Jump to content
IGNORED

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


Message added by The Computer Audiophile

Important and useful information about this thread

Posting guidelines

History and index of useful posts

Most important: please realize this thread is about bleeding edge experimentation and discovery. No one has The Answer™. If you are not into tweaking, just know that you can have a musically satisfying system without doing any of the nutty things we do here.

Recommended Posts

19 minutes ago, seeteeyou said:

 

Hi Larry,

 

Ouch, I only talked about ground loops because I was trying to make sure that batteries are OK as long as they're off the grid. In other words, I was simply trying to poke some holes at my game plan (just in case I wasn't aware of other potential issues) while I didn't have doubts about what you've accomplished already.

 

I was being "extra" cautious because some audiophiles were also having battery-powered source components like older MacBook etc. (I read that a long time ago and hopefully that's actually how it went.) Somehow some of them liked it better when it's plugged in while the battery was removed, the others would prefer that to be unplugged from the wall and therefore it's only powered by the battery.

 

BTW, that Intel Wireless-AC 9462-D2W module came with MHF4 connectors as follows

 

https://www.intel.com/content/dam/support/us/en/documents/mini-pcs/nuc-kits/NUC7xJY_TechProdSpec.pdf#page=28

 

Are we supposed to get a pair of pigtails first?

 

https://www.data-alliance.net/mhf4-to-sma-male-cable-3-inch-4-inch-5-inch-6-inch-8-inch-16-inch/

 

And then add a pair of extension coaxial cable cords between those pigtails and antennas?

 

https://www.ebay.com/itm/2X-10ft-WiFi-Antenna-RP-SMA-Extension-Coaxial-Cable-Cord-for-Wi-Fi-Router-Line/192196205184

Hi CTU,

 

You are good. Yes, I ordered pigtails from Amazon but the ipx end was too big. So I unsoldered the sma connector on the pigtail. I then unsoldered the stock Intel NUC antenna freeing the cable and resoldered it to the sma connector. Thank God for magnifying glasses. It worked! This was a DIY miracle.

 

There are two antenna connections on the WiFi card, one for 2.4ghz and the other for 5ghz. I used the 5GHz so only one pigtail and antenna wire is needed.

 

Larry

Pareto Audio aka nuckleheadaudio

Link to comment

 

19 hours ago, lmitche said:

NUC Update:

 

Ok, so all is working well with roonserver on the Nuc7cjyh in the dirty zone booting from and hosting the 4tb music disk powered by its stock smps.

 

In the clean zone is the Nuc7cjysal with headless Audiolinux and a 10 foot coax wire to the wifi antenna sitting inches from the wifi router.  Throughput is 433mbps, so more than enough for roon. The second connection is the uspcb to the iso regen, and the third 19 volt power from a sigma 11 lps.There is no chance for a ground loop here as there is only one powered device attached to the roonbridge Nuc. This Nuc boots from the onboard 32gb emmc memory and powers it's own internal wifi card. Roonbridge starts at boot.

 
The SQ is the best I have ever heard anywhere, just breath taking. 
 
It is hard to describe the level of openness and sense of presence without a single harsh sound. I can hear tremendous detail such as breath sounds for the first time on well known albums, players with instruments I never heard before and an expansive sound stage with height, depth and breadth. The SQ is there at any resolution and format, PCM or DSD. I can easily hit ear bleed levels which is a bit dangerous. I'm running 3 to 4 db lower volume then usual. Coherence is off the charts and eveything you hear makes sense in space and time. 
 
I am very grateful to Piero Olmeida for his Audiolinux OS and outstanding service and support. I am pleased to be using Audiolinux fulltime again and won't miss that other OS.

Sounds complicated to setup... why would one choose this route vs a microRendu for Roon endpoint?

Regards,

Dave

 

Audio system

Link to comment
12 hours ago, the_doc735 said:

In this scenario, you could get your electricity company/provider to install a second line to your property with a second electricity meter purely for Hi Fi usage; that way would provide greater isolation, "BUT!" ~ back at the electric companies SUB STATION there will still be a meeting point somewhere! Unless of course, the electric company are willing to build a dedicate electricity generator solely providing you with power i.e. isolated from the rest of the grid! Your own power station, no less!

