Maximizing Ethernet SQ (Roon and/or MPD)

0 Members and 1 Guest are viewing this topic. Read 3800 times.

zoom25

  • Full Member
  • Posts: 983
Maximizing Ethernet SQ (Roon and/or MPD)
« on: 30 Nov 2017, 06:14 am »
This is based on the trials and experimentation in the past 3-4 months with various cables, power supplies, powerline adapters, network switches, and routers.

Here is the album with pictures and descriptions: https://imgur.com/a/2XIAV

I will be adding my thoughts regarding what worked, what didn't work, what sounded worst/bad/good/excellent, what made the most difference, and what made no difference. I will also offer thoughts on how the sound differs vs. local storage (flash drives and portable hard drives and hubs). I would love to hear from others on this subject.

Useful links and posts:

https://www.bicsi.org/pdf/conferences/winter/2009/presentations/Mythbusting%20Takes%20on%20Shielded%20Cabling%20-%20Herb%20Congdon%20and%20Brian%20Davis.pdf

https://www.siemon.com/us/standards/Screened_and_Shielded_Guide_7_Antenna_Myth.asp

https://www.siemon.com/us/standards/Screened_and_Shielded_Guide_9_Why_Use_Fully_Shielded_Cabling.asp

http://www.ti.com/lit/an/snla107a/snla107a.pdf

https://community.roonlabs.com/t/validity-of-audiophile-ethernet-cables/30251/25
« Last Edit: 30 Nov 2017, 06:02 pm by zoom25 »


zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #2 on: 30 Nov 2017, 04:04 pm »
https://en.wikipedia.org/wiki/Observer-expectancy_effect

 :lol:

https://arstechnica.com/staff/2015/02/to-the-audiophile-this-10000-ethernet-cable-apparently-makes-sense/

 :lol: x 100

Thanks CM! Excellent feedback as usual.  :thumb:

It's funny that you didn't wait for me to even post my results and already assumed or hinted at what would be my results. You do this quite often. Hmmmm. Anyways, I wouldn't have it any other way. It would feel odd otherwise. Routine is a good thing. Maybe give this a read:

https://en.wikipedia.org/wiki/Psychological_projection  :D

CanadianMaestro

  • Restricted
  • Posts: 1760
  • Skepticism is the engine of progress
    • Hearing Everything That Nothing Can Measure
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #3 on: 30 Nov 2017, 04:08 pm »


Ha ha, you're a good sport, I give you that much  :thumb:

You had stated:
"I would love to hear from others on this subject."

With or without waiting for results.
So I chimed in.

Enjoy the musik.


pssst: The Ethernet cable has no effect on the SQ of a BDP-1. Simply put, the EN port and circuit inside BDP are electrically isolated from the rest of the digital circuitry, so....
(If they're not isolated, then shame on Bryston's engies... :duh:)


zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #4 on: 30 Nov 2017, 05:28 pm »
ETHERNET CABLES

First up, by popular choice ( :lol:), is ethernet cables.

Almost all of the Cat 5e, 6, and 6a were UTP. The only S/STP was the Cat 6a.

The setup was as follows: iMac w/Roon <----ethernet-----> Switch <-----ethernet-----> BDP-1

I would use the spare Linksys E3000 router initially to get these unmanaged switch working and connected to each other. Once the BDP-1 was recognized on the Mac, I would unplug the router and its power supply. No other devices were present on the switch. Only 2 ethernet cables at a time. No other switching power supplies besides the bare essential for the audio networking.

All the audio gear on Torus. The stock power supplies for the router and switch on separate strip outside the Torus. I kept the distance between all the gear and path of the cables constant.


The Cat 6a SSTP came with graphs and readouts as well. Each cable had the propagation delay (ns), delay skew (ns), resistance, and other performance characteristics measured individually.

Sound:

Used HD 800 and all Amphion monitors/amps/cables monitor setup.

