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

drjimwillie · February 19, 2019

Shot it down twice today in the use of Optane drives.....

I have read this thread from the beginning. I found it in July and I just finished it today!! I like to read from the beginning before I’ll post something. Also, it slows down time, because you’re trying to catch up to the present moment.

This is where I am today: I’m saving to build a NUC.

While I’ve been doing that I am simplifying the equipment I already have.

I needed a new NAS, so I bought the Fanless- Qnap HS 453dx NAS. It has a Celeron J4105 processor and 4GB of RAM. So basically it’s a NUC with 2 HDD Drive days and 2 M.2 SSD Drives.

I’m going to use it as a Music server. The advertising shows it running Roon. I will use it to run Roon and hold an HDD with my music. I was not planning to use an SSD because of the noise. But, with reports of Optane being quiet enough to not make a difference, I thought that I would try it for the performance boost it would give Roon. So I got a 32 GB Optane drive. After I installed it, I was told by Qnap that it does not support the octane drive.

Can you make a recommendation of SDD to replace the Obtane, please. M.2

Do you think Roon will run fine on the 6TB WD Red drive, so I do not need the SSD?

Also, I have a CAPS Pipeline server. I am getting it ready to be a server for my NUC. I am going to run the Pipeline in AL/Ram. For the moment it will be the endpoint, only running HQPe. I don’t want to Boot from the stick, so I planed to use an Optane drive. I had been told by supermicro tech-support that I can use Optane on this motherboard because it supports NVME. Today when I called to find which slot would be best, they told me the Optane would not work. No NVME support. After I told him I was using it as a server for music, that the Optane would replace a 1TB SSD to be removed because of “noise”, he became interested. He told me I would not be able to boot from the Optane. He told me that he would help me, possibly make it work. It feels a little risky to me.

I am wondering because I have the SOTM SATA filter on the SSD drive in the Pipeline currently, maybe I use 32GB SATA drive, on that SOTM filter?

What do you think I should do, please?

It is always a wild ride.

Thank you

Bill

drjimwillie · March 8, 2019

Risk=Reward

Humor is good between comrades 🙂

I would like to share my listening experience - I recently added the Qnap HS453dx, described to me as the audiophile NAS. Fanless, Cleron J4105, 4GB RAM. I could not get the Optane to work so I did not install an SSD.

I have a CAPS Pipeline as designed by Chris. I am running windows 10, Roon and HQ player desktop. Music internally on a SATA, EVO 1TB SSD. External HD Plex 300W ATX LPS.

I decided to move the music onto the HDD on the NAS. To reduce processes on the Pipeline, I decide to try Roon the NAS. Also because I was planning to run AL in RAM on the Pipeline, and I would not have to worry about the Roon database or updates. The Pipeline would be a player/endpoint only.

On the NAS, Roon is faster and more stable. I put on Abbey Road from Tidal, and started at Because, immediately I noticed that the music sounded better. The piano going into You never give me your money was sharper, and the part with the crickets, I heard detail in the background that I have never heard before, little chimes and things, going into the Sun king, John’s breath between verses was prominent. This was consistent with music from the HDD’s also, there is more space around the instruments.

Now for the terrifying part. Last night I ventured into the barren wilderness of AudioLinux and HQ Player embedded. Previously I had help from the great tech-support at supermicro for changes to my BIOS, to get the pipeline to boot from the USB drive and also in preparation to try to get an Optane drive to work. What may be of interest to others is that the pipeline is not NVME. Also Optane only comes in the M.2 form factor. To get it to work I have put it into a Startech adapter for the PCIe slot. This is the first time I have opened my pipeline since it was built for me by small green computer. I did my homework and needed the three guides I found to load in configure AL headless. I loaded HQPe and, could not get sound. Turns out I needed to choose the XMOS driver for my Holo Spring DAC. Listening sounded good but not dramatically better than It had previously. Then I loaded the OS into RAM. I would say my impression is that the sound was quieter. And then I realized I had not loaded the files for my Acourate convolution engine. By the time I had worked through this my wife was putting my child to bed. I had to listen and very low volume, so it’s hard to get a good impression. I wanted to listen to Abbey Road but I was having an issue with Tidal for the last day or so. I knew I was exceeding my timeline for working on my system but I could not stop because of my compulsion. I solved the Tidal issue by reloading it after I reset my password. But by this time my wife came down and yelled at me for spending so much time on my system. She reminded me of all of my responsibilities and suggested I could possibly take over some of hers. If she only knew The depths of my obsession.

She is working from home today so I am having the mental debate of, do I go in and listen to the system as it is to find out more about the subtleties or sneak in, pull out my computer, open it remove the SSD before I’m discovered, put it back together and then listen to it. But that would be breaking the rule of making more than one change at a time.

From first impression I would say that moving my music and Roon to the NAS had a greater effect then AL in RAM with HQPe.

