I use a direct eth connection between my dev and mac mini. no router or switch inbetween. will soon replace the aq cinnamon with the diamond.
no idea how the direct eth connection compares to wifi
(25-May-2016, 00:54)johnsap Wrote: [ -> ]any glue why Air via wifi sounds differently than Air via Ethernet ?
From what I've read there is no difference in the digital signal that is received by the Devialet in either case. The likely explanation for the difference in sound is that different types and levels of 'noise' are introduced into the system by the two transport mechanisms.
With wi-fi the 'noise' emanates primarily from the wi-fi board within the Devialet and there's not much/nothing an end user can do to reduce it.
With ethernet cabling there are two types of 'noise' which are typically present - common mode (coming from the signal source in this case usually a router or switch often powered by a noisy SMPS), and airborne RFI. Some people use shielded ethernet cable, optical isolation or add ferrite chokes to ethernet cables to try to reduce the levels of noise. I tried ferrite chokes on my normal Cat 5e cable but noticed no changes in my system.
I was facing very painful cracklings with AIR 3.0beta and my D900.
After gathering a substantial amount of evidence, including recordings of the problems, I submitted a complaint to Devialet and had a quite constructive discussion with their support/dev today.
It appears that :
- the cracklings I heard are real and not the results of my imagination
- the network interface of the D900 I was suspecting to be defective is working ok
- the cracklings are produced when the Mac does not have enough resources to flush the network packets to the network interface at a fast enough pace
- they are working on an improvement but the subject is technically arduous
After this discussion, having a better understanding of what was happening, I decided to use the optimization scripts from CAD :
http://www.computeraudiodesign.com/osx-a...imization/
I have not noticed further problems so far, but I will continue testing to see if improvements are stable after deactivating all the services which I am not using anyway.
Just in case this helps anyone else...
(25-May-2016, 15:50)SwissBear Wrote: [ -> ]I was facing very painful cracklings with AIR 3.0beta and my D900.
After gathering a substantial amount of evidence, including recordings of the problems, I submitted a complaint to Devialet and had a quite constructive discussion with their support/dev today.
It appears that :- the cracklings I heard are real and not the results of my imagination
- the network interface of the D900 I was suspecting to be defective is working ok
- the cracklings are produced when the Mac does not have enough resources to flush the network packets to the network interface at a fast enough pace
- they are working on an improvement but the subject is technically arduous
After this discussion, having a better understanding of what was happening, I decided to use the optimization scripts from CAD : http://www.computeraudiodesign.com/osx-a...imization/
I will continue testing to see if any improvement occurred after deactivating all the services which I am not using anyway.
Just in case this helps anyone else...
Hello Bear
I am curious - what specification of Mac are you using ? Their explanation regarding "resources" sounds a bit suspect to me. I am using an iMac from 2008, with just 4Gb memory and a Core2 Duo processor, but with AIR3b the performance has been near-perfect with ETH and also WiFi, including HiRes, the same machine is also running Dirac and Roonbridge, so there is quite a bit going on inside this elderly machine . . .
Is the rest of your network behaving OK?
(25-May-2016, 16:44)Zappydev Wrote: [ -> ]I am curious - what specification of Mac are you using ? Their explanation regarding "resources" sounds a bit suspect to me. I am using an iMac from 2008, with just 4Gb memory and a Core2 Duo processor, but with AIR3b the performance has been near-perfect with ETH and also WiFi, including HiRes, the same machine is also running Dirac and Roonbridge, so there is quite a bit going on inside this elderly machine . . .
Is the rest of your network behaving OK?
Had the same thought. I can push DSD64 through a Raspberry Pi2 on Roon and I can't imagine AIR is hugely different resource-wise. I'd imagine any intel based Mac could handle it fine in isolation.
What format were the files out of interest?
How did Devialet test your D900's NIC out of interest - did they take log data, or did they remotely connect to it somehow and run some diagnostics?