Don't laugh...

 

https://gizmodo.com/obsessed-audiophiles-in-japan-are-installing-their-own-1785291714

Link to comment
59 minutes ago, lmitche said:

There are two antenna connections on the WiFi card, one for 2.4ghz and the other for 5ghz. I used the 5GHz so only one pigtail and antenna wire is needed.

 

Larry

 

Hi Larry,

 

Many thanks for your prompt replies and there's something I still couldn't understand. Is it any different if I were leaving NUC as is without having to solder anything or add any pigtails, and then insert those extension coaxial cable cords between my wireless router and its bundled antennas instead? I just watched this video linked below that's why I really wasn't so sure about how much would that Akasa aluminium fanless case affect the signal strength

 

https://youtu.be/26OxCwEHoTk

 

Supposedly I'm placing a bundled antenna (that's extended all the way from my wireless router) inches away from Akasa, do you think the maximum speed @ 433Mbps is still good to go without any dropouts?

 

Alternatively, I was also speculating what would happen if we're bypassing the wireless router altogether since we've got both NUC7CJYH (dirty side) and NUC7CJYSAL (clean side) that are equipped with the same soldered down Wi-Fi card. Let's say headless version of AudioLinux could be configured to join an ad-hoc network while maintaining a stable connection @ 433Mbps, that's essentially the wireless flavor of what Roy was doing at the very beginning when this thread was started

 

https://wiki.archlinux.org/index.php/ad-hoc_networking

https://wiki.archlinux.org/index.php/Wireless_network_configuration

https://www.reddit.com/r/archlinux/comments/4tyi92/how_to_set_up_wifidirect_between_arch_pc_and/

 

Wireless router is outta the picture, just like how Roy was getting rid of the switch back then. In other words, the packets no longer have to jump through all the hoops while we could even try to disable the encryption just in case the SQ might be affected. I understand stand that good fences make good neighbors but in this particular case we don't really have much (if anything) to lose.

 

Rather than "butchering" that Wi-Fi card on the dirty side, we could also add this USB adapter instead and then that extension coaxial cable cord is inserted between the antenna and the RP-SMA connector accordingly

 

https://wikidevi.com/wiki/ASUS_USB-AC56

https://www.amazon.com/Asus-USB-AC56-Dual-band-Wireless-AC1300-Adapter/dp/B00FB45USW

68Q3kbM.jpg j6lb0Fs.jpg

 

If that were doable and the connection is stable, hopefully adding one more USB device on the dirty side wouldn't affect the SQ while we no longer have to mess with those tiny little pigtails and internal antenna cables inside the NUC itself.

 

Does it make sense at all or am I going off the wall?

Link to comment
On 9/20/2018 at 12:18 PM, lmitche said:

Andrew,

 

Yes indeed, you are right. Coincidentally I ordered a copy seconds before reading this. It may even sound better.

 

Regular Audiolinux will run very happily in 4gb. Ramboot takes the larger 16gb footprint. Headless should Ramboot with 8gb.

 

Larry

@lmitche 

Hi Larry

 

Thanks for sharing your experiments with your NUC setup and congratulations on achieving the superb results.

Can I double check with you about the above - that AudioLinux headless Ramboots with 8gb. Does this mean Roon core can be run adequately in this scenario without using the 16gb option...?

 

Cheers Patrick

 

Topaz 2.5Kva Isolation Transformer > EtherRegen switch powered by Paul Hynes SR4 LPS >MacBook Pro 2013 > EC Designs PowerDac SX > TNT UBYTE-2 Speaker cables > Omega Super Alnico Monitors > 2x Rel T Zero Subwoofers. 

Link to comment

CTU, in moving the nuc board to the akasa case the wifi antennae were left behind. I first rescued them and placed the wires through the antenna holes in the case and taped them to the antenna to the outside. That works but then realized I didn't want the RF near rhe NUC and DAC so the pigtail was hacked and 10 foot antenna installed on the 5ghz side only.

