Build a dedicated Audio PC

[jkeny]

Welcome back, Pearse - we've been busy little bunnies while you've been away enjoying yourself in LA :)
Great with the laptop but it won't play for long with the fan disconnected.

[Fran]

So in the interests of keeping things on track, here are the main questions from earlier:

1. Is fanless really necessary? Apart from the noise of the fan itself, is there any effect on SQ

Yes, you should use either fanless, or else run the fan from a separate supply

2. Is there some kind of magic in the DH77EB that Chris Connaker got lucky on or will other mobos work OK (I realise that the ease of passive cooling on the DH77EB maybe be important if 1 is shown to be true).

3. Do we really need an i7 - would i5 or even i3 affect SQ negatively.

4. Importance of picoPSU/linear supply?

Hopefully find this out next week if it works out with nige2000



So at least there is a plan for 2 out of the 4 questions.

The other 2: CPU and DH77EB. Anyone got any real comparisons on this?

Pearse - did I see somewhere that you had previously done a dedicated PC build? What board did you use for that?

From the computer audio nuts who are hanging around this thread, have you guys heard differences using different computers/laptops with the same spec? Eg John - you have 2 different laptops. Have you ever heard a difference depending on which one you used for the audioPC - or Tony?

I suppose the premise I'm working towards here is that if you used any mobo + cpu but implemented picoPSU, fanless & SSD would they all sound the same?

Fran

[Sligolad]

Fran, the build I last put together was back in 2010 using an Intel Core i7 980X 3.33G 12MB on an ASUS RAMPAGE III EXTREME MOBO, all in a passive Zalman TNN 500 case.
It is a bit of a monster to be lugging around and I use it for all AV and Sat stuff but did use it wirelessly initially for 2 PC setup when JPlay went to V5 and it worked well.
It would be too difficult to roll it out for comparison and it is at the extreme end of performance with a six core extreme cpu and Mobo it still a bit OTT for a dedicated JPlay build.

Given the Mobo suggested by Chris Connaker costs only $100 it might be worth going with this even without the ability to tweak to much in bios. When you get into tweaking bios it could definitely turn a lot of people away from PC based audio.

The processor is the tough one and it would be nice to hear comparisons but given the way JPlay is going and people finding i7 processors running at 50% load when loading tracks in JPlay I would err on an i5 as lowest spec.

[Fran]

Yes, I agree - there is no saving in moving to a different motherboard - in fact the opposite is probably more likely.

I have my eye on an i5 3470 at the moment, but also keeping an eye for i5 3335S = all >3ghz quad cores using 22nm.


Fran

[jkeny]

I think we have to approach these experiments systematically & be careful in judging processors SQ. I reckon that differences between CPUs may be subtle, if at all. In other words we have to deal with the major detriments to SQ first before we can hear if the CPU or other changes have any effect.

The major components that cause SQ degrade seem to me to be:
- fan (running off the main PS)
- HDD

Both of these are spinning devices & have varying current demands & this may be the reason they effect SQ.

I seem to remember that Marcin & others have claimed that even when running SSD, it improves SQ if run off a separate PS. Fran, have you tried this?

So the general rule seems to be to eliminate anything that causes power fluctuations to the main board. Could this be simply from PS fluctuating or from varying EMI/RFI (EMC?) or both? If it's just PS fluctuation then trying to stabilise this is the route to take. If it's just EMC emissions then using ways of shielding components from EMC would be beneficial. If it's both then ........

It makes me wonder if there is any other low hanging fruit that we can test with this in mind? In other words, is there anything else which can cause fluctuating PS to the motherboard - WiFi/network cards, graphic cards, etc. Another thing that springs to mind, does varying temperature cause similar fluctuations in the operation point of devices & varying current draws? I think that Marcin also recommends underclocking CPU & RAM.

As Fran asks, are there any other things to consider that we are missing? Any computer nut wish to chime in?

Fran, I never noticed any obvious difference between my different laptops but then I haven't set out to A/B test this so there may well be? Also, from what I said above the spinning fans & HDD in all my laptops may well mask any differences there could be between them?

[tony]

Ok back from my walk! Now I am not going to offer any scientific evidence but I think you have no option but to go i7 on a build. I spoke to John about this earlier but if you look at any of the serious computer audio heads on any of the forums the spec of the pcs are all at this level and most are way above what we are thinking will be enough. Running the zuma headless at the moment so cant check the ram on it but I bet it is 32gb.

My gut feeling is for stability and that extra special SQ it will need something that is more than fit for the job. The zuma spec is under no pressure barely gets warm.It is over engineered but that is what makes it work so well. Chris Connaker has done the hard work we just need to come up with the cheapest way of implementing it.

From what the jplay guys are saying and from talking with Pearse/Ken emi wadding/padding whatever is the next move. Personally I think it is getting to the point where it is going to be very hard to make any more real improvements. It will be back to the the green sata cable sounds better then the red one.

I think the zuma spec with a H5 case is the most reasonable bang for buck in that space. Maybe the memory type Nige identified which had a slower clock speed but you got twice the ram. The one item that I think is superflous is the most expensive bit (SOTM USB). But to confirm it I would need to be down to just testing that for awhile a/b. I cant hear any obvious difference.

Hopefully testing tomorrow night proves me wrong.

[Sligolad]

Another option for dedicated audio PC fanless

http://www.quietpc.com/sys-sidewinder

[jkeny]

The price on this core i7 CPU NUC-style (4" X 4") might be worth waiting for - launch in April, price not yet announced. http://www.engadget.com/2013/01/09/giga ... i7-due-in/

Aluminium case & better cooling, it seems?


Also Intel's own new NUC models are due to be announced in April also - core i5 & i7 models
Seems to be worth waiting to see what April brings (besides showers)

[mick]

Quiet pc do an NUC style mini pc:
http://www.quietpc.com/sys-ultranuc
Any thoughts?

[Fran]

After hearing the various bits at Tony's over the weekend, I went today and bought a picoPSU - its a no-brainer really. I found the cheapest place to go was ebay, but there may be other options out there. Logic supply is the cheapest but their shiping is ridiculous.


Anyway, another slight step on the way.

Fran