Tír Na HiFi

Posted this on the JPlay forum but it might be just as at home here!

I have been on a bit of a journey with PC based playback this past couple of months and I thought it about time to recount my experiences for those who may have been on a similar sort of journey particularly when it comes to single PC and Streaming solutions.

For those who may not be familiar with my background with Jplay I have been a long-time advocate of Jplay and particularly the streaming solution using 2 well specified passive PCs before I went down the clean power supply route. I also has Audiophile server core running on both PC’s.

About 3 months ago I was at a gathering of PC based audio enthusiasts from TirNaHiFi where each of us had various streaming solutions running with Jplay where we had been exploring the benefits of various implementations of PCs with both Battery and Linear power supply solutions. We were fortunate to also have in attendance several Vinyl enthusiasts with years of experience with vinyl rigs between them.

We each were enjoying the benefits of these power solutions with all the good LAN cables with various isolator solutions on a very good audio system, we also had a good Turntable and Phono stage to reference off as well to make sure we were not blowing wind up one another’s ass!

At the end of the evening one of the guys had a copy of MQN which he thought sounded good on a single PC so we all agreed to have a listen to a track or two! What happened next left me and everyone there amazed and confused, we all preferred easily the single PC with MQN and even the vinyl guys liked the MQN single PC playback. Needless to say this ended the night with a lot of very quiet Jplay enthusiasts heading off into the night very confused but yet excited about what we heard.

Roll on about a month later where I had taken time away from streaming and also Jplay trying various versions of MQN and loving the experience on a single PC. During this time I continued to improve this single Zuma PC with all PPA gear including SD SATA drive, USB PCI Card, 2 x SATA and 1 USB Red cables with good improvements. I also added linear power direct to the Mobo to 5v and 12v and everything was sounding really good.

During that month I discovered I could take a wave file and play it back without MQN through Total Commander and I was shocked to hear no real loss or difference in sound quality. I then fired up Jplay Mini and again no difference to speak of. Since that moment I have been back on Jplay Mini 5.1 and I have been loving the sound from my system for over a month now.

My current conclusion is that networking between 2 PCs no matter how well built they are is not as good as a single well-built PC with clean dc power to as many parts of the PC as possible, and also having a well stripped Server 2012 installation with either Scott’s optimisation or Phil’s Server optimisation. No network connection at all!!

I also think maybe Jplay and MQN do a good job of minimising the negative influence of switch mode power supply solutions in the windows environment and that Jplay does a good job of minimising the negative impact of current LAN solutions and the noise that comes with this approach. Maybe some of the recent discussions on connecting PC’s via fibre for streaming might be the future.

Saying all of the above I could be wrong and in the new year I will give streaming one more try now that I have what is a really good audio PC and fire up the Cirrus 7 again as a control PC to see if streaming can match what I am hearing from one PC, I doubt it but I could be wrong!!
I would be interested to hear if anyone else has gone back to try a single highly tuned audio PC with good hardware against their streaming solution?

I am looking forward to trying Jplay 5.2 but not expecting to hear any real difference as I believe most of the improvements to be had are now to be gained from good hardware implementations with a lean server 2012 R2 solution and possibly a lean Win 8.1 solution. Would be nice to be surprised again though.
Cheers, Pearse.

Nothing much to add, except that its good to summarise where you are and have a think as logically as possible about it.Never was this as true as with PC audio. The rate of development is so fast that its very easy to get lost/left behind. My own thinking is along the lines of your Pearse - I think power feed and power consumption are major factors at play. Do anything to improve one, other or both and the SQ improves. its just not always easy to figure out the information.


Fran

I'm starting to think that playing music through a cpu is fundamentally flawed

This seems like a better solution http://www.the-ear.net/review-hardware/ ... mer-server

Interesting journey. Pearse - thanks for taking the time to write it up.

Yes, I find your conclusions interesting - it would seem that improved software or improved PS result in the same or similar audible benefits. I guess the question is what underlying mechanism is being improved - noise, timing, both or something else? I still don't know which. Your experiences suggest noise might be the issue but I have done tests to completely isolate noise (including ground) on the USB receiver side of my devices but still found that playback software made an audible difference. It's possible that I didn't fully isolate the PC side but my preliminary conclusion was that something else was getting through & it had to be in the USB signal from the PC. The only thing that seemed to qualify was signal timing - USB packet jitter, for instance. Of course, noise gives rise to jitter!

sbgk wrote:I'm starting to think that playing music through a cpu is fundamentally flawed

This seems like a better solution http://www.the-ear.net/review-hardware/ ... mer-server

But most options still use processors of some sort so the question is still what are the fundamental factors that affect audibility of computer (for want of a better term) audio? Only when we answer this will we be able to use informed choices about the characteristics necessary for such processors.

