A novel way to massively improve the SQ of computer audio streaming

[romaz]

If this works for you then great.

 

I would like to point out the misconception that a "direct" connection is somehow fundamentally different than using a switch or router. Routers themselves very frequently run a variant of Linux e.g. pfSense so all this really depends on is the quality of the hardware : many many people here obsess over details of PC hardware / disc drives/ different types of power supplies yet hardly any thought regarding different types of switches and routers.

 

An FMC is itself a simple switch ... and replacing the power supply and/or clocks can be considered for not a great deal of $$

 

So all else equal: no big deal. However it's taking a lot or work to get Windows to work in bridged mode: duh -- it's not designed to be a router/switch.

 

 

Sent from my iPhone using Tapatalk

I don't know if this "direct" connection is fundamentally different or not. All I know is with the very same router in place (in fact, no other variables have been altered), this "direct" connection in my system sounds considerably better.

 

I agree with you completely about the importance of the network and how everything in the network path seems to make a difference including not just the switch or router but also the internet modem. In my earlier post on this thread, I had indicated my desire for someone to build an audiophile-class modem/router/switch that incorporated low noise linear regulators, an audiophile-class clock and integrated optical isolators. Having experienced first hand how my modem/router was transformed when connected to my Paul Hynes SR7, no one has to convince me that this matters. With that said, the FMCs that we use also incorporate switching regulators which are not ideal. The TP-Link FMCs use switching regulators that downconvert the 5-9V that it receives to 3.3V. Ideally, I would like to bypass these noisy regulators and just feed the FMCs 3.3V directly from my LPS-1. Whether this results in better SQ, I'm not sure but that will be a project I will want to tackle at some point.

[romaz]

I am fairly comfortable messing around in basic network stuff, but don't have a lot of theoretical knowledge, especially in things like bridging adapters.

 

In my setup, I don't connect to the internet at all. It is currently a very simple configuration:

 

Zotac Z-Box LMS Server <-> Netgear AV switch <-> FMC <-> FMC <-> R-Pi (Squeezelite)

D-Link Wi-Fi Router setup as access point--^

 

I do have DHCP running on the D-Link to make it easier to insert new gear. I can easily turn it off.

 

Both the LMS Server and the R-Pi(s) have static IP addresses..

 

All are of course on the same subnet

 

If I connected the Zotac LMS Server directly to the R-Pi, should it work? If it does, I could of course control it by plugging a monitor, keyboard, and mouse into the Zotac.

 

Or do I need to add a 2nd ethernet adapter and bridge them? This I could do with a USB-Ethernet adapter, the Zotac is a 2-smartphone sized micro with no internal expansion ports.

 

If I can get it to work, longer term I'd like to go back to controlling via the LMS Server webpage either on a smartphone or a laptop that I connect to the D-Link's WiFi network.

 

The Zotac has built-in Wifi that I could turn back on to connect to the D-Link. Or I can connect a USB WiFi dongle to the Zotac for the same purpose.

 

But I thought I saw one couldn't bridge ethernet to WiFi.

 

Those with more knowledge of more advanced networking techniques, what do you think?

 

Greg in Mississipppi

I couldn't get wifi and ethernet to bridge in Mac OSX. In fact, after repeatedly failing, I did some reading and OSX expressly states this isn't possible. It might be possible in other OS's. Give it a try.

 

If both server and NAA have static IPs and are on the same subnet, you could be good to go. Whether you would need a crossover ethernet cable, I'm not sure. As you've stated, you'd be forced to control your NAA only from the computer your NAA is directly attached to and it may not be what you would want to do long term but at least you could hear how this direct connection sounds in your system.

[romaz]

Hi romaz:

Any chance you or someone else here can write me a dummies guide to making this work in OSX using the Manage Virtual Interfaces GUI window of the Networks Prefs Panel?

 

My goal is to use my microRendu in NAA mode via a single Ethernet cable from my iMac (where both HQ Player Desktop and my music files reside).

