PC upgrade - why does it sound better?

Posted by: mike k burke on 06 January 2011

OK, i had a bit of spare cash and decided to see if i could make my music PC sound better as it was only sounding moderately better than my general purpose quad core PC. I figured I could always use the bits and improve my main PC if didn't make a difference.

Bearing in mind all the discussions re jitter/streaming etc I would like members opinions on why the following changes work, as they undoubtedly do.

Firstly the things that stayed the same:-
Power supply+ case
ESI juli@ sound card
Storage disc for WAV's (SATA2)
Mouse using PS2 input (all USB is disabled through BIOS by the way)
W7 Operating system and music playing software (CMP2)
optical cable and coax cables both connected to the nDac

Changed
Operating system drive from Sata2 to SSD
Ram 512MB DDR2 to 1 GB DDR3 (couldn't get DDR3 512 at the time)
Processor & motherboard from old AMD 3200/socket 939 to intel i3/socket 1136 m/board
Bios settings on the motherboards are different but both were underclocked as much as possible to reduce power consumption

Basically following cmp2 methodology for those who are aware of it.

Result suprised me by being much better particularly in the Bass where it is notably more tuneful and, for want of a better word, forceful. The other thing that seems to have altered is my preference had been for the optical connection (just), it has now swung the other way to the coax.

So as i understand it from other discussions i haven't changed my spdif clock/sound card, power supply or even the cables in use. Why does it sound better?

Posted on: 06 January 2011 by james n

Whilst you haven't physically changed the power supply you are running a different processor board, so demand on the supply will be different and so will the spectrum of noise on the internal supply rails. This will have an effect on the internal clocking and ultimately the quality of the resulting s/pdif output. I'd also suggest that both conducted and radiated emissions wil have changed too and this will affect different systems in different ways depending upon setup.

James

Posted on: 06 January 2011 by Jack

Pure guess work......

- less noise from HDD access after switching to SSD affecing SQ?
- more memory therefore less disk access to HDD?

The above guesses are on the assumption that noise is causing sound degrading artefacts (e.g. jitter) - although the nDAC design is supposed to deal with that so not sure that makes sense? Sometimes logical explanation doesn't seem possible

I would be interested to know how much of the XP config tweaks documented for CMP2 you applied to W7 and which worked and which didn't. Incidently are you using ASIO or WASAPI?

Posted on: 06 January 2011 by js

512 should have been enough for his use with the seperate sound card and no display so I doubt he was going virtual. In fact the extra RAM could represent a larger PS load which is probably wht he used 512 to begin with.

Unfortunately bits sound different. Found that out with firewire chipsets. Many also prefer the sound of SS drives and as James has said there are other factors also. Lack of vibration from the hard drive may also have an effect. This is exactly why I prefer ready made by companies that already bothered to check and even make custom bits but nothing wrong with having some fun and good tunes along with it.

It would be interesting to test a beefy linear supply but they also vary substancially.

Posted on: 06 January 2011 by mike k burke

Hi James/Jack

I choose the i3 as it is the latest low power chip with on chip graphics. The one thing i have noticed is the cpu operating temperature was 40-45 degrees now is barely above room temperature. Damn, I could have saved £20 on the big CPU cooler!

Only other physical difference thinking about it is the VGA is now onboard, again reducing load.

As to the tweaks with CMP2, most work although have had a few problems when it comes to the autoruns tweaks, particularly winsock related. I think this is one of the areas w7 differs from XP quite considerably so no suprise.

Registry tweaks all work ok but not convinced they make much difference.

Most of the features you turn off are just in different places so it is just a case of finding them.

Kernel mods are i believe non applicable to W7 as some are now standard and others don't exist

Also have now only got about 20 services still running to minimise the system latency when measured with DPCLAT (this doesn't get much coverage on the cmp2 site but lots of info elsewhere, particularly with people building PC's to record music)Key one is to kill the user profile service. This at least gets close to using minlogon on XP as i understand it.

