Building a DIY Music Server

[MarcelNL]

15 minutes ago, Tatomek7 said:

I will repeat what @Nenon already said, wiring matters a lot, so consider to do that once and good. I know it's difficult to not check immediately, how it will sound, but if you will change wiring after, you will go through burn in process twice. BTW DC connection to Taiko ATX also impacts on sound ( a lot ). I just changed that few days ago and I was surprised, how big impact it has. 

 

 

I already had Vcaps ( x 3 ), KBG ( x 4 or 5 ) Mundorf Supreme SGO ( x 2 ) and Miflex KPCU ( x2 ). Each of them impacts on the final sound signature. VCaps gives lot of space, details, micro information, Miflex gives more bottom and reduces micro information. So, rolling them you can voice the ULPS to your liking. At the moment I have 1Vcap, 1 KBG, 1 Mundorf and 2 x Miflex. Still no 100% my cup of tea.

 

Duly noted! And I agree with what you say, it is just that I want to save myself the trouble of doing yet another full wire harness once I make the final case for the beast...the last Cap has three wires (6 actually)into the  Taiko ATX.

 

Bypassing will be a task for after burn in and I know what that can do from previous experience building and using a Le Monstre amp...I've got a pretty wide range of potentials available and testing will be extensive. (if only a computer would be less sensitive so I could swap caps while listening (using some alligator clips), powering down, listening, burn-in, listening, power down, next iteration....time consuming!

[MarcelNL]

BTW; did you try a Jupiter beeswax cap for bypass?

[Nenon]

2 hours ago, MarcelNL said:

Bypassing will be a task for after burn in and I know what that can do from previous experience building and using a Le Monstre amp...

 

This is different than an amp power supply.

As you know the Taiko ATX uses GaNFets that switch at a very high frequency. Bypass caps are needed to reduce the output impedance at high frequencies. Without them, the Taiko ATX may see too high impedance and not start. I can't predict if that will happen or not in your particular version, but just FYI. They are a part of the design, not just a voicing tool. 

[MarcelNL]

I was not aware of that fact, thanks!

 

 

[MarcelNL]

I have lift off, not listened to the system yet but Daphile works and is playing music...first listening will be done during an audio gathering with some friends, which is good as I'm deaf in one ear with a constant high pitched tone due to a cold and won't hear any difference anyway. Listening to music last night gave me the impression things might be sounding good (standard PSU) but that was all....

[MarcelNL]

First impression is : WOW....one good ear and all taken in to account...

[Exocer]

1 hour ago, MarcelNL said:

First impression is : WOW....one good ear and all taken in to account...

Wait till you have two fully functional ears again! I'm long overdue for an ear cleaning methinks. Wondering how much of an SQ improvement to expect with all the wax buildup gone 😂.

 

In other news, things have been a bit stagnant on my end with regards to the server. I've had the good fortune of being able to sit back and enjoy the tunes. Every once in a while the desire to tweek comes back, but, I usually come to my senses once listening resumes. For those on the fence building this design, just do it. You're in for a treat. The power to your server is of utmost importance when striving for the best SQ you can extract from your system.

 

Worth mentioning, I am probably going to build up a separate LPS using the unused 19V rail of my SR7T and put it up for sale in the coming months. I no longer need that rail to power my server so some lucky AS'er will be able to pick that up at a reasonable price.

 

In recent news, I built myself a new DC input cable using solid core Neotech UP-OCC of 16awg, gold plated molex pins (as usual), gold plated pure copper lugs. I braided it for "neatness" and do not know if there are any sonic benefits in braiding this way but i've done this with my EPS cables as well just to keep a neat look:

 

This is my first experience with this wire in this function, but what I will say is it sounded a bit thick at first compared to the stranded 16awg wire it replaced. It eventually sounded more refined, with improved imaging/separation. Is it a night and day difference over the Oyaide 16awg wire it replaced? No. Am I glad I tried it? Yeah, it satisfied my curiosity.