 

On the iMac I have an Apple Thunderbolt>Ethernet dongle (EN1 I think) for connection of the computer to my Cisco switch. From the built-in Ethernet port (EN0), I always have a 25-ft. BJC/Belden Cat6a cable running to my tweaked Mac mini, and that is the cable I want to instead put into the microRendu.

I don't use wifi at all. It is turned off.

 

So I need to know both what settings (IP, etc.) to alter for the built-in network, and then what to set in the Manage Virtual Interfaces window as far as bridging. And then is there anything I need to change in the microRendu? (Of course I know to tell HQP to use NAA as Back-end once it is set up.)

To date with the microRendu, I've only been able to use it as NAA when running it through my switch, with my DSL modem/router providing DHCP service.

 

Many thanks to anyone that can walk me through this. I just don't have the time to do my own trial and error these days---and John is coming for a visit next week so I'd like to have this working so we do other fun things.

 

Best,

--Alex C.

Sure, Alex.

 

System Preferences > Network > Manage Virtual Interfaces

 

Once in Manage Virtual Interfaces, click the "+" sign at the bottom left of the screen and select "New Bridge".

 

Give this new bridge whatever name you'd like. You should have a selection of interfaces listed for you in the box to include in this bridge. Specifically, you should see your native ethernet port called "Ethernet". You should also see listed your other ethernet port that belongs to your Thunderbolt device. In my instance, this is called "Broadcom NetXtreme Gigabit Ethernet Controller" although yours may be called something else. Regardless, check the two ethernet ports you want included in this bridge and click "Done."

 

This should take you back to the previous screen. In the left window pane, you should see this new bridge you've created and its status. If it states "Connected", you may already be good to go. Regardless, make sure this new bridge is highlighted. Then look at the right window pane to see the specifics of this bridge. In the Configure IPv4 line, feel free to keep the "Using DHCP" option. This works just fine for me although if you feel compelled to create a static IP, you can do so. If you decide to use the "Using DHCP" option, then you don't have to bother hitting the "Advanced" tab, just hit "Apply" if you're not already connected.

 

You should be good to go. On my Mac Mini, it was that simple although you may require a reboot.

 

To verify that your microRendu is online, open up Safari and type in "www.sonicorbiter.com" and the control screen should come up. From there, make sure the HQPlayer NAA option is selected and you should be ready to play music. Let me know if this doesn't work.

[romaz]

Thanks Romaz.

Have been following this thread since first post.

Just took 3 minutes to set this up on my MacMini.

 

Will report later on SQ.

Yes, please do!

[romaz]

Gents,

Chris briefly posted a description of the upcoming Roon 1.3, then pulled it.

I remember seeing something to the effect that an "improved RAA" (my words) was in the pipe.

That will be another interesting development to follow, plus all the other improvements, especially the built-in HQP-type capabilities.

Did they partner with Jussi? I dont see them replicating all the work....

Yes, I think 1.3 will be big for many of us. I'm sure you've seen John Darko's piece already:

 

Coming over the hill: the monstrous Roon 1.3 | DAR__KO

 

Hopefully, this improved RAA sounds just as good with this direct connection.

[romaz]

Quite the elaborate thread (skipped through it) - interesting read.

 

Has anyone compared the "bridged LAN-connection" to the recent, just updated wifi-feature through the sMS-200? Wouldn't a succesful wifi-connection here potentially "deflate" some of the efforts sought on the LAN-connectivity?

It could sound better and I'm open to anything that sounds better. Based on what May Park of SOtM has said, however, this wifi connectivity with the sMS-200 sounds no better than the standard connection, it is supposed to be more of a convenience option. When I connected a USB drive directly to the SOtM's USB port, it still didn't sound as good as this direct connection between my Mac Mini and the sMS-200.

[romaz]

romaz, we have been directly connecting devices for some time now. I have a how to guide for OS X where WiFi is used for internet and the Mac is used for playback directly connected to a microRendu. If you develop a similar guide for Windows (including how to undo it) that would be great. This "should" avoid having to use two ports on one motherboard which hardly anyone has.