 

I am sure one can buy a pigtail with the appropriate connector and avoid the pain of the DIY approach. Have you found a source?

 

I didn't realize Roy was talking about an adhoc network link. Yes that could be done in theory with an ethernet connection on the dirty side nuc and an adhoc connection bridged to the roonbridge machine and bypassing the router. But, as we already have galvanic isolation, I'm not sure it would buy much. Did Roy test this or was it theory? I thought he was using Adrian's etherregen-ish switch.

 

We are both regularly off the wall.

 

?

Pareto Audio aka nuckleheadaudio

Link to comment
18 minutes ago, tapatrick said:

@lmitche 

Hi Larry

 

Thanks for sharing your experiments with your NUC setup and congratulations on achieving the superb results.

Can I double check with you about the above - that AudioLinux headless Ramboots with 8gb. Does this mean Roon core can be run adequately in this scenario without using the 16gb option...?

 

Cheers Patrick

 

Patrick, I haven't tried it as two 8gb boards are in my nuc. However I would be very surprised if headless Audiolinux didnt run roonserver ramboot with 8gb

Pareto Audio aka nuckleheadaudio

Link to comment
57 minutes ago, lmitche said:

In one of the posts above I acknowledged that this NUC exercise is a recreation of a ethernet to usb streamer, a wireless one. And yes it is complicated to setup especially given the immature state of the NUC bios. But that challenge is part of the fun of this, at least for me.

 

I would like to say the sq is better than a microrendu or sms200 and their variants, but I can't. Years ago,  I had early versions of both here, and they were immature products with operational issues that prevented an adequate evaluation. Given this history I have nothing to say about SQ comparisons.

Thanks, always looking to see if someone has found a better mouse trap.

Regards,

Dave

 

Audio system

Link to comment

@lmitche let's say we call the SQ of your latest config (AudioLinux NUC streaming to another AudioLinux NUC) a 10.

 

Where would you rank your

- standalone AudioLinux NUC

- standalone Windows 10 big rig

- Windows 10 big rig streaming to AudioLinux NUC (not sure if you ever tried this)

Pareto Audio AMD 7700 Server --> Berkeley Alpha USB --> Jeff Rowland Aeris --> Jeff Rowland 625 S2 --> Focal Utopia 3 Diablos with 2 x Focal Electra SW 1000 BE subs

 

i7-6700K/Windows 10  --> EVGA Nu Audio Card --> Focal CMS50's 

Link to comment
16 minutes ago, rickca said:

@lmitche let's say we call the SQ of your latest config (AudioLinux NUC streaming to another AudioLinux NUC) a 10.

 

Where would you rank your

- standalone AudioLinux NUC

- standalone Windows 10 big rig

- Windows 10 big rig streaming to AudioLinux NUC (not sure if you ever tried this)

Rick,

 

You have the ranking right, but I'm not ready to rate just yet. I've done the build in many small increments, each an improvement, but it makes it hard to say what each jump is worth.

 

The big rig is put away at the moment as I'm using the 19 volt lpsu for the NUC so no comparisons there.

Pareto Audio aka nuckleheadaudio

Link to comment
6 hours ago, seeteeyou said:

 

Thanks Larry for documenting your journey, we're so glad to have you and Roy here who share your findings with everyone.

 

I've got a question regarding what we could do to optimize the power supplies on the clean side since Roy mentioned this before

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=361&tab=comments#comment-860642

 

Given that output of LT3045 couldn't go any higher than 15V, do you foresee any potential issues if we're combining two of them (i.e. 7V + 12V in series) into 19V with a Y-cable as shown below?

 

https://www.computeraudiophile.com/forums/topic/31554-diy-dc-power-cables/?page=24&tab=comments#comment-678934

y9WMd0V.jpg

 

Let's say that's causing no harm whatsoever, now it's just a matter of providing enough current @ 19V

 

https://nucblog.net/2018/04/gemini-lake-nuc-review-nuc7cjyh-linux-htpc-conclusions/

 

Granted running AudioLinux in headless mode with RoonBridge shouldn't draw THAT much current to begin with, though we'll just play it safe by going for 19V × 1.5A = 28.5W like this

 

http://www.ldovr.com/product-p/dxp-1a5dsc.htm

XJHj6nH.png xWq9oZs.png

 

Besides feeding LT3045 with Sigma 11 LPS, what else do you recommend?