I noticed that on the HQPe configuration page that the driver had reverted back to USB from XMOS and I’m afraid to touch it because this is what had enabled the system to work previously.

Oh, I forgot to mention that two weeks ago I had rotator cuff surgery so I am working with one arm 😛

It’s a great time to be an audiofool.

This is so much fun!

drjimwillie · March 9, 2019

Thanks for the encouragement. I tore it snowboarding. I was barely moving when it happened.

All of the learning and preparation to make this quantum shift in the evolution of my system has been a good distraction.

I’m still not sure what to make of the change in the sound. It is quieter. Like the “extra “energy is gone. I remember back when I got the audio Phileo, and it removed a lot of the Jitter. I remember thinking, this is what music sounds like.

The music is more of a cohesive whole.

I think it’s going to take a little bit to wrap my mind around what is going on.

It looks like tomorrow I’m going to have most of the day to myself do some more listening and enjoyment.

Then I can do some surgery myself and remove that SSD.

If I’m really brave I’ll get into the BIOS and see what I can find from those first screenshots in this thread, to remove. Although this is a supermicro board and the BIOS is completely different.

Like Count Floyd from second city television would say “very scary “

drjimwillie · March 9, 2019

Disaster

I was going to tell you how great everything sounded, because it did. I took notes for a great listening report.

The caps pipeline is in an HD Plex H5 case with the fancy switch on the front. This switch never really worked right. It would take multiple presses of the button to power on the Pipeline. But, I could always get it to power on . Recently Larry sent me a new switch. The day I was going to replace the switch there had been a power outage so the pipeline was off and I could not get it to power back on. ( I have an ultra isolation transformer The surgery was not an issue )When I opened case I found that one of the two metal posts had lost it’s nut and the nut was sitting right there. I replaced the switch. It did not power on. The guys at supermicro had me remove the battery and touch these two metal nubs to reset the CMOS. That worked. And the switch is working just fine. That was last week.

This morning the music sounded great. Power down pull out the pipeline, put on my rubber gloves and my grounding strap, opened it up and pulled out the SSD. Put it back in the rack. It would not power on. Pull it out of the rack and remove the only new item which was the Optane drive. Put it back in the rack. Those ATX LPS connectors are a pain with one hand, compounded by an awkward reach. Pulled it out of the rack remove the battery reset to CMOS, that not work either. I thought one of the great values of this motherboard is that it has an IPVM which would allow you to access the bios even when it was powered down well, when you plug in the ethernet cable into the special port the light on the switch does not light up because there is no power in the computer.

Right now all I have is anguish anger and beer in my stomach ( I waited to have this beer until I was done with my diagnostics )

My wife is going to come home and say “what did you do while I was gone“

Well honey nothing on the list you gave me.

I am out of ideas

drjimwillie · March 31, 2019

A lot of water under the bridge.

When last we saw our Hero, he had gone from the height of rapture having installed AL in RAM on his CAPS Pipeline, to plunging to the depths of despair trying to extract an extra ounce of sweetness by removing his SATA SSD and his server would not power on. The horror!

I removed the battery from the Supermicro motherboard, touched the two nubs to reset the CMOS and let it sit overnight. (Although the Pipeline had been running with the Optane drive in the PCIe adapter installed, I removed it, just in case.) There was a plastic ribbon cable touching the six prongs of a jumper thingie, so I moved that. The Pipeline fired up – kowabunga!? I am not sure what fixed it?

I found out that I have one month of warrantee remaining on my motherboard that covers labor. I have had trouble powering up in the past. I have to decide if I want to send my Pipeline across country to have Supermicro see if they can find the issue or keep listening to music, now that it is working. What would you do?

I am running AL/RAM but because I have Roon on my Qnap HS453dx fanless (audiophile) NAS, I do not have to worry about saving the database on the Pipeline, I did not reinstall the Optane & have returned it. The Pipeline is only running HQPe and is currently being used as a music player/endpoint. The NAS, with its Cereon CPU and 4GB RAM is the music server with my music files on HDD, (no SDD), Roon and Tidal delivered over Ethernet. Eventually I will have a NUC as my endpoint, bridged from the Pipeline player. This will be a new three-computer typology.

The extra energy. The excess is gone. There is more space around the instruments.

Previously, I may have been using the wrong convolution curve for Acourate in HQPe.

Now the music is even more three-dimensional and fully formed. The piano notes that I had heard as sharper before are now full, multi-fingered, chords. The sound feels like it reaches up to the ceiling.

Bass impact is visceral. Bass drums kick you in the chest and techno like Jain - Makeba thumps with force.

There is a volume boost. The potentiometer on my preamp maxes at 60 and I’ve been keeping it 10 lower than I usually do. Maybe a bit of harshness. I am happy again.