 

Have you loaded at this thread:

http://www.computeraudiophile.com/f26-sonore-sponsored/sonicorbiter-direct-connection-mac-pc-beta-28606/

 

Thank you for this link, Jesus! I was not aware of it and so it appears you've blazed this trail already with your Sonicorbiter. Interestingly, it appears the intent was for convenience and no one really made any comments on whether they thought this connection resulted in sonic superiority (perhaps no one considered it might sound better this way and so no one did any critical comparisons although to my ears, the improvement isn't subtle).

 

On my Mac Mini (El Capitan), I've succeeded in making a connection with a single ethernet port (obviously no bridging involved) but this was only useful for playback of music from local storage. For those who wish to stream from a NAS or from Tidal, then you're stuck.

 

I will be testing Windows for myself soon but it appears this direct connection is a bit trickier with Windows for many who have tried it thus far. Perhaps the best person to develop a guide for Windows would be Andrew Gillis. As he is the developer of the microRendu's OS, I would think he would be in the best position to develop a "no fuss" direct path to the microRendu. He has already suggested he can configure a sonicTransporter that could be used for direct connection straight out of the box and so this could represent a convenient option for many.

[romaz]

Stupid question perhaps:

 

Does the way the wired LAN-connection is configured matter when a LAN-isolator just before the network player is used? As in, will the LAN-signal on the side of the network player, after the LAN-isolator "treatment," benefit from a less polluted signal (in the wake of a bridged LAN-connection) prior to the LAN-isolator treatment?

I have the same LAN-isolator that you do, the SOtM iSO-CAT6 and I am using it between my Mac Mini and SOtM-200. While the difference I can detect with it in the chain is small (meaning I could easily live without it), it seems to improve it and so I have kept it in. I have yet to try my optical isolators (FMCs) between Mac Mini and sMS-200 to see if they would make a difference.

[romaz]

I have this running now. I relocated my 2008 Mac Pro that was running Roon Server into the listening room, connected the WW Startlight leading to the microRendu to port 1, and a generic Ethernet that leads to my FMC to port 2.

 

Weird thing is that after rebooting my Mac Pro could not be found on the network nor could the Mac Pro be found. The microRendu could be found though. I ultimately deleted the bridge and created a new one. All good now. This is just a short term solution to asses this method of connection, so I'll just leave well enough alone.

 

Now for the listening...

 

 

Sent from my iPad using Computer Audiophile

I'm looking forward to your impressions, Kenny. It seems those with an sMS-200 have been unanimous that there is an improvement. I'd like to hear more from microRendu users. I should get my microRendu back in a couple of weeks.

[romaz]

So, how would you rate the sonic difference between bridged LAN-connection with the iSO-CAT6 implemented (between the host computer here and network player), and that of the LAN-connection to either router-direct or via switch (still with the iSO-CAT6 implemented) - insofar you've tried out these configurations with the iSO-CAT6 implemented in every instance? I'm asking to find out whether the LAN-isolator may (or may not) diminish the difference, and if there's still a significant leap in sound quality to gain with bridged LAN-connection compared to the router/switch connection - with the iSO-CAT6 implemented in all cases.

In my system, regardless of where I put it, the iSO-CAT6 offers only very minor improvement. In fact, I could easily live without it and I question whether I would buy it again if I were to start over. In fact, I find my SOtM dCBL-CAT6 cable to offer much more impact which is why I have purchased another one to use in my direct connection. To my ears, it improves upon the BJC CAT6A and equals the performance of AudioQuest's much more expensive Diamond ethernet cable. SOtM's dCBL-CAT7 cable may be even better but I haven't tried it.

 

As far as the step change offered by this direct connection, in my system it almost rivals the improvement I am getting by connecting my sMS-200 (or microRendu) to my Paul Hynes SR7 which is saying a lot! I equate it more to a component upgrade than a minor tweak.

