unicast streams or a single multicast stream in BOOST mode?
One question: I am still trying to diagnose audio dropouts on one of two paired speakers with lossless (ALAC) audio played from my '17 iMac. As best as I can tell, it works well with 2-3 rooms but starts to falter in Party Mode. This leads me to believe it is a network bandwidth issue since I don't experience the dropouts with (lossy) Internet audio streams, which are much lower bandwidth. This happens even with three BOOST devices. Diagnostic code: 8345731
For a given audio source, is Sonos streaming audio to the Sonos speakers as multiple unicast streams or a single multicast stream in BOOST mode?
Hi Mike. I have only had a chance to skim through your blog article and look forward to reading it more carefully, but it looks to be technically accurate (insofar as I am able to judge), clearly structured and well written. Which basically ticks all the boxes. Great job.
A few minor feedback points
1. it is a bit Sonos One-centric in a couple of places, reflecting your own experience, but could be made more general, e.g. you say 'If at least one Sonos One speaker can connect via Ethernet to your router...'
2. There are further areas you could cover in due course if you felt so inclined, for example:
(a) The Bridge and SonosNet 1.0 (although maybe that's a detail too far)
(b) How things operate with a Playbar/base surround system
3. I think turning everything green in the matrix may not be as important as might be inferred.
Those are just minor quibbles and suggestions. It's a great contribution to the forum, which many people will find useful. (If they can find it given the rubbish search tool on the forum!)
(In answer to your question I think Sonos uses multicast, but don't trust that, wait for official confirmation!)
A few minor feedback points
1. it is a bit Sonos One-centric in a couple of places, reflecting your own experience, but could be made more general, e.g. you say 'If at least one Sonos One speaker can connect via Ethernet to your router...'
2. There are further areas you could cover in due course if you felt so inclined, for example:
(a) The Bridge and SonosNet 1.0 (although maybe that's a detail too far)
(b) How things operate with a Playbar/base surround system
3. I think turning everything green in the matrix may not be as important as might be inferred.
Those are just minor quibbles and suggestions. It's a great contribution to the forum, which many people will find useful. (If they can find it given the rubbish search tool on the forum!)
(In answer to your question I think Sonos uses multicast, but don't trust that, wait for official confirmation!)
Multiple unicast, at layer 3. Unfortunately the picture at L2 is not entirely clear from the matrix, since Direct Routing will try to short-circuit the STP topology between grouped (and paired/bonded) nodes.
IME having multiple wired nodes might, perversely, give rise to a suboptimal L2 topology in some cases. Specifically, wiring otherwise wireless nodes together can result in Direct Routing being defeated.
And of course the choice of group coordinator can be a significant factor in terms of link bandwidth utilisation.
IME having multiple wired nodes might, perversely, give rise to a suboptimal L2 topology in some cases. Specifically, wiring otherwise wireless nodes together can result in Direct Routing being defeated.
And of course the choice of group coordinator can be a significant factor in terms of link bandwidth utilisation.
@JohnB. Thanks for the feedback. I based the article on devices that I own. If I buy a Connect:Amp or a Playbar (after it is updated with HDMI-ARC), then I will fold those devices into my articles.
I didn't cover Bridge / SonosNet 1.0 since those are now obsolete.
My understanding is that ANI levels correspond to the cell color codes. 0-3 is green; 4-5 is yellow; 6-7 are orange; 8-9 are red. However, it isn't clear to me why the color codes differ on the horizontal/vertical axes versus their intersection. I would think they would all show the same color.
I didn't cover Bridge / SonosNet 1.0 since those are now obsolete.
My understanding is that ANI levels correspond to the cell color codes. 0-3 is green; 4-5 is yellow; 6-7 are orange; 8-9 are red. However, it isn't clear to me why the color codes differ on the horizontal/vertical axes versus their intersection. I would think they would all show the same color.
Colours in the left column reflect ambient noise conditions. Colours of cells of the matrix body relate to wireless tunnel signal strength.
@ratty If it is multiple unicast streams, then that likely explains why I am having audio dropouts with lossless music in Party Mode for 14 speakers given that all speakers are using the same 2.4Ghz band and channel.
If I understand what you are saying, my use of three BOOST devices with Ethernet backhaul between them potentially hurts my configuration?
If I understand what you are saying, my use of three BOOST devices with Ethernet backhaul between them potentially hurts my configuration?
It's the choice of group coordinator which is key. This is the room which starts the group (and, in the case of a pair, the left unit). Best results should be obtained by choosing a wired coordinator or, failing that, one with a strong wireless link direct to a wired node.
The 3 wired BOOSTs should be fine, though traffic won't necessarily follow the STP tree. Where it can, Direct Routing will be used between bonded/grouped nodes. The matrix doesn't show this.