 

I'm just wondering if one power bank for 9V plus another for 15V were any good, batteries that are disconnected from an AC outlet should be able to prevent any ground loops from happening

 

https://www.ravpower.com/26800mah-Type-C-external-battery-charger-black.html

https://www.amazon.com/RAVPower-Portable-26800mAh-Recharged-Nintendo/dp/B01LRQDAEI

https://www.bhphotovideo.com/c/product/1313332-REG/ravpower_65_02058_101_rp_pb058_26800mah_usb_c.html

 

Other than that, do you guys have other suggestions for 19V?

I have been told these battery banks contain switching technology? (to achieve their various voltages)...

Link to comment
2 hours ago, lmitche said:

In one of the posts above I acknowledged that this NUC exercise is a recreation of a ethernet to usb streamer, a wireless one. And yes it is complicated to setup especially given the immature state of the NUC bios. But that challenge is part of the fun of this, at least for me.

 

I would like to say the sq is better than a microrendu or sms200 and their variants, but I can't. Years ago,  I had early versions of both here, and they were immature products with operational issues that prevented an adequate evaluation. Given this history I have nothing to say about SQ comparisons.

NUC Vs. microrendu or sms200 ? ...comparisons?

Link to comment
1 hour ago, lmitche said:

Patrick, I haven't tried it as two 8gb boards are in my nuc. However I would be very surprised if headless Audiolinux didnt run roonserver ramboot with 8gb

Great thanks Larry.

Topaz 2.5Kva Isolation Transformer > EtherRegen switch powered by Paul Hynes SR4 LPS >MacBook Pro 2013 > EC Designs PowerDac SX > TNT UBYTE-2 Speaker cables > Omega Super Alnico Monitors > 2x Rel T Zero Subwoofers. 

Link to comment
4 minutes ago, seeteeyou said:

Once again Roy also talked about why double regulated is better than single regulated counterpart here

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=361&tab=comments#comment-860642

On August 10, 2018 at 3:01 AM, romaz said:

Regarding power supplies for this NUC board, yes, it scales extremely well to power supplies and better than my DFI board.  This board can accept 12-19V.  During my testing with my SR7s, 19V sounds a little better than 12V but DR (double regulated) 12V sounds considerably better than 19V.  The ideal SR7 for this board would likely be a DR 19V which I do not have.

 

My DIY experience with multi-stage regulation very positive for SQ. May do more in future (got low hanging fruit).

 

But confused about 12-19V range from @romaz quote. I unaware of (non-ps) circuitry consuming 19V, thought input V all down converted to 12v, 5v, 3.3v, 1.?v or similar. Clarification of what-is-what in load through the 12->19V range, maybe helpful to understand why "19V sounds a little better". Maybe point way to dropping inconvenient high voltage requirement? 

Link to comment
2 minutes ago, look&listen said:

But confused about 12-19V range from @romaz quote. I unaware of (non-ps) circuitry consuming 19V, thought input V all down converted to 12v, 5v, 3.3v, 1.?v or similar. Clarification of what-is-what in load through the 12->19V range, maybe helpful to understand why "19V sounds a little better". Maybe point way to dropping inconvenient high voltage requirement? 

 

Roy should have explained something about current draw and voltage back in March 2017

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=47&tab=comments#comment-644456

Link to comment
2 hours ago, lmitche said:

I am sure one can buy a pigtail with the appropriate connector and avoid the pain of the DIY approach. Have you found a source?