Cat 5e and 6 UTP:
-All the Cat 5e UTP sounded the same.
- All the Cat 6 and 6a UTP sounded the same.
- The only tiny difference I felt at times between 5e UTP and 6 UTP was that the cat 6 had a tiny bit better focus. Still, nothing compared to other tweaks.
- Sound tonality and presentation was the exact same.

Cat6a S/STP:
- These came with metal plugs on both ends.
- The BDP-1 has shielded ports.
- With the DGS-1005 (unshielded) and DGS-108 (shielded), I could try connections where the grounds would be attached on both ends or only one end.
- In a home environment, I could not ground the DGS-108 switch
- Eventually, I even took out the metal shields on both ends, which allows me to verify whether what I was hearing was due to any differences between the switches and their power supplies.

- I would NOT recommend having the grounds connected on both ends when both devices have shielded ports.
- In general, I would not recommend having the ground connected on the BDP-1 end.
- I eventually had the ground disconnected on both ends.

- I did find that the Cat 6a S/STP had the blackest background among the bunch and this stayed regardless of whether the grounds were connected or not on one or both sides.
- When you have one or both the grounds connected, there is a specific characteristic change to the sound.
- I felt this character was really exemplified when grounds were connected on both ends.
- The transients were oversmoothed and this was noticeable on material with a lot of drums and kicks.
- I compared my 5e UTP, 6 and 6a SSTP with the ground lifted to the ground connected 6a cables. The former all presented drums and kicks with clean rise and decay. They all sounded correct this way.
- With the ground connected, the drums didn't have the proper energy and bite.
- There were other changes I noticed and to some they may or may not like it.
- I just knew that the sound wasn't correct and so ditched the shields on both end. Better this way for me.

- With the shield disconnected on both end on both pair of Cat6a S/STP, this is where I found the sound was the best.
- Clean transients and correct sound like the rest of the UTP, but with better clarity and focus to the image and a black background.
- There can be a number of reasons for why there might be a difference.
- The Cat 6a S/STP not only eliminates pair to pair crosstalk, but is also better suited EMI/RFI wise in an audio environment in rejecting or radiating noise.
- This can reduce interactions with other nearby cables and gear.

Two useful links:

https://www.siemon.com/us/standards/Screened_and_Shielded_Guide_9_Why_Use_Fully_Shielded_Cabling.asp

https://www.siemon.com/us/standards/Screened_and_Shielded_Guide_7_Antenna_Myth.asp


TL; DR:

- If going with UTP, go with Cat 6 and forget about it.
- No sound difference between any of the Cat 5e and Cat 6 UTP cables.
- For best sound, go with a S/STP design, but make sure that the grounds are not connected.
- Introducing the ground can change the sound in a very specific manner (softened transients, over smoothed sound)
- Shield and screened cables can potentially offer a very quiet background, and better clarity. You can read up on the phase performance of both the HD 800 and Amphion monitors.
- YMMV based on your environment and setup.
« Last Edit: 4 Feb 2018, 05:42 pm by zoom25 »

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #5 on: 30 Nov 2017, 05:45 pm »
Link to the Cat 6a S/STP: http://www.infinitecables.com/cat5e-cat6-cat6a/cat6a-molded-shielded-patch-cables/cat6a-sstp-black/

The Cat5e and 6 was all the stuff I had gotten from a data centre for free. Everything was generic, cheap, and measured well.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #6 on: 30 Nov 2017, 05:53 pm »

Ha ha, you're a good sport, I give you that much  :thumb:

You had stated:
"I would love to hear from others on this subject."

With or without waiting for results.
So I chimed in.

Enjoy the musik.


pssst: The Ethernet cable has no effect on the SQ of a BDP-1. Simply put, the EN port and circuit inside BDP are electrically isolated from the rest of the digital circuitry, so....
(If they're not isolated, then shame on Bryston's engies... :duh:)

With the cables and power supplies, I'm not even going to attempt to say that this particular thing improved or changed the sound for this particular reason. On my end, I tried to keep things to a minimum and change one thing at a time. However, because of the interactions between all the gear and components, there is simply no way I can say anything with full conviction of "this is exactly why X happens". It would be arrogant to do so.