A wire strung directly between two nodes -- which jointly remain wireless -- is generally not a good idea. It messes up Direct Routing.
The 3 wired BOOSTs should be fine, though traffic won't necessarily follow the STP tree. Where it can, Direct Routing will be used between bonded/grouped nodes. The matrix doesn't show this.
A wire strung directly between two nodes -- which jointly remain wireless -- is generally not a good idea. It messes up Direct Routing.
@ratty
Great feedback! I folded your information about color codes, multicast audio, and group coordinator into the article.
It seems odd that whatever specific room/speaker a user selects first (and thus group coordinator) makes a difference in overall Sonos system performance. I would expect this nomination to be handled automatically and, perhaps, even run within a wired BOOST.
However, I am seeing something that appears different: ALAC music plays without dropout for a few rooms. Play all rooms (14 speakers) in party mode and I consistently hear right channel dropout. Keep playing the audio and reducing the number of rooms, doesn't help. I can even go all the way back to a single room, where I started and will continue to get dropouts. Stop the song and start again in the original room and I won't hear the dropouts. And no issues with Internet radio stations.
I uploaded another diagnostics test for Sonos to confirm: 8347153
Great feedback! I folded your information about color codes, multicast audio, and group coordinator into the article.
It seems odd that whatever specific room/speaker a user selects first (and thus group coordinator) makes a difference in overall Sonos system performance. I would expect this nomination to be handled automatically and, perhaps, even run within a wired BOOST.
However, I am seeing something that appears different: ALAC music plays without dropout for a few rooms. Play all rooms (14 speakers) in party mode and I consistently hear right channel dropout. Keep playing the audio and reducing the number of rooms, doesn't help. I can even go all the way back to a single room, where I started and will continue to get dropouts. Stop the song and start again in the original room and I won't hear the dropouts. And no issues with Internet radio stations.
I uploaded another diagnostics test for Sonos to confirm: 8347153
BOOST is a 'simple' layer 2 bridge. It has no audio-related capabilities.
I uploaded another diagnostics test for Sonos to confirm: 8347153
They'd better advise. Bandwidth may be getting choked somewhere, resulting in mangled buffers. The matrix is at best a snapshot. Excessive physical layer errors could result from interference bursts. The history in /proc/ath_rincon/phyerr would be your guide.
Spoke with Sonos tech support who looked at the diagnostics upload. Some additional details for others:
* for anyone, like me, who starts out in Standard networking mode, then changed to Boost, you should first do a network reset (the button is hidden below the iOS keyboard) under Advanced Settings
* they recommend against connecting more than one BOOST device via Ethernet. In my case, the other two BOOST devices should be connected wirelessly to avoid "root bridge contention".
But my ultimate problem is that Sonos reportedly can't stream lossless audio to as many Sonos devices as I have in Party Mode. And, given bandwidth contention, the right paired speaker will drop off first.
Interestingly, the Sonos rep said that they do support multicast audio. Assuming that's the case, I don't understand why the number of Sonos speakers makes a difference then (as opposed to number and tonnage of streams on your SonosNet network).
* for anyone, like me, who starts out in Standard networking mode, then changed to Boost, you should first do a network reset (the button is hidden below the iOS keyboard) under Advanced Settings
* they recommend against connecting more than one BOOST device via Ethernet. In my case, the other two BOOST devices should be connected wirelessly to avoid "root bridge contention".
But my ultimate problem is that Sonos reportedly can't stream lossless audio to as many Sonos devices as I have in Party Mode. And, given bandwidth contention, the right paired speaker will drop off first.
Interestingly, the Sonos rep said that they do support multicast audio. Assuming that's the case, I don't understand why the number of Sonos speakers makes a difference then (as opposed to number and tonnage of streams on your SonosNet network).
Boy, that'll be the day!
"Root bridge contention"? You have an external root anyhow. None of the BOOSTs is root, and only one is secondary. Wireless, your BOOSTs may well be bypassed.
Interestingly, the Sonos rep said that they do support multicast audio.
Maybe they misunderstood, or they're thinking of some nuance of Direct Routing. When I've measured the Tx bandwidth from a wired coordinator it increases in proportion to the number of physical players in the group.
"Root bridge contention"? You have an external root anyhow. None of the BOOSTs is root, and only one is secondary. Wireless, your BOOSTs may well be bypassed.
I agree. Network performance is worse now that I have disconnected two the BOOST devices. I updated the article to show the new network matrix. According to the matrix the Ethernet connected BOOST is now serving all Sonos speakers. The other two Sonos BOOSTs that are now wireless appear to be disregarded. There are a number of yellow cells now that weren't there before. I don't understand why Sonos doesn't prefer a multi-channel hub & spoke model (multiple BOOSTs with Ethernet backhaul spread over channels 1, 6, and 11) versus a mesh network, particularly since that would free up more bandwidth for lossless audio.