[romaz]

This is all very encouraging news. I think consensus is building that with at least the microRendu and the sMS-200, this direct connection results in a very worthwhile improvement in SQ and could spur further refinements in either these NAAs or in the music servers that connect to them. What remains troubling to me is that there remains no good explanation for why this should sound better. Clearly, this ethernet "frontier" with respect to audio is not well understood and there could be other discoveries around the corner. Would it help to have the ethernet output at the server reclocked? Would a slower 100Mbit speed sound better than 1000Mbit? Is it really the switch that is the problem. I will attempt to insert just my Paul Pang switch with TCXO clock in this "direct" path to see if it results in degradation or improvement.

[romaz]

Great! I just got in a couple of microRendu's and am going to give this a try this week. Is it safe to assume if one has a external thunderbolt HD and since thunderbolt daisy chains devices that the thunderbolt to Ethernet adapter will still work when daisied off a TB drive connected to mac? I'll find out I guess when I give this all a try. :-)

I don't see why a daisy chained Thunderbolt ethernet port wouldn't work. Please report back your findings.

[romaz]

Ok, another new revelation. The improvement I am hearing with this direct connection has nothing to do with bypassing the switch. At least, it has nothing to do with bypassing my Paul Pang 100Mbit switch with TCXO clock powered by my LPS-1. This switch has no noisy switching regulators. It is a 5V device as is the TXCO clock. Previously, as I had stated, it resulted in a small but meaningful difference in my system. I have now connected it in my "direct" path. It is flanked on both ends by a short SOtM black LAN cable. One cable connects directly to my Mac Mini and the other connects to my sMS-200. The improvement is now even larger with a bigger soundstage and even crisper details, easily heard with guitar plucks. While the improvement is not anywhere as huge as the direct connection, it is now MUCH more pronounced than when I had it connected before my Mac Mini! This switch is staying exactly where it's at.

 

I bought this switch used from another CA member for $99 and figured it was worth the risk. It's turning out to be a truly excellent buy. For those interested, you can buy one new from Paul Pang himself. It appears he no longer sells the slower 100Mbit switch that I have although it's unclear to me if one speed sounds any better than the other.

 

SHOPPING AREA: AUDIO GRADE SWITCH

 

Regarding the LPS-1, this is an excellent indication for this PSU. Compared to my switching 5V PSU, the improvement is very significant.

 

Tomorrow, I will see if my FMCs in this "direct" path results in any benefit.

[romaz]

Ooh this sounds really intriguing!

 

So are you saying this configuration is even better than the direct connection without the switch? And you attribute it to the SotM cables?

 

Trying to understand. Given my difficulties getting bridging to work, this may be a more viable option.

No, with this switch in the "direct" path, you will still need to bridge. I am seeing no way around this.

 

This switch now has no direct connection to the router like I had it before.

 

Before, it was internet modem/router > ethernet cable > FMC > optical cable > FMC > ethernet cable > Paul Pang switch > Thunderbolt ethernet port on Mac Mini > Mac Mini > native ethernet port on Mac Mini > 50 cm SOtM LAN cable > iSO-CAT6 isolator > 30 cm SOtM LAN cable > sMS-200.

 

What I have done is I have replaced the iSO-CAT6 isolator with the Paul Pang switch and connected the receiving FMC directly to my Mac Mini. Nothing else has changed. I had previously verified that the iSO-CAT6 was causing no harm although it didn't result in any major improvement either. As I have swapped out the iSO-CAT6 isolator and swapped in the Paul Pang switch, the improvement is quite notable.

 

Two lessons learned:

 

1. Having the switch in my network path (at least in my case) was never the problem and so bypassing this switch has nothing to do with the improvement heard with this direct connection.

 

2. It appears this direct pathway can be improved further.

[romaz]

These kind of discoveries make me wonder if an Ethernet based direct connection to a renderer have anything over a server only tweaked USB stream. Whereas the galvanically isolated power supplies (LPS-1) have changed the game for both.