Interestingly the latency was slightly better with the AMD setup (12-15microsec vs 15-20)

Have been using Asio4all but have now got the Juli@ Asio driver working with the coax spdif which now allows me to upsample to 96K and sounds better for it

Posted on: 06 January 2011 by Simon-in-Suffolk

Hi- well my thoughts are probably the biggest difference is caused by the change in motherboard. The motherboard manages some regulation and decoupling on the power lines. You might find your Intel motherboard has better decoupling and noise filtering. A noisy power line will find it's way into the digital link and pass along as RFI and pollute your DAC.
The fact you say the coax is now superior than before points to this as well in my opinion
Simon

Posted on: 06 January 2011 by Jack

When I built my XP machine I followed most of the tweaks on CMP with the exception of the specific hardware and disabling of networking - the DPC latency runs at 6 microseconds.

I've recently made the machine multi-boot with W7 and done some tweaks but not all the XP recommended settings from CMP - cant get the DPC latency below 145 microseconds although plays without any problems.

Will try some of the other CMP settings when I've sort my current J River problem!

Posted on: 07 January 2011 by js

Jack, are you going Wasapi sans kernel or asio? Latency isn't necessarily a bad thing in straight playback but sometimes it will come down if you go wasapi exclusive and forgo the others which were favored in XP. Sometimes there's no option. Don't mean to tell you things that you may already know.

Posted on: 07 January 2011 by Jack

I'm using WASPI but only because I have been experiencing some problems using ASIO with high res files.....I seem to think ASIO sounds better?

I have set it so the audio application can take exclusive control of the audio interface. I've also set the performance priority to background rather than programs

Posted on: 07 January 2011 by js

Interesting. Interfaces all sound different and if it's very slightly 'glassy' that yu're hearing, that was my take on the program's character. Here's a wierd thing to try. I found I preferred JR with the fade setting in and fast. Makes no sense at all as I generally take everything out when testing but it came up because we were trying to eliminate a tic in a customers setup and it was better for him. Didn't fix the click. There's probably a buffer involved.

Posted on: 07 January 2011 by mike k burke

Hi Jack,

When I first put together the rebuilt PC the DPC latency was around 90-120micro sec compared to my main PC which is between 100-150. I disabled USB from the outset and I know from before this does make a difference.

Stopping most of the unneeded processes reduced it to around 70-90

Changing the memory frequency/timing as per cmp2 website reduced it to around 20-25

Stopping the user profile service which i left to last for convenience knocks it down to 15-20

hope this helps

Posted on: 07 January 2011 by mike k burke

quote:
Originally posted by Simon-in-Suffolk:
Hi- well my thoughts are probably the biggest difference is caused by the change in motherboard. The motherboard manages some regulation and decoupling on the power lines. You might find your Intel motherboard has better decoupling and noise filtering. A noisy power line will find it's way into the digital link and pass along as RFI and pollute your DAC.
The fact you say the coax is now superior than before points to this as well in my opinion
Simon

Hi Simon,

If what you say is correct then this bears out why other people on other forums are using linear power supplies or dual switched supplies as well as some even powering the juli@ independently.

Problem is this takes messing around with easily available std parts into the realms of DIY tinkering and I am not sure my basic electrics/electronics is up to it. Maybe have to get better at reading motherboard specs!

Posted on: 07 January 2011 by mike k burke

or maybe Naim end up making a power supply for a PC.....any name proposals?

Posted on: 07 January 2011 by js

Phonosophie (sp) ps.

Posted on: 07 January 2011 by Simon-in-Suffolk

quote:
Originally posted by mike k burke:
Problem is this takes messing around with easily available std parts into the realms of DIY tinkering and I am not sure my basic electrics/electronics is up to it. Maybe have to get better at reading motherboard specs!

Mike, I suspect as you found with modular components that you have in PCs there is a lot of trial and error, and that is where the Market is created for specialist audio pc's etc, where someone has done that trial and error already. I doubt whether motherboard specifications will give you the info you are looking for.

BTW a clean power supply will do little to help poor regulation or poor decoupling on the motherboard itself.

Fun isn't it...

Simon