See 8349487 (issues with lossless) and 8349557 (dropouts of right channel even with lossy audio in Party mode).
Maybe they misunderstood, or they're thinking of some nuance of Direct Routing. When I've measured the Tx bandwidth from a wired coordinator it increases in proportion to the number of physical players in the group.
Indeed, I can play lossless audio to a single speaker without issue. So, if Sonos supports audio multicast, then it's not clear to me why it can't play the same stream to all speakers without issue on SonosNet.
Multi-channel hub-spoke could only apply in a non-mesh configuration. The radios are single-channel. Substituting WiFi APs in place of the three wired BOOSTs and moving to Station/Standard Setup would permit simultaneous use of channels 1, 6 & 11. However what you gained on the swings you could potentially lose on the roundabouts, since AFAIK Direct Routing between grouped players can only work across WAPs if they share the same channel (as well as SSID/key obviously).
Since the WiFi environment is about to be steamrolled by mesh networks, I wonder how the question between standard network and Boost is going to play out. My concern is interference in having two mesh networks. My initial guess is that one would only want one mesh network in the house. A mesh network with a dedicated back haul frequency would seem to take performance out of the equation.
@Kenneth Goodwin
If you can, you want to use a BOOST network for your Sonos devices (either Ethernet connected Sonos player or BOOST device). And you want to dedicate your BOOST network to its own channel so it isn't overlapping with your home network. Assuming that is the case, then the type of home WiFi network that you have won't impact your SonosNet network.
If you can, you want to use a BOOST network for your Sonos devices (either Ethernet connected Sonos player or BOOST device). And you want to dedicate your BOOST network to its own channel so it isn't overlapping with your home network. Assuming that is the case, then the type of home WiFi network that you have won't impact your SonosNet network.
All,
I have significantly rewritten my article to address multicast flooding in addition to SonosNet and RF coverage. I believe that the multicast flooding is related to Google Cast devices and am currently working with Sonos Escalations on that.
I have significantly rewritten my article to address multicast flooding in addition to SonosNet and RF coverage. I believe that the multicast flooding is related to Google Cast devices and am currently working with Sonos Escalations on that.
Great feedback! I folded your information about color codes, multicast audio, and group coordinator into the article.
It seems odd that whatever specific room/speaker a user selects first (and thus group coordinator) makes a difference in overall Sonos system performance. I would expect this nomination to be handled automatically and, perhaps, even run within a wired BOOST.
However, I am seeing something that appears different: ALAC music plays without dropout for a few rooms. Play all rooms (14 speakers) in party mode and I consistently hear right channel dropout. Keep playing the audio and reducing the number of rooms, doesn't help. I can even go all the way back to a single room, where I started and will continue to get dropouts. Stop the song and start again in the original room and I won't hear the dropouts. And no issues with Internet radio stations.
I uploaded another diagnostics test for Sonos to confirm: 8347153
I am seeing the same behaviour with my 12 play:1 setup. and am digging around on this forum for some clues as to how to get things working better. This thread has been invaluable. Many Thanks.
Thank you @andrewlydon
Since I first published this post, I have upgraded all of my home networking gear to Ubiquiti Unifi. For a full write-up about Ubiquiti Unifi and Sonos best practices, please read here:
http://freetime.mikeconnelly.com/archives/6373
Unfortunately, there still seems something wrong with Sonos in networks that have a significant amount of multicast traffic (e.g. in my case, a number of Google Cast enabled devices). Two months on, I am still working with Sonos Escalations to resolve the issue so that I can play music in party mode. For my home network, three pairs of Sonos smart speakers seems to be the breaking point before I experience one or more speakers dropping out.
Since I first published this post, I have upgraded all of my home networking gear to Ubiquiti Unifi. For a full write-up about Ubiquiti Unifi and Sonos best practices, please read here:
http://freetime.mikeconnelly.com/archives/6373
Unfortunately, there still seems something wrong with Sonos in networks that have a significant amount of multicast traffic (e.g. in my case, a number of Google Cast enabled devices). Two months on, I am still working with Sonos Escalations to resolve the issue so that I can play music in party mode. For my home network, three pairs of Sonos smart speakers seems to be the breaking point before I experience one or more speakers dropping out.