------------

The NAS with stock SMPS is in a closet, with the FIOS router, attached to a Netgear FS108 >30 ft Ghent JSSG Ethernet > Cisco 2960-SI > 3 ft Ghent JSSG360 > Pipeline.

The Qnap has two Ethernet ports, 10Gb & 1 Gb. There is a utility to create network or virtual switch, which will bridge about any interface you want, including wifi. This utility and a little help from the great Qnap tech-support, made the bridge easy. I thought I had to make a bridge from he router to the NAS and then a bridge from the NAS to the Pipeline. It was much simpler, one bridge. Connect the stock Ethernet cable from the 10Gb NIC to the FS108 (as a moat to the router); connect the 1Gb NIC with the 30’ Ghent, direct to the Pipeline, check two adapter boxes on the NAS utility & wa la – a bridge is built.

The volume boost is diminished, some of the openness is filled in with midrange, in a good way, I think, but the sound is a bit congested.

Rather than go direct to the pipeline, I connected the 30’ cable to the Cisco Switch > 3’ JSSG 360 to the Pipeline, the congestion is cleared up some. Definitely better. I guess the music data did not like the 30’ cable.

I moved the fanless NAS out of the closet and into the rack. FIOS Router > FS108 > 30’ cable > Cisco switch > Stock cable > NAS 10Gb NIC -bridge- NAS 1 Gb NIC >3’ Ghent JSSG 360> Pipeline.

I have the 300W HDPlex ATX LPS. The 12V out was powering the LPS1 for my Iso-Regen. I gave this 12V to the NAS and had to reenlist the SMPS to power the LPS1 (I had eliminated all SMPS), I plugged the SMPS into the homemade plugstrip, after the ultra-isolation transformer, not good, noise and a rhythmic spitting noise.

Moved the SMPS to the second dedicated circuit (I tied both dedicated grounds together, to reduce ground loops) Ahh much better.

Any harshness is gone. The detail of the AL/RAM is there, with the mids filled in and the bass rather than just hitting me in the chest affects my whole body. It is throwing a large sound field.

The speakers in my other room are Altec Lansing A7 Voice of the Theaters, my brother described those as throwing a sound field that enveloped you. This System has Snell C-IVs which were described in comparison to have Laser point precision. Now, Snells are throwing a sound field like the voice of the theaters. I just put on Paul Simon- Homeless, when I closed my eyes, the venue hall was bigger, than the fairly good sized room I’m sitting in.

Quiet Drums in the background of songs are super fast.

So from the beginning of this note, things were brilliant and stark, now things are fuller and some of the warmness is back. Listening to the volume at 01 is the full experience. You can see further in, at low levels, because the grunge is gone.

I want you to know that I write this because I admire you all. I am so smitten, that I wrote this twice. (chagrin face) I had almost finished this note last weekend on my phone, then I was called for family duty and was not able to retrieve the note. This is my second attempt.

I have been dying to ask three questions, please.

Austinpop-Sir, on the 400W HDplex LPS, did you bypass the exterior ATX interface and insert the cable through a port on the case and plug directly into the ATX interface on the motherboard? If not, can I do that and eliminate connectors and the second cable in the case and its noise? Is the interior a different connector form the exterior?

I now need another Ethernet cable. I think Alex suggested using a cable without a metal head, to prevent the possibility of magnetic coupling. I use the Ghent with the metal Metz head. What the best quality, non-metal Ethernet connector, please? Maybe Ghent can use this head with the JSSG 360?

About a month ago, I contacted the wonderful HDPlex and asked if they would install LT3045s on each rail of my 300W ATX LPS. It is the full sized unit with the same case as the 400W ATX LPS. He said that they couldn’t because it would be basically a rebuild and suggested I get a 400W. With all of the recent good words about the 400W I was considering it. I was thinking it had LT3045s in it. The smaller 400W HDPlex has them. I read in a comparison that the small 400W sounded good, but the 400W ATX LPS was “a different animal.” That sounded impressive to me.

HDPlex said that My 300W is very much like 400W. OK, I should be happy, but I am almost disappointed….. Does anyone have experience with the two of them?

I want to use regulators with the ATX LPS and the 300Ws other outputs.

I am not using my 19V rail. I would like to give it to the NAS. The NAS is 12V. I need someone to build me a Sigma-11 or a HPULN or something with a circuit that gives balanced sound. Can you make suggestions, please?

I was thinking of a Ghent ATX cable, with a box in the middle, with a regulator for each rail. This would connect directly to the motherboard ATX interface.

The 12V and adjustable outputs would need something also.

It would be much simpler to put the regulators in (or on top of) the case, directly off the rails.

I have been listening to music, alone, while writing this, wow.

Thank you for your patience.