(25-May-2016, 16:54)Hifi_swlon Wrote: [ -> ] (25-May-2016, 16:44)Zappydev Wrote: [ -> ]I am curious - what specification of Mac are you using ? Their explanation regarding "resources" sounds a bit suspect to me. I am using an iMac from 2008, with just 4Gb memory and a Core2 Duo processor, but with AIR3b the performance has been near-perfect with ETH and also WiFi, including HiRes, the same machine is also running Dirac and Roonbridge, so there is quite a bit going on inside this elderly machine . . .
Is the rest of your network behaving OK?
Had the same thought. I can push DSD64 through a Raspberry Pi2 on Roon and I can't imagine AIR is hugely different resource-wise. I'd imagine any intel based Mac could handle it fine in isolation.
What format were the files out of interest?
I think the limited resources which are at stake are the network resources. So the processor specs are of little help to judge whether they are right or not.
Yes, the network is very clean, with a dedicated Airport Extreme just under the network interface of the D900, a dedicated 2.4 GHz band for the Devialet, aso...
I was mainly streaming from Qobuz, upsampling on the Mac to 24/192 into the D900. This was convenient as this was allowing the exclusive, direct and integer modes on Audirvana to be used and not conflicting with any sampling rate changes.
I changed my mind and am not upsampling anymore.
Cracklings disappeared following my script execution. This is good news.
Will keep updated.
Devialet asked to conduct a certain number of checks and listened to the recordings of crackling I sent. They acknowledged that the problem was reproducible under certain circumstances I had detailled. So their explanations were most coherent and I do feel comfortable with what they say. My purpose here is just to share my positive experience
(25-May-2016, 00:54)johnsap Wrote: [ -> ]any glue why Air via wifi sounds differently than Air via Ethernet ?
No i dont,but just more meat on the bone in my setup
Air3 with wifi is just airy imo,have switched back and forward ,and ended up with a PP switch and from there audioqest diamond to Devialet
(25-May-2016, 09:03)PhilP Wrote: [ -> ] (25-May-2016, 00:54)johnsap Wrote: [ -> ]any glue why Air via wifi sounds differently than Air via Ethernet ?
With ethernet cabling there are two types of 'noise' which are typically present - common mode (coming from the signal source in this case usually a router or switch often powered by a noisy SMPS), and airborne RFI.
Precisely. There is indeed spec for CM noise and primarily depends on the Supply. even a DC variation in supply affects CM noise and what it does is; Example instead of reading a '1' sent from Driver, the Rx will read it as a '0' a simple example.
CM noise : Rise and Fall time of the differential signals cross at the center. Will it ? not really.
However the ETH protocol is packetized. there is CRC which actually checks the correct data sent. Each word is padded with pre-fix and post fix to the actual data sent out. However i m not sure how this is implemented in Audio system. In true (Financials) protocols if data is corrupted then the protocol sends back ACK to send back the data once again. Anyways not going in details good cable helps in making the EYE diagram appear good wide squarish waveform at the Rx end.
Hence I always forgive if AIR drops or pauses because I know its not the AIR protocol but our own system needs enhancements = Hence i add more delay in AIR shamelessly. ETH CAT 5,6,7 cables makes the signal appear very good at the RX. The Square waveform sent from Driver appears like a human EYE at the Rx. these good cables make it appear widened and better rise/fall time (faster rise/fall time, more squarish it is) at Rx.
(25-May-2016, 17:27)SwissBear Wrote: [ -> ]My purpose here is just to share my positive experience
I have to admit that my enthusiasm has been short lived. Upon the suggestion of a French forum participant, I launched, from a different machine, the port scan of the D900 interface, which started to crackle just as before
This just shows that the network board of the Devialet is not able to handle a reasonable level of traffic without serious trouble...
I will wait until things settle in a professional way from Devialet side and switch back to the Amethyst and its reliable interfaces