This is a very good question. I have a lot of experience with various single box purpose-built music servers with USB streams including a TotalDac-d1 server, Aurender N100 with Ken Lau PSU, Aurender N10, Aurender W20, Lumin U1, CAD CAT, Auralic Aries with femto clock and LPS, various custom-built CAPS servers and more. Here is the thread I started on Head-Fi:

 

REVIEW: Comparison of 5 High End Digital Music Servers - Aurender N10, CAD CAT server, TotalDac d1-Server, Auralic Aries, Audiophile Vortex Box

 

None of those single box units ever sounded anywhere close to what I am hearing now. That is how special both the microRendu and sMS-200 are, imo. This direct connection has much to do with it also but so does my Paul Hynes SR7 which I just got recently. I am finding these things to be the key pieces so far.

[romaz]

But in this scenario, why use the switch at all? Is anything else connected to it besides the Mini and the SMS-200? Even with the clean LPS on the Pang switch, it's still one more active device in the chain?

I did this as an exercise to see if bypassing the switch was responsible for the improvement I was hearing with this "direct" connection as many have assumed. I wasn't necessarily expecting to hear an improvement, I was just wanting to see if placing the switch directly in the path would result in deterioration which would suggest that yes, the switch is the guilty party and so it was a pleasant surprise to hear that SQ actually improved.

 

Another question I wanted to answer for myself was whether relocking the signal just prior to sending it to the sMS-200 would make a difference as my Mac Mini has no special clock of its own. Both the microRendu and sMS-200 have their own audiophile-class clocks but if I made their jobs easier by sending them a freshly reclocked signal, would it make a difference? The concept is identical to the concept behind the USB Regen, that if you feed a component a high integrity signal, then SQ potentially improves. That may be what is going on here.

 

I did try using just a short 30 cm LAN cable between Mac Mini and sMS-200. Initially, I wondered whether such a short LAN cable might result in worse SQ due to reflections but that doesn't appear to be the case. It sounds great with just this short cable. As I stated, adding the iSO-CAT6 isolator to this chain resulted in no harm and maybe a touch of improvement and so I kept it in.

 

Tomorrow, I will try the exact configuration you have requested.

[romaz]

But then you didn't have the LPS-1 to make the comparison fair?? The LPS-1 has transformed my USB stream.

I have high respect for my LPS-1 and it is my highest value PSU by far. For sure, it is playing a role in why my current setup sounds better but presently, it is only powering my switch and one of my FMCs. Compared to my SR7, the SR7 is better although to be fair to the LPS-1, the SR7 is better than PSUs I have from Kenneth Lau, Paul Pang, HDPlex and Teradak by an even wider margin. It is expensive but it is the finest PSU I have ever experienced and it has very significantly transformed my sMS-200, mR, Mac Mini and internet modem/router. I now have another Paul Hynes SR7 on order but this time, it will be Paul's very best as the outputs will have double regulation and between 150-160dB of line rejection.

 

If I had to prioritize the key reasons for why my current setup sounds so good (better than anything I have tried before), as I stated, I believe it is largely due to the sMS-200/mR, this direct connection and the Paul Hynes SR (in no particular order) but as always, everything matters.

[romaz]

romaz,

did you try the SGM 2015 as well?

Matt

 

No, I am very intrigued by that device although I am in a unique situation from many. My Chord DAVE DAC upsamples much higher than HQPlayer does with its FPGA. With Chord's new M-scaler (which is built into the new Blu Mk2) which I have on order, combined with my DAVE, I will soon be up to 1 million TAPS. HQPlayer will never come close to this with even the most powerful computer. Nonetheless, I bought HQPlayer and having tried HQPlayer with upsampling either to DXD or DSD512, DAVE by itself sounds better.

[romaz]

Which server did you set-up with HQPlayer and which DAC did you feed with DSD512?

Thanks

 

Matt

HP Z820 workstation with dual 8-core Xeons and Nvidia Quadro K5000. Unfortunately, a very noisy machine. Like I said, I have a Chord DAVE DAC which can handle DSD512 and PCM up to 768kHz but internally, it upsamples much higher, all the way to 2048FS.

