Conducted and Radiated Noise + Other Mechanisms?
One of the reasons I've held off on making a final suggestion is because some of the results make sense, while others go against the common held theory about ethernet and that's it a 100% perfect non-realtime asynchronous not susceptible to any changes provided the buffer is filled ahead of time. For our purposes, it means that it shouldn't make any difference in the final SQ if we tweak or alter the network setup, provided the buffer is filled and plays without any dropouts.
So before revisiting that point, the first thing I wanted to do in my experiments was to address other variables and if possible eliminate them (which I have)
. The two of them that I've been able to test are conducted and radiated noise. This is a test that I've been doing for almost 3 years every few days. I've done this test a good 100+ times on separate days just to see if the result will ever differ. It never has.Conducted noise
is noise that you might introduce inside the BDP by plugging in an ethernet cable. This noise can either be picked up by the environment or from the leakage of power supplies. We are also assuming that these cables are ungrounded (so UTP or shielded cables with floating ground) in order to keep the ground plane isolated. John Swenson does have data and measurements on leakage and how to fix it. As for the cable picking up noise, Cat6 is immune to 30 Mhz. The floating shield can give even further protection. Additionally, the levels of noise are not supposed to be anywhere this bad, so a standard UTP cable should work fine.Radiated noise
is noise that is coming out of the ethernet cables and affecting other gear nearby or other cables.
Before looking into anything else, the first test you want to do at home in your setup is to see if conducted or radiated noise is a problem in your setup. There's a simple test for this.Test:
First you have to make sure that there is only 1 ethernet cable in the vicinity of your gear and coming into the rack. Keep all the other ethernet cables a good 6+ feet away just to be sure. Use MPD and play music off a directly attached flash drive. Plug and unplug the ethernet cable from the BDP and see if you notice any difference in sound.Methodology:
I tried best and worst case scenarios of this test. I have a 10 and 18 feet AES cable so I can keep the BDP-1 in the rack with all the gear or pull it far away from the rest of the gear as a control. I also have a Torus isolation transformer for the audio gear. I also controlled for whether the power supplies for the router and switches were plugged in or not, or actively transmitting or not. In some cases I kept the switches and their power supplies 10+ feet away from the rig. In other cases, I put all that networking gear directly on top of my DAC and the digital and analog cables in order to try and force some kind of noise. As far as cables go, I've used 30+ ranging from Cat 5e, 6, 6a S/STP floating shield, belden bonded at various lengths (1 feet to 60+ feet). All Fluke tested from a data center where I get them for free.
I've used a total of 9 devices
with which I made a direct ethernet connection to the BDP-1:
1) Mid 2012 Retina Macbook Pro (Thunderbolt-ethernet Adapter + USB-ethernet adapter...direct ethernet port - no router/switch in middle)
2) Late 2009 iMac (direct ethernet port - no router/switch in middle)
3) Mid 2010 Macbook Pro (direct ethernet port - no router/switch in middle)
4) Hitron CGN3CSMR (modem/router)
5) Linksys E3000 (router)
5) Trendnet TEW-432BRP (Fast ethernet router)
6) D-Link DGS-1005 (Gigabit switch - both with stock power and Teradak LPS)
7) D-Link DGS-108 (Gigabit switch - both with stock power and Teradak LPS)
D-Link DES-105 (Fast ethernet switch - both with stock power and Teradak LPS)
9) D-Link Powerline AV500 adaptersResult:
Plugging and unplugging the ethernet cable or their power supplies made absolutely zero difference
to the sound regardless of where the BDP-1 or the rest of the networking gear was placed, or what ethernet cable was used. Again, this was with playback from a USB flash drive and the ethernet cable was only used as a controlling feature to view Manic Moose. I kept the Manic Moose page open to make sure the ethernet cable was transmitting the entire time when plugged in.Good news:
If you are playing music off a USB device and using an ethernet cable for controlling the BDP, then conducted noise shouldn't matter with the BDP's. You can still test for radiated noise by doing the plugging and unplugging test to see if any other gear is susceptible to this noise. In my case it had zero difference even in the worst conditions. Essentially, you can use any ethernet cable or router/switch for controlling the BDP. The power supply of the networking gear also makes no difference to the BDPs so need to spend money on a LPS here.
Ethernet is working as expected here! The ethernet isolation transformers are working perfectly. No noise goes through. Ground plane is kept isolated. Hooray!
This is what 99% of MPD users should go for, at least if they have a BDP-1. Put music on local USB drive and control with ethernet. It very well may also apply for BDP-2 and BDP-3, but I cannot be certain as I don't have them for testing (which I really wish I did because there is one experiment in particular that I cannot run with the BDP-1 but can with the BDP-2 and BDP-3 due to its gigabit port)Bad news:
When these same ethernet devices are used to send data for Roon (RAAT) or MPD (NAS drives), then you start getting subtle differences between each device or whether a stock vs. linear power supply is used or even the ethernet cable used (I found something unusual here in another direction that almost no one has bothered considering or controlling for...doesn't concern shielding but rather length...will go into it later on. Belden has info on this as well regarding impedance mismatch and return loss. Nothing fancy required here, however some consideration is needed IMO. Very cheap solution)
When used as a control, all these ethernet devices behaved like you would expect with noise and ground plane isolation. However, when used for sending audio files or PCM the differences start emerging. In all of these setups, there were never any dropouts and so no bits were dropped. Not once!Why does this happen?:
A lot will jump to tell me that it's all in my head. In my defence, this is something I've been doing for over 7+ months and in my desktop setup, the way the computer desk and rack is placed with the networking gear hidden on the ground, I never know which device is being used. I'll have a bunch of ethernet cables coming from below and into the wooden rack (closed back) and they are all made of the same material and colour. It's hard to explain here, but essentially there is no way of knowing which ethernet cable is plugged into which device. I can reroute the stream and make it come from any of the devices in a second.
Because I was able to test and remove conducted and radiated noise as an explanation, I have 2 theories as of right now. I'm not sure if they are related at all or if it's only one that's happening. People seem to REALLY DISLIKE
even considering them because ethernet is non realtime and asynchronous without any clock associated to it and isolated from any noise. I think John Swenson is looking into this perhaps, but he said he doesn't want to say anything at the moment and will wait at least half a month to a year. To be fair, those guys might be right, and this is something that shouldn't be happening or should be controlled for...yet it happens with the BDP-1.
I'm not an EE or an IT person so I cannot confirm anything or say with 100% confidence that this is why. Neither do I have any gear for measuring this. So the way I approach these tests is by trying to control for as much as possible and eliminate variables along the way. I've also done quite a bit of testing with regards to jitter so I know what to listen for. I'm not sure what the mechanism is between the differences, but it ultimately results in variable pattern of jitter depending of the networking device, PS, and cable used.
Try this and find out how good your ears are: http://www.cranesong.com/jitter_1.htmlWill you hear this difference?:
Yes, no, maybe so?
Since I haven't tested the BDP-2 or BDP-3, this might not be an issue on those two devices. Or if your DAC is really exceptional and immune to all jitter, then it also might not matter. There are a few tests you can do still do. See if you hear any differences between FLAC (compression level 5 or higher) and WAV when playing music from MPD off a USB drive. Try the Crane Song jitter test and see if you can hear extreme and intentional cases of jitter in your setup. It's good to know if your ears/brain or setup are capable of identifying jitter.
More to follow...