There is already quite a bit of discussion on other forums from developers and manufacturers that are far more talented about what might be happening in an isolated case or a system.

At best, I can only offer suggestions based on the repeated patterns that I observed over the last 4 months while changing things one at a time.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #7 on: 30 Nov 2017, 06:00 pm »
Powerline Adapters

The D-Link AV500.

These can deliver consistent performance in terms of speed. I still found my wireless faster.

These devices are simply way too noisy in how much noise they put out on the mains as well as radiate. When connected to the BDP-1 directly, the sound was at its worst. Dynamically flat and unengaging. BDP-1 sounded worse than my iMac as a source.

Additionally, on my MacBook Pro running off of battery (no power bar connected), with headphones plugged into the headphone jack, I could hear the buzzing noise interference. Again, no electrical connection in this case.

NOT RECOMMENDED. There is simply too much baggage to deal with this.

Krutsch

  • Full Member
  • Posts: 568
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #8 on: 30 Nov 2017, 06:16 pm »
Quote
I did notice that when putting in the various Cat5e of 5-10 feet, it could sometimes take anywhere up to 5-10 seconds. The Cat 6 was faster. The Cat 6a SSTP was near instantaneous.

Let me guess: you tested these cables in the same order? So, with the Cat5e, the associated MAC address and IP address are not cached in the router and Roon server, so look-ups took a little more time. When you tested the follow-on cables, physical and logical network addresses are already cached and look-ups are instantaneous.

Trust me: at 5-10 feet, there would be no measurable difference in network transmission latency or throughput.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #9 on: 30 Nov 2017, 06:25 pm »
Power supplies for the switches:

Relevant reading from Texas Instrument: http://www.ti.com/lit/an/snla107a/snla107a.pdf

Switches used: D-Link DGS-108 and DGS-1005.

Biggest problem with the supplied power supplies on these switches is that they are not grounded.

These stock switching power supplies radiate a lot of EMI which I confirmed with an AM radio. You could even hear their noise pattern change as the load varied. So, that alone can impact other gear nearby at some level.

I tried the 5V Apple charger as well which worked fine. Then I added the Jitterbug to the Apple charger. I did not like it there. It cleaned up some things, but added this fatigue that would hurt my ears. Same thing I experienced when I used the Jitterbug with the portable hard drives connected to BDP-1 directly without a hub. The Jitterbug doesn't seem to like being in the middle of things being actively powered.

When I went back to the stock switching D-Link, everything worked fine and sounded right. The piercing effect was gone.

Then came the Teradak U9 linear power supply which is grounded. This was by far the biggest improvement in comparison to cable changes or adding Jitterbugs or other generic switching power supplies. I use the 5V usb output which is reported to be cleaner than its DC output. There are noise measurements scattered in the wild.

John Swenson has some good info on this subject as well. Grounded power supplies seems to be the one thing that every one out there agrees on.

The Teradak U9 is not only grounded but is also better at not putting noise back into the mains or putting airborne noise.

I haven't yet tried plugging the Jitterbug into the Teradak, even though it's pictured there in the album as a thing I may try in the future.

The Teradak brought a change that I think almost anyone could identify. It was not subtle. Among all the tweaks listed on this thread and in the past regarding local playback with various flash/portable drives and USB hub, this would be my #1.

TL;DR: Teradak U9 LPS for the network switch brings the most improvement. First thing I would recommend.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #10 on: 30 Nov 2017, 06:35 pm »
Let me guess: you tested these cables in the same order? So, with the Cat5e, the associated MAC address and IP address are not cached in the router and Roon server, so look-ups took a little more time. When you tested the follow-on cables, physical and logical network addresses are already cached and look-ups are instantaneous.

Trust me: at 5-10 feet, there would be no measurable difference in network transmission latency or throughput.

I've tried them in all kinds of order just by sheer chance over the months. Sometimes, the switches and router were left powered on during the swapping. Other times, the switches was also powered off while swapping. Regardless of the situation, I still found the Cat6 SSTP to lock on almost instantly.