Please enjoy your day,

William

drjimwillie · May 1, 2019

3 hours ago, Balázs said:

You are absolutely right and I didn't invent to design a new PSU. I just wanted to have a single source of good quality DC instead of many smaller PSU based on my existing HDPLEX 200. I really like what the 3045 did to the HDPLEX. It can be that it's not the 3045 that makes the difference but the HDPLEX represents a rather low quality power.

I have the 300 W HDPLEX ATX LPS. I am not using my 19 W output. I’m having a regulator built for me to step down and 19V to 12 V for my NAS.

Currently the HDPlex ATX powers my pipeline server/HQPE player. The 12V powers an LPS 1.2 for my iso-regen.

I would like to use regulators to clean up the power from my ATX but if I put LT3045s in line won’t it drop the voltage and reduce the compatibility with the computers requirements?

Also I would be using the adjustable output to power the LPS-1 for my SOTM USB card but the DC cord from the LPS-1 does not fit into the SOTM USB card. Has anyone else had an issue like this? I ordered a new DC cord from Ghent hopefully the tolerances will be better? SOTM confirms that the external power input is 5.5, 1.2

drjimwillie · June 4, 2019

PCIe Riser Ribbon Cable Data Point

There has been talk about the motherboards with more than one PCIe slot, and the conflict with the passive cooling cases. I had a CAPS Pipeline built for me by one of our favorite reputable suppliers/Gurus. He said that he liked the HDPlex H5 case better than the Streacom case, also I was one of the first to use the HDPlex 300w ATX LPS and he said that it would match. When I opened it and saw SOtM tx-USBexp PCIe card mounted sideways with the giant ribbon cable, I scratched my head. (see attached picture) I thought that has to degrade sound quality. I decided to put the card into the slot. While I had the Pipeline server pulled out, I did a couple of no cost power upgrades. I unplugged the internal ATX harness and pulled the harness from the ATX LPS directly into the server, making the internal harness obsolete. I also re-routed all of my power cables to the outsides of the rack, leaving the data and interconnects behind the rack. The USBexp card had a full height back plate on it, I tried to get a half size plate, but I did not want to wait for it to get shipped from Korea. There are 3 PCIe slots and only one lined up with a slot on the back of the case. The slot is extra short, the USB jack sits on the bottom of the slot, but the card seats fully. If I had the back plate I do not think it will fit. I was very upset. I thought that the builder of the Pipeline did not do proper research to coordinate the PCIe slots. I was thinking that if I went with the case that Chris recommended, then the slots would line up. Is this a common problem?? If I wanted to add other PCIe cards am I S.O.L.? I was thinking that I would have to chop a big hole in the back of the case? I do not have a Dremel and I am afraid of the metal filings…. What do others do?

OK - to the punch line: (unfortunately I did 3 things at the same time so I can’t attribute the effect to one thing, but I think that removing the two sided riser and the ribbon cable was the major upgrade)

The improvement was very noticeable. (I did a few upgrades, which I will list, but this was the one I was compelled to take notes) “Music sounds like music. Instruments had more body. The piano sounded like a whole piano, there was more texture and bass notes, it was easier to listen to. You could hear that singers were further away from you because they had depth and from the acoustics in hall.”

Other upgrades:

My QNAP HS253dx (slient-audiophile) NAS has two Ethernet jacks, which I bridged, it was a big improvement. But I was short one quality Ethernet cable to my Cisco switch. I bought a Ghent JSSG Metz Ethernet cable. The harmonics in the music were improved.

Sean Jacobs built me a regulator (12V) to power my NAS, it is being fed from the 19V of the HDPlex. I got it hardwired with cable glands to cut down on cables and connectors and lower impedance. He said it will take 4-6 weeks to break in. My NAS has my music on HDD, no SDD and is running Roon. (has anyone else tried this? This NAS was suggested by lmitche, but I do not know if he has one?) Sean said that the regulator is equal to the DC2, but with higher amperage. When I replaced the 12V from the HDPlex with the new power, the sensation was AHHHHHHHHH, that is how it should sound. This will only get better.

I replaced the LPS-1 feeding my Iso-Regen with LPS-1.2. It was a nice improvement, not as huge as I was hoping for. My intent was to move the LPS-1 over to the SOtM tx-USBexp PCIe card but the DC jack would not fit. Rather than force it I ordered a Gotham JSSG 360 DC cable from Ghent. I was joking that the improved power to the USB card should not make a difference because it was feeding the Iso-Regen. When the Gotham cable came in I put it on the LPS 1.2 and moved the Ghent DC cable with the Oyaide plugs to the USB card with the LPS-1. It made a very nice improvement, I was surprised. Smoother, fuller, more relaxed. More of an impact than the upgrade from the LPS-1 to the 1.2. Alex suggested that it may be because I was feeding the 1.2 with HDPlex 12V and I should try the power supply it came with, plugged directly into the wall. I currently am and I think that it is more dynamic, I did not listen critically because it is a new power supply.

