Devialet Chat

Thanks for trying it, @daniel.avasilichioaei. Shame it didn't give us anything more interesting.

If the idea is to get closer to RAAT, I wood recommend using the options -t and -l 60k.

-t is to use TCP like RaaT instead of using ICMP
-l 60k is to try to force the usage of the same kind of jumbo frames as RAAT seems to use.

Jean-Marie

@Jean-Marie I did think about using a TCP "ping" but wasn't sure there was a sensible choice for the TCP destination port (i.e. one that would be open and able to handle this kind of abuse!).

By the way although there are what look like 60 KB frames in Daniel's WireShark traces, I guess that's probably not a true reflection of what's actually on the wire; more likely it's an artefact of the "long send offload" option used by the Ethernet NIC. Hard to be sure though.

(21-May-2019, 16:03)mdconnelly Wrote: [ -> ]When I first started experiencing this problem (when Roon 1.6 was released and Devialet released their latest OS/firmware update), I tried everything I could think of since there were some users that were not experiencing the problem.   As time went on, many of those that claimed it worked fine realized that they were, in fact, operating at 10/100 mpbs network speed.   But some, as described in this thread and in Roon forums, had it working at gigabit speeds without issue.      

Here's a list of the major changes I methodically tried:

-- different PCs running RoonServer or Roon Core
-- different ethernet runs (including different cabling - 6, 6a, 7 - and dramatically different lengths) from PC to router to Devialet
-- different Devialet configs including eliminating all but ethernet input and no wifi
-- two different routers (Pace 5268ac, Dlink)
-- numerous different router settings

The problem was always readily reproduced.   It wasn't until I changed the network speed did I find a workaround.   Initially, I just changed the network speed to 10/100mbps on the PC running RoonServer.   That worked!    I then set the PC back to gigabit and placed a cheap 10/100mpbs switch in front of the Devialet.   That also worked.    Happy it did but clueless as to how/why.

What I have not tried is having two ethernet runs from my RoonServer PC such that one goes to the router and the other establishes a subnet to the Devialet as others have successfully implemented.   To do so in my setup would be far more complex than just using the 10/100 switch.

My frustration with both Devialet and Roon is that I have to believe they know what the issue is but are remaining silent about it.   It's never surprising that Devialet is silent, but in my experience, when Roon is quiet, it's because they know it's not their problem to fix.   Of course, this did pass Roon's QA testing so they are partially responsible for the headaches.

As with all things Devialet, it will get fixed eventually.  I suspect it will also require us to pester them periodically.   I'd love to better understand the issue, but fortunately, the lower network speed works perfectly (as does using the AIR protocol) so I'm willing to wait patiently (yet read the forums frequently hoping someone shares some further insight ;-)

Thanks for that. It looks like we can scratch my idea that it has something to do with having both ethernet and wifi inputs enabled.

(21-May-2019, 12:45)thumb5 Wrote: [ -> ]@David A - another possibility is that there are differences between the hardware, such as different revisions of the Ethernet PHY or even the processor on the Core Infinity board.  Thinking aloud (again) these might result in subtle timing changes that tip the balance between Gigabit Ethernet working reliably and not.

@thumb5

That's a possibility but I have to wonder about how likely it is. They started upgrading Expert Pro's in late November/early December 2017 and they had to have bought enough boards to handle the number of upgrades expected (ie the number of Expert Pros they had sold up to then) plus expected new production for a reasonable period. It's unlikely they were buying boards in small runs, they would have made a bulk purchase to cover upgrades plus production for a period simply in order to keep their costs down and all boards from that bulk purchase are likely to be identical. It's unlikely revisions or changes to the hardware would have been made during the run.

There's a fair chance that nearly all of us on Devialet Chat with CI machines are using boards that came from the initial run that Devialet took delivery of, almost certainly all of us who had our amps upgraded with the new board are. I think the odds are that there are no differences in the hardware, at least for most of us. If there are differences then I'd say the people with the different hardware would be those who had their amps upgraded closer to February this year when the Roon Ready update came out, or people who bought new amps close to that time and that those of us who have the "original" hardware versions are those of us who had our amps upgraded with the new board or who bought new amps in the months after the upgrade process started.

If we asked the question, I suspect we'd find that there would be no relationship between when people with or without the problem had their amps upgraded or bought their amp and whether or not they have the problem.

Fair enough. I don't have any insight into Devialet's procurement practices which may indeed be as you suggested. I do know from experience of working with embedded hardware and writing device drivers for interfaces like Ethernet that the slightest difference in hardware revision can sometimes cause substantial difference in behaviour. All I would say is that I am a little surprised that you seem to be ruling out some potential causes without any evidence.

I'm not ruling anything out, I did say it was a possibility. What I said was that I wonder about how likely that is as a cause. As for ruling something out without evidence, there's also no evidence of the kind of problem you suggested as far as I'm aware. If, as you suggest, the problem is related to a particular hardware version that has been changed on the CI board at some time then you could try opening a poll thread and asking people when they got their CI upgrade/bought an amp with the CI board, and also whether they get the problem or not (limit responses to those using RAAT and a wired gigabit connection so things don't get messed up by people who are not at risk of the problem don't respond and distort the data). If you're right then we should be able to identify that those people who do have the problem tended to get their amps/upgrades either before or after a certain date but if there is no date correlation with the problem and the incidence is relatively even over the whole period since the boards started reaching users then that would tend to indicate that a change in hardware on the CI board isn't related and perhaps even that there haven't been any hardware revisions on the board.