 

Next up will be Mundorf SG but with expenses adding up I do not expect to have that completed until next year.

 

Happy holidays all, and most importantly...happy listening!

 

Cheers,

-Rob

 

 

[oneguy]

About how much thermal paste do you guys estimate I’ll need for the CPU and the heatpipes?

[Soul Analogue]

On 12/11/2021 at 3:06 AM, Nenon said:

 

This is different than an amp power supply.

As you know the Taiko ATX uses GaNFets that switch at a very high frequency. Bypass caps are needed to reduce the output impedance at high frequencies. Without them, the Taiko ATX may see too high impedance and not start. I can't predict if that will happen or not in your particular version, but just FYI. They are a part of the design, not just a voicing tool. 

Yes.. i have a customer of ulps having issues powering up the pc, most likely due to this reason

i did not put bypass caps in the PS

this incompatibility issue has been the first i encountered 

What strange to me that this “Requirement” had not been mentioned officially by Taiko … but only by Nenon in this thread… maybe i missed the info somewhere?

ps. i do not like to add bypass caps to CLC filters, as they could mess up with frequency coherence and sound weird if not properly implemented and tested

 

it would be another story if they are necessary for proper functioning of the system

[MarcelNL]

10 hours ago, oneguy said:

About how much thermal paste do you guys estimate I’ll need for the CPU and the heatpipes?

I used one full syringe of 10g for my build, you may need a bit more

[MarcelNL]

So the feedback was that it sounded ridiculously good, not as digital was supposed to sound at all ;-)  (3 Vinyl enthusiasts can be a difficult crowd). Today we learned that the neighbours also thought it sounded good.

Hope to regain my hearing this week as I'm very curious, the bright side is that burn-in can happen during my bench penalty.

 

The few fragments when I was able to hear a bit sounded promising with loads of overtones and great tonal separation, one remark was "I can hear much deeper into the music, with many details I never knew were there but they now stand out"

I heard effortless dynamics without hints of glare that were still present with the previous PSU(Corsair AX1200i)

 

Listening was done on two open baffles with KL405 fullrange units, an Audio Note 300B amp that is a nice amp but nothing special, we briefly rolled some tubes to quickly revert to the original mercury vapor rectifier after testing a Russian 'komrad' of the original.

[Nenon]

15 hours ago, Soul Analogue said:

What strange to me that this “Requirement” had not been mentioned officially by Taiko … but only by Nenon in this thread… maybe i missed the info somewhere?

 

As far as I know this is the only place talking about the Taiko ATX. Or at least I haven't seen other places. It has been mentioned a number of times that low impedance of the power supply is one of the most important aspects of the ULPS. I have also mentioned quite a few times that having the right capacitance is important. Some people disagree with me, and that's all fine.

 

The Taiko ATX is a DIY product, so I assume people who are designing and selling power supplies for it would have tested it in different scenarios and would have discussed the design with Emile first. The v2 and v3 I posted about were certainly "approved" by Emile before I shared publicly. 

[oneguy]

5 hours ago, MarcelNL said:

So the feedback was that it sounded ridiculously good, not as digital was supposed to sound at all ;-)  (3 Vinyl enthusiasts can be a difficult crowd). Today we learned that the neighbours also thought it sounded good.

Hope to regain my hearing this week as I'm very curious, the bright side is that burn-in can happen during my bench penalty.

 

The few fragments when I was able to hear a bit sounded promising with loads of overtones and great tonal separation, one remark was "I can hear much deeper into the music, with many details I never knew were there but they now stand out"

I heard effortless dynamics without hints of glare that were still present with the previous PSU(Corsair AX1200i)

 

Listening was done on two open baffles with KL405 fullrange units, an Audio Note 300B amp that is a nice amp but nothing special, we briefly rolled some tubes to quickly revert to the original mercury vapor rectifier after testing a Russian 'komrad' of the original.