I am really enjoying how all of this sounds and things are still burning-in.

So I have paid my dues with data point, listening impressions. I have a few questions, please.

(Bridging-how this thread started) -The QNAP NAS has two Ethernet jacks, it was easy to bridge them because there is a utility, I assigned the 10GB to the FIOS router and the 1GB to the server. (I think I am going direct to the server, I cant remember.) Path: FIOS Router – GS108 – Cisco Switch- NAS- Pipeline Server/Player. The Pipeline Supermicro X10SLH-F (data server) Motherboard has two Ethernet jacks. Should I bridge them? I am not using a NUC or other endpoint. The Pipeline is running Audio Linux and HQPe only. If I bridge the Ethernet jacks in the pipeline, do I continue to have a direct connection from the NAS? Do I set it up with two cables from the NAS to the switch and two cables from the Pipeline to the switch? Will the Pipeline use one cable for input and one for output?? Will this increase the POE noise in the Pipeline? Will you please help me think this through? A.L. has a bridging utility, can I assign input and output?

I am going to purchase the Uptone EtherRegen when it comes out. I wonder how this will effect things?

I have more questions, but I will wait for the answers on this one, please? Thank you

drjimwillie · June 4, 2019

It is two pieces. the first is two mail parts, one side for the PCI slot and the other for the ribbon cable.

I was made for me by Andrew of Small Green Computer, you can ask him.

Although my point is that it sounds much better with out it.

drjimwillie · June 5, 2019

Creating an ethernet bridge is the original topic of this thread. If you go to the first posts you will learn much more than I can explain.

If you have two ethernet jacks in your device/server you can create an internal switch that is also known as a bridge to connect in going and outgoing ethernet signals. It creates a direct connection Mostly intended for use with an ethernet NUC.

( I hope I get that right, it’s a lot of pressure with all the smart guys around here)🙂

drjimwillie · July 9, 2019

If a tree falls in the forest does it make a sound? How does my different typology sound compared to other systems; I may never know?

I am glad that we are talking about the impact of the NAS. I tried to bring this up previously and the subject did not get traction.

If a NAS does not have a fan, is fed clean power and has no SSD, is it still considered noisy?

My old NAS used to be located in a separate closet. I did not use it with my system. My CAPS Pipeline had a SSD with my music and Roon Server with Tidal. The closet with the FIOS router & Netgear GS108, is 30 feet away. The 30 foot Ghent JSSG ethernet cable was connected directly to the Pipeline server. I installed a Cisco switch near my equipment rack, connected the 30' cable to it and then a 3 foot cable to the Pipeline and it was a nice increase in SQ. I purchased a QNAP HS-453DX fanless NAS and installed it in the closet. (I did not install the SSD) Music from the NASs HDDs did not hurt SQ. I moved the NAS into the rack so I could connect it directly to the Cisco switch and had a bump up in SQ. I removed the SSD from my Pipeline, installed AudioLinux to run from RAM, I moved Roon to the NAS to lighten the load on the Pipeline and of course after all that, there was an increase in SQ. The NAS has two ethernet ports, 10Gb & 1Gb, they look different so they are easy to identify. The NAS has a utility to create all types of bridges. I created a bridge using the 1Gb connected directly to the Pipeline and the 10Gb to the switch and got a very nice increase in SQ. I added a custom Sean Jacobs regulator for the NAS and this is icing on the cake. I wonder how this sounds compared to other systems?

I am looking forward to the Uptone EtherRegen to replace the Cisco Switch. I will connect both bridged cables the EtherRegen and run a third cable from the "clean" output to the pipeline.

The Cisco switch does not allow for an easy power upgrade, the AC cord is hardwired. The Cisco switch is connected to my Ultra-Isolation transformer. Which would you consider nosier, the switch or the NAS? I am not sure if I have the will power to test it and may just wait for the EtherRegen.

My Pipeline has two ethernet ports. Do you think that there would be value in bridging them and connecting them both to the EtherRegen. (I can find how many ports it will have)

My plan is to eventually add an endpoint to this setup. Something like the SOtM motherboard with one of these high-power low watt Intel CPUs that are coming out.

I have added a sketch incase anyone would like to make a suggestion.

Thank you

Bill

drjimwillie · July 17, 2019

On 7/9/2019 at 3:08 PM, drjimwillie said:

If a tree falls in the forest does it make a sound? How does my different typology sound compared to other systems; I may never know?

I am glad that we are talking about the impact of the NAS. I tried to bring this up previously and the subject did not get traction.

If a NAS does not have a fan, is fed clean power and has no SSD, is it still considered noisy?