You are most probably correct and it could be exactly that. For my purpose, there was zero difference after the initial setup.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #11 on: 30 Nov 2017, 06:48 pm »
I think that's about all I have to say on this topic. On the other older thread, I've already exhausted my views on local attached playback. The only thing left is the Wifi Dongle and how wireless compares. I may consider that if I move my setup somewhere else.

Roon through this setup sounds considerably different to how I experienced Roon last year or even the start of the summer this year. I don't find whatsoever like in the past that Roon is lacking to MPD local playback. I'm perfectly happy with the sound coming out of the Bryston and my headphones/speakers.

No more tweaking for awhile. Looking forward to hearing the Audeze LCD-2C again.  :thumb:

audioengr

Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #12 on: 30 Nov 2017, 10:12 pm »
I've done a bit of experimenting with routers, switches, cables and isolators.  Here is what I discovered:

1) if you are using wired Ethernet, the most important thing you can do is tie the DC common of your router or switch to earth ground.  This will significantly reduce the "leakage" noise that comes through the Ethernet transformer.  you can do this with a LPS or a switcher/wall-wart.  A bit of a kludge with a wall-wart though.

2) change the power supply on router or switch that is driving your Ethernet interface to a fast-responding LPS.  I have heard good things from other customers about the Sbooster products for this purpose, although they probably don't have #1.  You will need to run a separate wire.

3) add a good isolator in the Ethernet cable.  I have found that a .5m cable from the router to the isolator and then a 1-3 meter cable from isolator to the DAC input works good.  I like the EMO EN-70e. Inexpensive too.

4) Change your cables to a reputable CAT7 type.  I like the TNP brand from Amazon.  Cat6S is also an option.

Steve N.
Empirical Audio

jinjuku

  • Jr. Member
  • Posts: 8
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #13 on: 2 Dec 2017, 12:43 am »
1. Used UnShielded cabling. UTP is noise immune to 30 mega hertz.
2. CAT 5e is good for 10GBe, and is supported, by a number of manufactures for 45 meters. This should put into perspective how over kill a 15 year old Ethernet standard is for audio data rates (it's awesome :) )
3. A switch/router Achilles heal is the single ended power supply. High end switches and routers are often have a dedicated DC power port. Data centers often run DC to the entire rack line.

I took a 315 foot generic CAT5e at $0.30 a foot and a 3 foot Nordost Heimdall II at $233 a foot. When blinding and going NAS(Netgear DS)<>Switch (Cisco SG 200-8)<>Cary Audio DMS 500<>Amp<>Speakers. No one could hear the difference.

The best way to do this testing, and it's what I did, was to place switch ports into a LAG and hang multiple cables off the switch and swap them at will. I don't understand why a CAT6 would bring up the interface on the Bryston any quicker than any other compliant cable. That's never been my experience.

Me personally? I like wireless for a myriad of sound (pun intended) technical reasons.  For $80 I have dedicated AP on it's own channel and SSID the I routinely get 38MB/S throughput on. That's 38 time more bandwidth than needed for 24/192 PCM.

jinjuku

  • Jr. Member
  • Posts: 8
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #14 on: 2 Dec 2017, 12:48 am »
Let me guess: you tested these cables in the same order? So, with the Cat5e, the associated MAC address and IP address are not cached in the router and Roon server, so look-ups took a little more time. When you tested the follow-on cables, physical and logical network addresses are already cached and look-ups are instantaneous.

Trust me: at 5-10 feet, there would be no measurable difference in network transmission latency or throughput.

I have to agree with you there. Putting my multi homed machine (4 Intel NIC ports) into an LACP team and creating a dynamic LAG on the switch allowed cable swapping regardless of 5e or 6 or 6a to bring the interfaces backup in the same amount of time.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #15 on: 2 Dec 2017, 11:06 pm »
Yeah, I have no idea why Roon/BDP was being weird with responsive time with different cables. Not an issue. I'm leaving my cabling as is.