My experience with JPLAY has only been on a single PC - turning off the network card made such a dramatic improvement to sound quality I never wanted to try a dual PC.

JPLAY 5.2 beta 14 & 15 is a major upgrade in SQ imo - I can't compare it to MQN as I am hoping for a 24bit version sometime soon but the Jplay 5.2 betas are much better.

Also, if using a Linear power supply I've found that slowing down the RAM timings makes a significant improvement, whereas the opposite is true with a switched mode power supply - which sounds better with over-clocked faster RAM timings. The differences are amazing - a good linear power supply sounds worse to me that the switch mode one if RAM speed set to fast!

It never ceases to amaze me, computer audio.

Clive

jkeny wrote:

sbgk wrote:I'm starting to think that playing music through a cpu is fundamentally flawed

This seems like a better solution http://www.the-ear.net/review-hardware/ ... mer-server

But most options still use processors of some sort so the question is still what are the fundamental factors that affect audibility of computer (for want of a better term) audio? Only when we answer this will we be able to use informed choices about the characteristics necessary for such processors.

Copying data through the cpu seems to be a bad thing, kernel streaming has some options that allow memory mapping between source and destination, but they can't be used with usb devices as they require memory copying. So maybe the NAD solution has memory which is loaded with the data and then dma'd to the destination.

Clive101 wrote:My experience with JPLAY has only been on a single PC - turning off the network card made such a dramatic improvement to sound quality I never wanted to try a dual PC.

JPLAY 5.2 beta 14 & 15 is a major upgrade in SQ imo - I can't compare it to MQN as I am hoping for a 24bit version sometime soon but the Jplay 5.2 betas are much better.

Also, if using a Linear power supply I've found that slowing down the RAM timings makes a significant improvement, whereas the opposite is true with a switched mode power supply - which sounds better with over-clocked faster RAM timings. The differences are amazing - a good linear power supply sounds worse to me that the switch mode one if RAM speed set to fast!

It never ceases to amaze me, computer audio.

Clive

i suspect underclocking cpu and ram frequency only works well with cleanly powered pc's

which makes your findings interesting
what was the ram settings you compared between smps and linear?
might explain some varying results with bios settings

nige2000 wrote:

Clive101 wrote:My experience with JPLAY has only been on a single PC - turning off the network card made such a dramatic improvement to sound quality I never wanted to try a dual PC.

JPLAY 5.2 beta 14 & 15 is a major upgrade in SQ imo - I can't compare it to MQN as I am hoping for a 24bit version sometime soon but the Jplay 5.2 betas are much better.

Also, if using a Linear power supply I've found that slowing down the RAM timings makes a significant improvement, whereas the opposite is true with a switched mode power supply - which sounds better with over-clocked faster RAM timings. The differences are amazing - a good linear power supply sounds worse to me that the switch mode one if RAM speed set to fast!

It never ceases to amaze me, computer audio.

Clive

i suspect underclocking cpu and ram frequency only works well with cleanly powered pc's

which makes your findings interesting
what was the ram settings you compared between smps and linear?
might explain some varying results with bios settings

From my recollection (not in front of PC at the moment) using ASUS motherboard the RAM timings that sounded best with a SMPS were 6,6,6,20,1 running at 833 Mhz (the fastest my system would allow and actually sounded very good) but using the Linear power supply at these settings it sounded as if Linear Power supply not run in, not good at all - dry and strained sounding. Using 12,12,12,25,1 and RAM at 1333 Mhz for the LPS gives alot more vibration, detail and depth, everything sounding spot on, but the SMPS sounded awful.

Just shows you can't go by one setting suits all.

Clive

6 6 6 20 sounds like a incorrect timing. no ram is rated as bad as that. you better try 6 6 6 23, or 6 6 6 22.

12 12 12 25 is also crazy. if you like 12 12 12 25, you need to try 12 12 12 23. 25 is unnatural sounding if you understand how ram is rated.