The big problem with a poll like that is whether or not you get a representative sample. There are 4,213 members on this board at the moment yet there are only 200 posts on this thread and many of the people who have posted have posted multiple times. There are probably less than 200 members out of the total membership who have posted on this thread and not all of those people have the problem. 200 people is roughly 5% of the membership here and we don't know what percentage of members have amps with the CI board and it's a poll relying on self selection when it comes to responses. That's not a good recipe for getting a statistically reliable poll result but it's the best I think we could do here.

Our problem is that we have a problem but that's all we have. We have a set of symptoms that some people get and some people don't. We know it's associated with RAAT, the ethernet input on our Devialet's, and a gigabit connection. Not everyone using the Roon Ready input with a gigabit ethernet connection gets it but we know little else. @daniel.avasilichioaei has provided some interesting data and you've provided some analysis but with only one source of data we can't be sure whether only people who have the problem would show results like those that Daniel has demonstrated or whether people without the problem would see similar results if they did the same tests. That means we can't be sure Daniel's test results are related to the problem.

In other words we've got a pretty big lack of hard information when it comes to anything apart from how the problem manifests itself. Devialet and/or Roon may know more but at present they aren't saying anything and I can think of 3 reasons for that:

1- they haven't been able to reliably duplicate the problem. We don't know the actual incidence of the problem, how many people with CI boards using gigabit ethernet connections get the problem and how many don't. All we see are the reports here and on the Roon forums and people posting on these 2 sites are probably a minority of Devialet owners. What's happening with the people who don't post on the forum here and at Roon? If the problem is only occurring for a minority of owners then Devialet and Roon may actually be having issues replicating the problem reliably.

2- they can replicate the problem but they're still working on identifying a problem or developing a fix and they don't want to announce anything until they have something to announce.

3- they know what the problem is but there isn't a fix and they're trying to hide the facts.

There are some people here who will probably make arguments for each of those possibilities with justification because the lack of comment from both Devialet and Roon for more than 3 months doesn't inspire confidence but we've got no evidence to indicate which one of those possibilities is the most likely.

Basically, we know how the problem manifests itself but apart from that all we have is suppositions and guess work and the only people who could give us any reliable information are Devialet and Roon and they're not talking.

So I'm not ruling anything out but I'm not too strong on generating suggestions about causes that we have no way of checking out. Yes, one of those suggestions may be right but the only way we're going to know is if Roon or Devialet tell us what the problem is.

In the meantime I think the best we can do in practical terms is to gather comments about what people who aren't getting the problem are doing in their system, things like what are they running Roon on, what their network connections are, and what their config files are like, and also gathering that kind of info from those who get the problem. Then people with the problem could try duplicating the setup of those without it and see if that solves the issue and those without the problem could try duplicating the setup of those with the problem and see if they get the problem, and if they all report their results here we can start to develop a list of things that are common to those with the problem, those without the problem, and which changes either trigger or solve the problem. There's a lot of work involved in organising that sort of data and the more people providing data the better because given the number of different amp models involved, the number of different computers/servers being used to run Roon, and the number of different routers and switches likely to be in use it's going to be hard to get enough people with each combination of things in use to allow us to make educated guesses in every case.

I'm not trying to be negative but I once had to design and run a research project as part of a university course and actually having to do that in a way which stands up reasonably well both statistically and scientifically teaches you that doing that kind of project isn't as easy as you thought it would be before you started doing it.

I am one of those having recurrent problems with a Devialet Expert 220 Pro and streaming via Roon. I decided to use a 100Mbps switch to connect the Devialet, as a Gigabit connection is a non-starter. It works fine for a few days this way and then the problem recurs. The short-term fix is to unplug the Devialet and pull the SD card so it reverts to a factory configuration; then I push the the SD card back in so it loads my preferred configuration. Obviously, this is not a long-term solution.

I have conveyed the issue to Devialet. Today, Karen responded with these two questions:

1) Does the Devialet Expert app on your phone still work? (i.e., control of the amp and visual feedback of the screen). The answer is no. The Devialet app, whether on my phone or iPad, just doesn’t connect.
2) the device continues to collect an IP address, this being visible from the info menu on the Ethernet or WiFi interface? The answer is no. The IP address (Ethernet and WiFi) on the information menu comes and goes, as if the device were getting a signal and then not getting a signal.

So, consider this an update for those of you having similar problems. Not sure what Devialet might be doing after months of this experience…

I have got the error in Roon too. And I am using 5GHz WiFi. I rebooted Roon Rock to fix the issue.

(03-Jun-2019, 17:42)maxijazz Wrote: [ -> ]I have got the error in Roon too. And I am using 5GHz WiFi. I rebooted Roon Rock to fix the issue.

I'll revert to 5GHz WiFi.  That works and isn't as wonky as the 100Mbps Ethernet connection.