Have you seen the Group Coordinator discussions here?
https://en.community.sonos.com/controllers-software-228995/sonos-group-coordinator-question-6803946
https://en.community.sonos.com/controllers-software-228995/sonos-group-coordinator-question-6803946
Is there a way to see exactly what paths SonosNet is using at any moment? And, even better, is there a way to see how much of the music traffic is unicast vs multicast? I am trying to work out what I need to change in order be able to play FLAC files in Party Mode. It is currently failing, presumably due to bandwidth restrictions somewhere on the 2.4GHz network. I am still a beginner with Sonos but hope to get up to speed soon.
Since I first published this post, I have upgraded all of my home networking gear to Ubiquiti Unifi.
http://freetime.mikeconnelly.com/archives/6373
Is there a way to see exactly what paths SonosNet is using at any moment? And, even better, is there a way to see how much of the music traffic is unicast vs multicast? I am trying to work out what I need to change in order be able to play FLAC files in Party Mode. It is currently failing, presumably due to bandwidth restrictions somewhere on the 2.4GHz network. I am still a beginner with Sonos but hope to get up to speed soon.
The STP topology can be determined from the Network Matrix at http://IP_of_player:1400/support/review. I don't believe there's anything user-accessible which displays the active 'short-cuts' in use by Direct Routing in a group situation. It might be possible to assemble the picture from the 'direct=X' information against each wireless tunnel in the individual STP pages (/usr/sbin/brctl showstp br0). A 'direct' parameter of 1 or 3 means that tunnel is a candidate for Direct Routing.
Since Sonos have access to the full diagnostic, and presumably will have written all kinds of tools to parse this into useful views of the system, you're probably better off consulting them.
@andrewlydon
FLAC and ALAC (Apple Lossless) compress PCM/WAV to between 33% (classical) to 70% (rock) of the original CD source. With some notable exceptions, most internet streams top out at 128kbps. So, VBR lossless tends to be 4-7x the bandwidth of a 128kbps CBR stream.
The way that Sonos engineers have explained it to me, if you use SonosNet (BOOST mode, not Standard WiFi), then both the control protocol and the music will be multicast. If you use Standard WiFi, then the control will be multicast but the music will be unicast. This leads to scaling issues quickly in Standard mode, not least because Standard mode is 2.4Ghz only (a software, not hardware limitation in the Sonos One). Which is why you should prefer BOOST mode if you have multiple Sonos speakers.
They caveated that even with BOOST mode exclusively using 3 BOOST devices, Party Mode--in my case, 16 Sonos Ones--may be susceptible to drop outs due to coordination issues where there is a significant amount of other non-Sonos multicast traffic.
In my case, I have 2 dozen Google Cast enabled devices (TVs, AVRs, STBs, soundbars, Google Homes, etc.) Even after (1) Google resolved the "mobile device wake up" multicast "storm" bug, (2) I properly configured STP and (3) I enabled IGMP snooping on my UniFi network, Sonos believes that there is still too much multicast traffic for Sonos to play a single lossless audio stream to more than 3 pairs of Sonos Ones. Similarly, I cannot play an internet radio station in Party Mode, though it does scale to more Sonos One speakers than lossless audio. (No, this doesn't make sense to me if the audio is supposedly multicast in BOOST mode).
Hopefully, we will get to the bottom of this issue soon.
FLAC and ALAC (Apple Lossless) compress PCM/WAV to between 33% (classical) to 70% (rock) of the original CD source. With some notable exceptions, most internet streams top out at 128kbps. So, VBR lossless tends to be 4-7x the bandwidth of a 128kbps CBR stream.
The way that Sonos engineers have explained it to me, if you use SonosNet (BOOST mode, not Standard WiFi), then both the control protocol and the music will be multicast. If you use Standard WiFi, then the control will be multicast but the music will be unicast. This leads to scaling issues quickly in Standard mode, not least because Standard mode is 2.4Ghz only (a software, not hardware limitation in the Sonos One). Which is why you should prefer BOOST mode if you have multiple Sonos speakers.
They caveated that even with BOOST mode exclusively using 3 BOOST devices, Party Mode--in my case, 16 Sonos Ones--may be susceptible to drop outs due to coordination issues where there is a significant amount of other non-Sonos multicast traffic.
In my case, I have 2 dozen Google Cast enabled devices (TVs, AVRs, STBs, soundbars, Google Homes, etc.) Even after (1) Google resolved the "mobile device wake up" multicast "storm" bug, (2) I properly configured STP and (3) I enabled IGMP snooping on my UniFi network, Sonos believes that there is still too much multicast traffic for Sonos to play a single lossless audio stream to more than 3 pairs of Sonos Ones. Similarly, I cannot play an internet radio station in Party Mode, though it does scale to more Sonos One speakers than lossless audio. (No, this doesn't make sense to me if the audio is supposedly multicast in BOOST mode).
Hopefully, we will get to the bottom of this issue soon.
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.