Before I was wondering if the Teradak LPS would be fine for audio uses as it was around $50. There are more expensive options out there, but I didn't want to spend anything serious on a power supply for a switch. So I was hoping for something that would get the job done or gets close. The experience was better than expected from the first moment. The soundstage depth, width, layering, and imaging just melted. Even on headphones, you can notice the difference from before. The other thing I noticed in comparison was that before it felt like there was this weird built up energy in the system. It just flows better now.

In comparison to MPD, there is less grain (hash?) which adds to the liquid aspect. I suspect it may be due to the USB drives (with or without hubs?) injecting noise. Perhaps, using MPD with a NAS might mitigate this. That actually might be a more fair way to compare MPD against Roon/RAAT, as both are getting data over ethernet. In that case, I suspect there would still be more happening inside the BDP-1 with MPD and processing vs. the incoming PCM from Roon.

Krutsch

  • Full Member
  • Posts: 568
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #16 on: 3 Dec 2017, 08:53 pm »

...

Perhaps, using MPD with a NAS might mitigate this. That actually might be a more fair way to compare MPD against Roon/RAAT, as both are getting data over ethernet. In that case, I suspect there would still be more happening inside the BDP-1 with MPD and processing vs. the incoming PCM from Roon.

It's actually the opposite, if I recall correctly from previous measurements (and on an older version of the RAAT app).

The Roon endpoint is constantly communicating back to the Roon Core/Server, in addition to processing PCM -> ALSA. That's why your network connection is so important. With NAS, you are mostly just pulling data down into the BDP, but with Roon there is traffic going both directions.

zoom25

  • Full Member
  • Posts: 983
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #17 on: 3 Dec 2017, 09:24 pm »
It's actually the opposite, if I recall correctly from previous measurements (and on an older version of the RAAT app).

The Roon endpoint is constantly communicating back to the Roon Core/Server, in addition to processing PCM -> ALSA. That's why your network connection is so important. With NAS, you are mostly just pulling data down into the BDP, but with Roon there is traffic going both directions.

Hmmm, interesting. The CPU currently lingers in at 11%. Although, I'm not sure how much that number translates to or what kind of noise each component (RAM, CPU) is generating, so it may be useless as a factor determining SQ. Best to probably listen.

jinjuku

  • Jr. Member
  • Posts: 8
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #18 on: 3 Dec 2017, 10:25 pm »
It's actually the opposite, if I recall correctly from previous measurements (and on an older version of the RAAT app).

The Roon endpoint is constantly communicating back to the Roon Core/Server, in addition to processing PCM -> ALSA. That's why your network connection is so important. With NAS, you are mostly just pulling data down into the BDP, but with Roon there is traffic going both directions.

I haven't spun up my protocol analyzer to map out Roons protocol but I'm with you on it's chatty nature.

If playing straight up through Tidal I can pull the Ethernet cable and the entire track plays. If I spin up Tidal via Roon soon as you pull the plug your done in a matter of seconds.

Not sure what's going on there but the behavior has me scratching my head at how sub-optimal it is. Suffice it to say I don't use Roon.

jinjuku

  • Jr. Member
  • Posts: 8
Re: Maximizing Ethernet SQ (Roon and/or MPD)
« Reply #19 on: 3 Dec 2017, 10:44 pm »
Hmmm, interesting. The CPU currently lingers in at 11%. Although, I'm not sure how much that number translates to or what kind of noise each component (RAM, CPU) is generating, so it may be useless as a factor determining SQ. Best to probably listen.

The computer is constantly making noise. I had an internal PCIe EMU 1212M Pro-Audio card. I took the balanced cable (TRS <> XLR) and on the XLR (input to the amp) I floated the shield drain conductor.

You could hear mouse movement, button click, keyboard, HDD access (SSD you could still hear but lower in volume and shorter in duration), paging operations. Soon as I soldered up the drain wire: Pristine audio.

It's the DAC's job to reject all of this. If what is going on in your computer is affecting your DAC you most likely have a defective DAC, or less likely defective cabling or computer.