I’m happy this is getting better and better and I hope your neighbors were fine with being able to hear it through the walls ;)

[Rovo]

Just to show you my version of Nenon ULPS v3. A prototype in Multiplex. Now that it all fits I can start making it from aluminium. The PC is used as a HQPlayer NAA endpoint. 

 

[Rovo]

And yes it was worth the time and effort (and money 😁)

[MarcelNL]

Looks great!

 

Why Aluminium? My experience is that gear sounds better in a non Metal case...I plan on using Panzerholz with EMI treatment on the inside to build a case, PSU in the bottom and MB on top.

[Nenon]

5 hours ago, Rovo said:

Just to show you my version of Nenon ULPS v3. A prototype in Multiplex. Now that it all fits I can start making it from aluminium. The PC is used as a HQPlayer NAA endpoint. 

 

Looks very good. The Extreme has some serious competition :). At this level, you may also consider using some copper instead of aluminum. Please share the final photos. Great idea to build a prototype from wood and use aluminum (and maybe some copper) later. 

 

 

4 hours ago, MarcelNL said:

I plan on using Panzerholz with EMI treatment on the inside to build a case, PSU in the bottom and MB on top.

Be careful with that EMI absorbing stuff. It sucks the life out of the music when too much is used. And too much was not much at all in my experience. 

 

 

[oneguy]

Very clean looking build!

[MarcelNL]

1 hour ago, Nenon said:

 

Be careful with that EMI absorbing stuff. It sucks the life out of the music when too much is used. And too much was not much at all in my experience. 

We auditioned a ton of different EMI blockers, and found one that works well and keeps the sound alive! Agree that it can take the life out of the music easily

1 hour ago, Nenon said:

 

 

[Nenon]

21 minutes ago, MarcelNL said:

We auditioned a ton of different EMI blockers, and found one that works well and keeps the sound alive! Agree that it can take the life out of the music easily

Do you mind sharing which one? 

[Exocer]

10 hours ago, Rovo said:

Just to show you my version of Nenon ULPS v3. A prototype in Multiplex. Now that it all fits I can start making it from aluminium. The PC is used as a HQPlayer NAA endpoint. 

 

Most definitely striving for this level of excellent! Well done. 

 

Yours and other more recent builds have convinced me that going with 2x47,000uF + 1x22,000uF caps is the way to go for V3 from a space savings POV. That being said I would like to understand more about your efforts with the chassis. Wonderful job! 

 

Note: I will be comparing SQ differences between 5x22,000uF and 2x47,000uF+1x22,000uF for those interested. It will take a while though as these caps take forever to break in.

 

Having one less massive chassis, shorter cabling between LPS/Taiko/Motherboard are beneficial from a space savings AND SQ perspective.

 

Well done!

-Rob

[oneguy]

On 12/10/2021 at 8:29 AM, elan120 said:

Have you check with Larry from HDPlex to see if HDPlex H5 case can physically accept the Gigabyte Z690 Aorus Master motherboard you plan to use?  According to HDPlex website, H5 case can accept ATX, mATX, and mITX,, but not EATX.  Since Gigabyte Z690 Aorus Master motherboard is an EATX motherboard, you might want to confirm with Larry to make sure.

 

 

 

Turns out it’s smaller than expected. 

[Rovo]

Thank you all for the compliments. It is nice to know that your work is appreciated. 

 

@Nenon

Temperature wise there is no real need for copper, the cooling block is big enough and the load of the CPU is relatively low. Or do you have other experiences, other advantages?

 

@Exocer

What kind of information are you interested in? Packaging is not easy as you perhaps can see from the pictures. 

 

[MarcelNL]

@oneguy Does the MB only have 3 PCIe slots but 5 (!) NVME slots?

 

[oneguy]

@MarcelNLyour eyes do not deceive you. Only 3 PCIe slots. One will be for the Optane drive so that only leaves two not spoken for.