My old NAS used to be located in a separate closet. I did not use it with my system. My CAPS Pipeline had a SSD with my music and Roon Server with Tidal. The closet with the FIOS router & Netgear GS108, is 30 feet away. The 30 foot Ghent JSSG ethernet cable was connected directly to the Pipeline server. I installed a Cisco switch near my equipment rack, connected the 30' cable to it and then a 3 foot cable to the Pipeline and it was a nice increase in SQ. I purchased a QNAP HS-453DX fanless NAS and installed it in the closet. (I did not install the SSD) Music from the NASs HDDs did not hurt SQ. I moved the NAS into the rack so I could connect it directly to the Cisco switch and had a bump up in SQ. I removed the SSD from my Pipeline, installed AudioLinux to run from RAM, I moved Roon to the NAS to lighten the load on the Pipeline and of course after all that, there was an increase in SQ. The NAS has two ethernet ports, 10Gb & 1Gb, they look different so they are easy to identify. The NAS has a utility to create all types of bridges. I created a bridge using the 1Gb connected directly to the Pipeline and the 10Gb to the switch and got a very nice increase in SQ. I added a custom Sean Jacobs regulator for the NAS and this is icing on the cake. I wonder how this sounds compared to other systems?

I am looking forward to the Uptone EtherRegen to replace the Cisco Switch. I will connect both bridged cables the EtherRegen and run a third cable from the "clean" output to the pipeline.

The Cisco switch does not allow for an easy power upgrade, the AC cord is hardwired. The Cisco switch is connected to my Ultra-Isolation transformer. Which would you consider nosier, the switch or the NAS? I am not sure if I have the will power to test it and may just wait for the EtherRegen.

My Pipeline has two ethernet ports. Do you think that there would be value in bridging them and connecting them both to the EtherRegen. (I can find how many ports it will have)

My plan is to eventually add an endpoint to this setup. Something like the SOtM motherboard with one of these high-power low watt Intel CPUs that are coming out.

I have added a sketch incase anyone would like to make a suggestion.

Thank you

Bill

Let me ask a bridging question that will be applicable to more systems.

If I understand correctly the typical bridging typology is a server with two ethernet ports, one port is in from the router/switch and the other out directly to a NUC or NAA.

If you are using a fancy switch or something like the EtherRegen, do you continue to use your bridge? Where as, you have two ethernet cables out from your server to the switch and then a separate cable out to your NUC?

Will the switch use one ethernet cable for incoming data and the other for outgoing data?

I thought that I had read that more ethernet connections made for a better sounding system. Can anyone confirm that?

I plan to test this with my switch but I am having a few issues at the moment that I want to resolve before I try it.

In theory I could use two cables out of my NAS and two cables out of my Pipeline server. Then later when I have an endpoint I could add one more cable, out to the end point.

drjimwillie · July 29, 2019

On ‎7‎/‎17‎/‎2019 at 4:02 PM, drjimwillie said:

Let me ask a bridging question that will be applicable to more systems.

If I understand correctly the typical bridging typology is a server with two ethernet ports, one port is in from the router/switch and the other out directly to a NUC or NAA.

If you are using a fancy switch or something like the EtherRegen, do you continue to use your bridge? Where as, you have two ethernet cables out from your server to the switch and then a separate cable out to your NUC?

Will the switch use one ethernet cable for incoming data and the other for outgoing data?

I thought that I had read that more ethernet connections made for a better sounding system. Can anyone confirm that?

I plan to test this with my switch but I am having a few issues at the moment that I want to resolve before I try it.

In theory I could use two cables out of my NAS and two cables out of my Pipeline server. Then later when I have an endpoint I could add one more cable, out to the end point.

On ‎7‎/‎21‎/‎2019 at 12:37 PM, austinpop said:

Obviously, we won't know until the ER is shipped and in our hands. Here is my experience with the sNH-10G, using not HQP but Roon.

It helps to think of the switch not for its switching functions, but as a cleanser/isolator. Obviously, you want to test multiple configurations, but the one I'd start with is this:

upstream <-> (Ethernet port 1) bridged server (Ethernet port 2) <-> (dirty side) reclocking/isolating switch (clean side) <-> endpoint <- USB -> DAC

if you have a NAS, it can either be somewhere upstream, or attached to the dirty side. This seems to be a personal preference thing. I prefer my NAS far away upstream in another room.

When used with Roon Core on the server and Stylus EP on the endpoint (both on Euphony OS), bridging the server definitely still sounded better. The unbridged case looked like this:

_________

upstream <-----------> | Switch | <---------------> endpoint

--------------

^

|

server <-------------------------|

When the Switch == sNH-10G, I found the bridged case to sound distinctly better. Only time will tell if this result holds for the ER.

Thank you for your reply. Although it leaves me a bit confused.

Are you saying that with the sNH-10G as the switch, using the bridged connection (2 ethernet cables) to the switch sounded better.

Here is a sketch of what I am asking about. (I am having trouble inserting this simple sketch as an image, into the post, how do I do it?)