 

Larry, let's see if stuff like this were the least painful way or not. So far the longest pigtail I could find was about 31.4 inches (i.e. 80cm) but they're coming from China

 

https://item.taobao.com/item.htm?id=540183870295

https://www.aliexpress.com/item/A-pair-of-IPEX-MHF4-80cm-31-4-2-4-5G-wifi-antennas-for-Intel-7260/32682007357.html

https://www.aliexpress.com/item/A-pair-of-IPEX-MHF4-80cm-31-4-2-4-5G-wifi-antennas-for-Intel-7260/32858347028.html

 

These guys make something up to 20 inches

 

https://www.data-alliance.net/mhf4-to-rp-sma-female-cable-1-inch-2-in-3-in-4-in-6-in-8-in-10-in-1-ft-14-in-16-in-18-in-20-inch/

 

Other than that, we could also try Pasternack

 

https://www.pasternack.com/t-cable-creator.aspx

 

They seemed to something that's gonna be a good match, though I wasn't able to select HMCX32 from their page The Cable Creator linked above

 

https://www.pasternack.com/pages/RF-Microwave-and-Millimeter-Wave-Products/ultra-miniature-coaxial-cable-assemblies-up-to-6-ghz.html

Quote

Compatible with Murata HSC® and IPEX MHF4®

 

Could 31.4 inches away from an Akasa case be far enough to mitigate the effect of RF noise? If not, do you think it could be taken care of if we're going for the one that's terminated with RP-SMA and then add another extension coaxial cable cord?

Link to comment
45 minutes ago, seeteeyou said:

Roy should have explained something about current draw and voltage back in March 2017

Yes, Thx!  Was nice balancing of I vs E over Noise, but theory not implementation. 1st question remains, but now details of regulation/conversion stages interesting.

 

Reading @romazmulti-subject reply post saw most answers consist of balancing of factors. I recognize mind of engineer- to know & balance many factors to achieve multiple goals. Also sometimes hear from other posters, but this one clear & comforting (so needed here & now :( )

Link to comment
18 minutes ago, look&listen said:

Reading @romazmulti-subject reply post saw most answers consist of balancing of factors. I recognize mind of engineer- to know & balance many factors to achieve multiple goals. Also sometimes hear from other posters, but this one clear & comforting (so needed here & now :( )

 

We just need to bribe Rajiv and then he'll be our spy, then we're gonna get the latest and greatest intelligence before that's posted by Roy.

 

I guess that he might have talked to Adrian about further improving Dream OS or something. For instance, we could tweak the Linux kernel by changing the value of CONFIG_HZ

 

https://linuxmusicians.com/viewtopic.php?t=9339&start=15#p29673

https://github.com/MintCN/linux-insides-zh/blob/master/Timers/linux-timers-1.md

http://www.my-hiend.com/vbb/showthread.php?5404-走向超值而極緻的-HI-FI電腦訊源(連載)&p=228756#post228756

 

That value is 250Hz (4ms) by default and we could increase that drastically in order to lower the latency. An audiophile in Taiwan tried some values such as 44100Hz (~0.023ms) 22050Hz (~0.045ms) 11025Hz (~0.09ms) since the vast majority of the files in his music library turned out to be in Redbook format. Finally he concluded that all three of them sounded pretty much the same and therefore 11025Hz was the chosen one with the lowest CPU usage.

Link to comment
1 hour ago, seeteeyou said:

 

We just need to bribe Rajiv and then he'll be our spy,

 

Not sure what you're saying, and not going to ask. Sorry guys - I'm very interested in the findings that result from these efforts, but much as I love to experiment, tinkering at the level that Larry et al. do is just not my thing. I'm happy to wait and see what results.

 

My plans for now are the same - I am still waiting to see if Adrian's DS-1, with OCXO system clock, powered by and SR-7 rail, can outperform my Zenith SE. 

Link to comment
1 hour ago, austinpop said:

 

Not sure what you're saying, and not going to ask. Sorry guys - I'm very interested in the findings that result from these efforts, but much as I love to experiment, tinkering at the level that Larry et al. do is just not my thing. I'm happy to wait and see what results.

 

My plans for now are the same - I am still waiting to see if Adrian's DS-1, with OCXO system clock, powered by and SR-7 rail, can outperform my Zenith SE. 

Will you consider to audition Adrian's DS-1 soon or are you waiting for someone else And see his findings?

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...