Ethernet streaming and occasional white noise on Windows

[Confused]

No, still on the old fashioned 7.1.0. (only just got SAM, so I am spending some time getting used to that, I don't like changing two things at a time) I sent my email query to Devialet late on Thursday, so I am hopeful of something back next week

[Georgethesixth]

It is possible to 'force' white noise on my system.

Background : MC19 with AIR Kernelstreaming (no Wasapi). This setting remains untouched.

Music is playing and sounds ok.
Go to Devialet Settings : turn off Bitperfect
Sound stops, but continues in MC.
Reboot MC server
Play music

Result : White noise ...

Reboot MC-server and Devialet : problem solved.

I tested it 5 times. Results are repeated. Is there anyone who can formulate conclusions or recommendations ? (other than 'Don't do that stupid !')
Can somebody confirm this 'pattern' ?

[Riddo]

Just tried the following. Full PC reboot, Jriver WASAPI only, 16/44.1 files only. No messing with playlist, pausing or anything else, just full album run-throughs. Ran ok for 48 minutes, then switched to almost silence, with just very scratchy sounding quiet music in the background. That's a new one! My quest to find a clear pattern in all this is, ummmmm, getting interesting. My confused tag started as a bit of a joke but it seems to be getting more and more appropriate with respect to this particular issue.
Perhaps my previous habit of switching between iTunes and Jriver has actually increased run time to white noise some how? This will make sense one day. (possibly in the very distant future)

Did you upgrade to 7.1.1 ? I'm silently hoping that might solve the problem. What I don't understand is the deafening silence of Devialet. If they would acknowledge the problem and suggest a workaround (USB ?), I could live with that.

Well, USB is my workaround after spending some time replacing a bass driver damaged by the "white noise" problem. I'll wait until the general consensus agrees that Devialet have solved the problem before I try ethernet again.

[pete86]

Devialet has always refused to acknowledge issues with AIR, from drop-outs in the 5.0 firmware (see the long thread on the what's best forum) to this white noise problem we have currently. I'm also using USB until I'm convinced it's resolved.

[AlexS]

@Confused, I'm puzzled as to why the issue seems to have worsened for you. With my precautions (restarting PC on a regular basis, and never putting the AIR application to sleep), I've avoided the white noise for over a month now (with 16/44 music).

@Georgethesixth, although your test case is a bit contrived (turning off BitPerfect midway), perhaps, it has a common root cause with the issue we experience naturally. If you can recreate the issue at will, Devialet can do a collaboration session with you. Thierry offered it to me, but my problem is sporadic.
I wonder what's in 7.1.1 vs 7.1.0. 7.1.1 must contain just minor bug fixes (Maybe, "Improved temperature with SAM")

---

I found the answer to my question about the new features of 7.1.1. There is nothing related to streaming:
New in Firmware 7.1.1

Software changes

USB bi-dir
Improve latence music between master and companion
Possibility to use the trigger out of the companion
SAM setting menu (in the Tone menu) only if SAM is ON
Improved stability of power supply
Improved temperature with SAM

---

One more thing: when you submit a request for help online, and receive a verification email, at least, I have to click on the link in the email TWICE, until I receive the following message: "Thanks for verifying your request." Without that, the request never reaches Devialet, which may explain why some of you never get any response back.

[Georgethesixth]

Always happy to assist troubleshooting. I'm free next week. As from August 2/8 I'm back to work so I'm not available during business hours.

[Confused]

Just for the record, I have always received responses to my emails to Devialet, it might take a day or two sometimes but respond they always do.

Agreed it is strange that the pattern of occurrence has changed, but I guess this can be a problem with occasional issues that vary, in that it inherently takes a while to build up a clear picture of occurrence patterns.

@Georgethesixth's slightly leftfield post has actually given me an idea. It might be interesting to turn off the Bit Perfect in AIR for a while, just to see if this changes or eliminates the problem. This might help to hone in on what is happening.

Listened to some vinyl and a good old fashioned cd earlier. No white noise, bliss!

[AlexS]

After having been able to avoid this probelm (the "dreaded white noise", as some people call it) for over a month, I experienced it yesterday. In a sense, I was not surprised by it, because my usage pattern that day was inviting the problem to happen: I kept the unit on for 5 hours, with AIR constantly connected to it, and listened to music for over 3 hours (with breaks, but without turning either Devialet or AIR off - testing SAM with Cremona M, BTW, which was a disappointment for me, but that's a rather different topic).
So, this confirms our main pattern: just using Ethernet streaming for a prolonged period of time. The only question is what resets AIR, taking it back to a clean state? Restarting it? Very likely. Disabling and re-enabling? Dubious, since @Confused observed that after the problem happens, disabling and re-enabling AIR only fixes streaming for about an hour. On the other hand, I think I was able listen to Ethernet streaming without issues for over 3 hours during a single day, keeping AIR running, but periodically turning Devialet off (which disables streaming in AIR). Or is turning Devialet itself off has a special effect on AIR, different from just disabling streaming in it?
Quite a few questions. But the pattern remains incontestable and consistent. I'd like to make some PC memory measurments, to see if AIR consumes more memory after streaming music, and if so, what causes this memory to be released. After that I'll be reporting my findings to Thierry.

---

Of course, there is a theoretical possibility that the problem lies with the Windows WASAPI mode itself. But then why would AIR restarts help to prevent it?
It may also be interesting to try a long streaming session with WASAPI disabled in AIR (and the Preferred Format set to match that of the music material). @Confused, I think you were going to try that at some point.

[Georgethesixth]

I've ordered a network-tap that allows to capture all network traffic between Devialet and other devices. Once it is delivered, I'll try to analyse what is going on, both on the PC part and the network part.

I did do a network trace on the switch but of course one can only see the broadcast traffic. The Devialet is a cheeky bugger Almost 600 broadcast in 30 seconds on port 45454: 50 % on the wifi interface and 50 % on the Ethernet interface. No problem for a home network, but the day someone connects a Devialet to a large corporate VLAN, they might be in for a surprise.

Suggestion to Devialet : why not ask for IP addresses of possible AIR-clients in the Configurator tool ? You can use broadcasting as the standard setup and unicasting as the advanced 'setup'. It is even a more secure . At least you need to have a configured IP-address before being able to eg. cycle the power or increase the output volume beyond repair.

[AlexS]

@Georgethesixth, I thought Devialet was using unicasting for streaming. AIR does know the exact IP address of the Devialet (and shows it, when it's connected). So, why would AIR need to broadcast messages? Well, that's a topic for a different thread (perhaps, one titled "Know your Devialet"
I wonder what useful info you'd be able to gather with a network tap, even if you catch it at white noise act. How would you know if AIR sends "noise" packets to the Devialet, or if the noise originates in the Devialet itself (needless to say, I suspect the former). Perhaps, the packets would contain only 1's (no 0's) - that's what white noise is, is it not?
What I'm after is to check if there is some memory misuse on AIR's part. Would one see used RAM grow as one continues playing music? The procedure should be simple:
After playing music, open "Task Manager" and note the used memory. Shut down AIR, and see how much memory was released. Take these measurements for music sessions of various lengths. If longer sessions consistently result in AIR's increased memory consumption, there is a memory leak, which should be reported to Devialet.
Yet I'm not up to lengthy experiments. It's in our best interest to communicate our concerns to Devialet as soon as possible (and the more emails, the better), to start them actively working on this (and we've collectively identified the pattern: a prolonged streaming session). For Devialet to work on this issue would - arguably - be a better investment of effort than SAMifying the speakers that simply don't benefit from SAM.