Server Bridge to Switch.tif

drjimwillie · July 29, 2019

Is anyone using this configuration?

How does it sound?

What did you compare it to?

What switch are you using?

drjimwillie · July 29, 2019

I understand that a bridge is a virtual switch inside your computer.

I am currently using a bridge, Port one in from my switch, port two out to my endpoint.

What I’m asking about is a little bit of out-of-the-box thinking.

What if you do what I have in the sketch?

Keep the bridged connection configured on your server and connect both ends to the switch.

Is anyone doing it?

Does it work?

How does it sound!

drjimwillie · July 29, 2019

I thought about this because of two reasons.

1. Let me please preface this with the fact that we read a lot of things here connected to audio. I thought that I had read somewhere regarding ethernet connections that more Ethernet connections made for a better sound. But I could not find the reference. From my memory the person had two ethernet ports on their server and they connected them both to their switch and because of the increased data flow the sound was better. I was assuming that possibly one ethernet cable would be used for the incoming signal and the other ethernet cable would be used for the outgoing ethernet signal.

Then either by pontification or test the theory was if you had two cards with 4 ethernet ports that 4 cables made even better sound. I ask the question because I wondered if anyone here had actually tried it?

2. The other reason I am asking this question is because I am doing my homework to be prepared for the EtherRegen. If there is complete isolation from one side of the switch to the other then there may not be a detriment to hooking up multiple cables.

drjimwillie · July 30, 2019

I read about trying to make my system sound better for a couple of hours every day, unfortunately this is more time than I get to listen to my system. 😕

Do that every day for a few years and some of the references get lost. And I’m sure some of the metaphors get mixed.

I remember reading about link aggregation but I’m not sure if that’s the basis for a better sound?

I thought it was kind of a wacky idea, which is why I asked the question here.

I was wondering if you could use or how you could use a switch in conjunction with the bridge

I would try my idea, but I’m having a couple of technical difficulties in my system is not conducive to critical listening at the moment. When I try it I will let you know. I am surprised that no one else has tried it up to this point.

I am not sure how the link to the JS article proves my point. I cannot find a correlation after reading it. Would you please explain what you’re thinking?

I have learned that I should try the switch after the server rather than before it. Although the switch before it did increase the SQ. I think the increase in SQ is because the ethernet cable from the router to the server is 30 feet long and it benefited from the signal regeneration.

Thank you

drjimwillie · December 10, 2019

In the write up by Romaz, he says the SMG Extreme uses a ASUS server motherboard. Can anyone report on exactly which motherboard?

drjimwillie · January 13, 2020

How much power is recommended to power the CPU separately?

And other system recommendations, please?

I have a C.A.P.S. Pipeline server/endpoint, AudioLinux loaded into RAM, only running HQPe. No harddrive. I have Roon running on my NAS. The Pipeline is powered by a HDPlex 300W ATX LPS. I was one of the first to power every component of my computer with a LPS. (Ancient history alert: at that time I was waiting for the microRendu to be released and because of my impatience, my available funds went to the LPS upgrade.) I was using the power taps on the HDPlex to power my SOtM USB card and IsoRegen, the 19V tap was powering the NAS. Now I am only using the molex connectors out of the HDPlex to the motherboard, w/o the molex for the removed harddrive. Every time I removed a load from the HDPlex, the System sounded better and the PSU ran cooler. When I heard the good reviews of the HDP 400W ATX LPS I asked Larry if I could have my PSU upgraded because I wanted cleaner power. He said that the 300W is basically the same as the 400W. Now I hear good things about the HDPlex 800W DC-ATX. What if I use the 19V tap for the 800W DC-ATX. I asked Larry if I used only the 19V tap, would I get the full 300W. He said that I would only get 190W max, & suggested I keep using the system as I am currently. Once again I am back where I started.

I have a 50W Zero-zone PSU that I am not using, so I thought it would be good to order an MPAudio HPULN for it, to clean up the power. I have not used the HPULN and do not have immediate use for it. (I now have an UpTone LPS 1.2 for every component that can be externally powered. My system is SOtM USB card > IsoRegen > Singxer SU-1 I2S. I thought with an LPS 1.2 on the IR & the SU-1 and USB card powered by the HDPlex LPS, that a third LPS 1.2 would not make a difference. I was wrong. Powering the TxUSBexp with a third LPS 1.2 was immediately much better) I thought that the HPULN had not shipped yet and a different way to solve the issue of getting better power into the Pipeline would be to have a separate PSU built for the CPU. I could use the almost $200 for the HPULN towards a PSU from MPAudio for the CPU.

With Austinpop’s reports on the importance of a muscular PSU for the CPU, the question is what is the required amperage? What is the wattage of the CPU that Austinpop successfully powered 12v/7A? (I think) The Pipeline has a Xeon CPU that is 80W. 80W divided by 12V = 6.6A. But, this is the rated power and we have learned that heavily loaded the CPU will use more than its rated value. Is 10A enough?