[romaz]

hi,

did someone test that direct link vs an optical network ? which one performs the best ?

I guess the optical network is better but that's just a guess...

thanks

Rgds

Ok, I swapped out my Paul Pang switch with TCXO clock from the "direct" connection and swapped in my TP-Link 100Mbit FMCs with the receiving FMC powered by my LPS-1. My Mac Mini is now directly connected to my internet modem/router and the Paul Pang switch is completely out of the picture.

 

Here is what my chain looks like: Modem/router > CAT6 cable > Mac Mini > 50 cm SOtM LAN cable > FMC (powered by generic 9V battery supply) > short optical cable > FMC (powered by LPS-1) > 30 cm SOtM LAN cable > sMS-200

 

The impact of having the FMCs in the "direct" connection is about the same as the SOtM iSO-CAT6 LAN isolator meaning there is an improvement with regards to a perceived slighter blacker background and maybe slightly better dynamic contrasts compared to just having the LAN cable in the direct connection but this difference is small. It is small enough that when I had my wife blind test me, I couldn't tell the difference. Out of 4 tries, I guessed right 1/4 times.

 

The difference with the having the Paul Pang switch in the direct path is noticeably better with crisper clearer details and a bigger soundstage. It is as if another thin veil has been removed. With blind testing, on 4/4 tries, it was easy to know when the switch was in the chain and so the switch stays.

 

This is an interesting and somewhat surprising finding for me. At least in my system, this leakage current/ground loop issue just isn't that big of a deal.

[romaz]

So do you think if PP would create a simple inline (or one output switch) with the TCXO clock that it would be a seller? And did you try FMCs with the PP switch (i.e after the switch)?

I believe so, Ted. I think it would be an equally ripe opportunity for Sonore, SOtM, and especially Uptone Audio (Ethernet Regen) since such a device would be right in their wheelhouse. It would also be another great application for the LPS-1. Of interest, Paul Pang used to sell a NIC with a TCXO clock which he shelved for some reason late last year. If I were to build a server today, I would ask him to resurrect that NIC and maybe even have him consider using his better OCXO.

 

I have not yet tried the combination of PP switch and FMCs and I will give it a try but I have my doubts the FMCs will add anything based on what I heard last night.

[romaz]

Larry, isn't the SotM ISO-cat6 he tested a passive?

Yes, that's right. The SOtM iSO-CAT6 is a passive isolation transformer.

[romaz]

Ted, you are right, but looking at a review of the SOTM solution it looks much more complex then the simple hospital grade isolation transformers such as the EMOsystems solution often discussed here. I had the latter in mind.

There are other solutions like the EMO, Acoustic Revive RLI-1 and GISO but I have never tried those devices.

[romaz]

I have gone through this thread's posts and it looks like no one has successfully bridged 2 Ethernet nics on windows 10 and got the main pc to connect to a microRendu to play HQP.

Do I have that right, and if not tell me who I need to contact?

I will be trying this on a Windows 10 Pro device later this week. I'll let you know how I fare.

[romaz]

Isnt the crux of the issue that the microrendu does not allow you to manually specify IP and subnet? If this was allowed, then at least we could have a direct setup with 2 separate NICs like how Jplay recommends it which I have successfully implemented in the past.

 

I think someone mentioned having to run a DHCP server on the windows machine to make this work in bridge mode?

 

Total aside, I have no idea why microrendu wont allow manual configuration of the network parameters. I asked this on the Sonore forum. Lets see what Jesus mentioned.

I mentioned this to Jesus on this thread last week and this was his response:

 

"I disagree, this is a complex server side / network based solution that can work with the microRendu...These specific solutions are best sorted out by the interested parties."

 

http://www.computeraudiophile.com/f10-music-servers/novel-way-massively-improve-sq-sms-200-and-microrendu-31110/index7.html#post620664