MPAudio says the 10A is 20 regulators, 22 for some extra headroom. Estimate 500 euro. How would this PSU design compare to the other popular PSUs?

Would it be worth the extra $30 to ship the now in-transit HPULN back to use the credit?

Now is where it gets crazy and I beg for a reality check. (sorry for the long post)

I want to make sure that a dedicated PSU for a CPU will stay viable over the long term. I am only running a one-box system because I am waiting for the dust to settle and an endpoint that will emerge that checks all of my boxes. I missed on the microRendu and things kept getting better with the off the shelf NUC. I fear that the day I order a NUC or iBox-V1000, the SOtM motherboard will come out or something from one of our other heros. Also, I am not in a hurry because my system with the EtherRegen already sounds magical! So, I do not mind waiting. But, while I wait I might as well make my existing system sound as good as I can. Upgrading the PSU for the CPU will do that, as long as I can reuse it later.

I use HQPlayer which has the new EC modulators would need a 5.0 GHz CPU to run DSD512, which currently does not exist. My Pipeline can run the EC modulator at DSD128, but the CPU is working so hard that it sounds like crap. I run PCM at PCM384 and it sounds great. The next Xeon CPU that Miska is looking at does 4.7 GHz turbo and is 155W. To future-proof for a dedicated CPU PSU with that load would be crazy. I could possibly move the new CPU PSU over to the future endpoint, but it will probably take 19V and what are the odds that it will have a separate input to power the CPU. Also, the endpoint PSU should be the “best you can get”, which probably means double regulated and I am not going to build a double regulated CPU PSU for the server.

I have the EtherRegen but have never sent ethernet to an endpoint, how important will the PSU for the CPU be then?

This is why I am reaching out to my friends here for advice, please. What would you do. What should I do.

Thank you

drjimwillie · January 14, 2020

4 hours ago, StreamFidelity said:

I also use the HQPlayer and listen enthusiastically with the EC modulator ASDM7EC with DSD 256. For this I use a CPU Intel Core i9-9900K (95W TDP), 3.6GHz - 5.0GHz OctaCore with 16 threads and 4.1GHz. 8.0GHz would be required for DSD516!

I have a lot behind me regarding the power supply. My HDPLEX 200W Linear Power Supply went up in smoke. Since I have significantly improved my power supply (Power 19V/16A - see picture), the sound has increased again. My consumption is a maximum of 19V / 5A.

How do you know that your max consumption is 19V /5A? Did you measure in line with both of the 8 pin CPU EPS at the same time, please?

or is that for the entire computer?

thank you

drjimwillie · January 14, 2020

7 hours ago, shahed99 said:

Romaz has mentioned in his last post that powering the CPU/EPS through HDPlex 800W DC-ATX sounds better than directly powering by a 12V rail.

My understanding is that Romaz has said powering CPU/EPS through the 800W HDPlex converter is better than powering it directly from the 12V of his 400W HDPlex, because the spec of the 800W is better (.1V variation).
Not from any other separate 12V supply. A separate 12V supply should sound as good or better as long as the specifications are the same.

drjimwillie · January 14, 2020

@StreamFidelity your system must sound really good with the two Keces and the 800W HDPlex powering them.
your system drawing is very clear and well done.

I can see from the picture that you have both PSU set to 20V and combined they are using about 4A, so the total power consumption of your server is 80W. (I forget, is this build the server or the endpoint or both?) But, we do not know what the CPU by itself is using.
It is interesting that your CPU is 95W and the entire system is only using 80 W. Were you playing music when you took this picture?

So, this is a good reference point.

It is also why I asked Austinpop what the power rating of his CPU is?

Can anyone of us, who is powering the CPU separately, see what the actual consumption is? Or, is that not as important as knowing that you need a lot of power to keep up with the instantaneous demand of the CPU.

drjimwillie · January 16, 2020

This is interesting and counter to what Austinpop reported. That higher amperage sounded better, than a better quality and less amperage PSU, when powering the CPU.

I wonder if he was powering the server or the endpoint? Or a one box system?

I wonder if increasing the amperage to the CPU of the server makes an improvement when sending data over ethernet to the endpoint?

drjimwillie · January 16, 2020

Bodiebill- how much of that 33A goes to the CPU and how many watts is the CPU, please?

drjimwillie · January 16, 2020

@bodiebill is the 400 W Seasonic currently feeding the 400W HDPlex DC – ATX?

What is the amperage limit of the 12V rail or output feeding the CPU/EPS?

drjimwillie · January 16, 2020

Also, are you saying that changing the PSU of the server did not affect the sound quality of the endpoint?

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

Recommended Posts

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Create an account or sign in to comment

Create an account

Sign in