Building a DIY Music Server

[MarcelNL]

13 hours ago, AudioDoctor said:

a water cooled CPU and GPU, 5800X and 3080

I was triggered by the 3080, my use case is different. No upsampling processing or DSP so I have an ancient video card that gets shut down by Daphile after booting.

[AudioDoctor]

31 minutes ago, MarcelNL said:

I was triggered by the 3080, my use case is different. No upsampling processing or DSP so I have an ancient video card that gets shut down by Daphile after booting.

 

HQPlayer uses Nvidia CUDA software to offload processing to make it possible too upsample to higher rates than using CPU alone. It won't be used for video output at all.

[Jud]

Too bad about Bitcoin.  GPU prices used to be somewhat reasonable, now one 3080 costs more than my last PC upgrade (ASUS mobo, Intel CPU, nVidia GPU, RAM, water cooling).

[AudioDoctor]

48 minutes ago, Jud said:

Too bad about Bitcoin.  GPU prices used to be somewhat reasonable, now one 3080 costs more than my last PC upgrade (ASUS mobo, Intel CPU, nVidia GPU, RAM, water cooling).

Yeah, its ridiculous.

[Tatomek7]

11 hours ago, elan120 said:

Are the other versions available that offers more?

If you will go back to page 121, you will see, I’ve changed in my ULPS Hammond choke for a custom made one. It was the only change I did.

After two weeks after implementation, I can say, every aspect of music presentation improved; soundstage is deeper, there is much more air around instruments. Localisation is more precise, bass is more differentiated, there is more details and resolution.

Negatives? Yes, I can hear difference between formats and unfortunately I started to prefer high resolution recordings, unfortunately there is much bigger difference between Euphony and Roon and as I really like Roon Radio, what allows me to discover new music, Euphony sounds much better.

 

Yesterday I’ve changed server from X99 MB to C621 ( same Apacer Rams, Jcat USB and Nic cards, same LPS to power cards and MB ) and unfortunately sound was not as engaging, as it was with X99. Probably I will have to change ATX and EPS wiring, in X99 it's Neotech and in C621 it's taken from HdPlex 800W. Before the change in ULPS I didn’t hear any major differences to force me to do more improvements. Of course it’s my guess, as it could be just different MoBo, but before this change, C621 was better even with power supply Corsair 1200i. So, I encourage everybody to check, if you have somewhere around someone, who can made you a choke. Price is not very high , but the difference is so big, it’s worth to experiment.

[MarcelNL]

Did anyone try various rectification methods, single side phase, double phase, double phase double side.....I know it takes a specific transformer each to end up at the same voltage going into the Taiko...so it's unlikely, but still...? 

[elan120]

6 hours ago, Tatomek7 said:

If you will go back to page 121, you will see, I’ve changed in my ULPS Hammond choke for a custom made one. It was the only change I did.

After two weeks after implementation, I can say, every aspect of music presentation improved; soundstage is deeper, there is much more air around instruments. Localisation is more precise, bass is more differentiated, there is more details and resolution.

Negatives? Yes, I can hear difference between formats and unfortunately I started to prefer high resolution recordings, unfortunately there is much bigger difference between Euphony and Roon and as I really like Roon Radio, what allows me to discover new music, Euphony sounds much better.

 

Yesterday I’ve changed server from X99 MB to C621 ( same Apacer Rams, Jcat USB and Nic cards, same LPS to power cards and MB ) and unfortunately sound was not as engaging, as it was with X99. Probably I will have to change ATX and EPS wiring, in X99 it's Neotech and in C621 it's taken from HdPlex 800W. Before the change in ULPS I didn’t hear any major differences to force me to do more improvements. Of course it’s my guess, as it could be just different MoBo, but before this change, C621 was better even with power supply Corsair 1200i. So, I encourage everybody to check, if you have somewhere around someone, who can made you a choke. Price is not very high , but the difference is so big, it’s worth to experiment.

I did miss your earlier post, and it is very interesting to know a custom made chock with same spec as Hammond can improve the sound so much.  Have you get to try adding the second chock yet?  Wonder if there are more changes or more improvements as single vs. dual Hammond chocks.

[ciccio1112]

On 8/17/2021 at 3:02 AM, Nenon said:

 

Goodmorning everyone.

 

@Nenon. Do we have any news about the Taiko case, and active mosfet rectification and GaN regulators for the unregulated power supply? I am always waiting for your news because I would like to complete the Taiko By Nenon with double Xenon. Thanks.

[Tatomek7]

2 hours ago, elan120 said:

Wonder if there are more changes or more improvements as single vs. dual Hammond chocks.

I will check it later. I have the second custom made choke on my shelf, but I will try it later. Firstly I would like to change the EPS wiring. 

[elan120]

37 minutes ago, Tatomek7 said:

I will check it later. I have the second custom made choke on my shelf, but I will try it later. Firstly I would like to change the EPS wiring. 

Looking forward to your test result when you get the chance to install the second choke.  I am considering trying Lundahl LL2773 and see if that will be an improvement.

[Nenon]

Just a small warning on chokes - if the choke is not right for the application some side effects like ringing and overshooting can happen. Plus, as I've already said many times, you are risking of killing the Taiko ATX. As a general rule, if you don't know how to do a proper simulation, stick to the Hammond. 

A common mistake I've done in the past is to perceive the extended treble caused by some of these effects as better sounding. There are better chokes than the Hammond, but the Hammond sounds very neutral to me... which means it will work universally in most systems. Some of the other chokes I've heard were colored and might be better in one system but not so good in another. Just something to keep in mind. Having said that, I would like to try the choke @Tatomek7 sourced if possible.

 

2 hours ago, ciccio1112 said:

@Nenon. Do we have any news about the Taiko case, and active mosfet rectification and GaN regulators for the unregulated power supply?

 

Taiko is setting up their new CNC machine:

source: https://whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-401#post-742178

 

I expect the DIY chassis to be their first project. 

A lot of people have good success with the SALIGNY Standard rectifier - https://evotronix.eu/main/

[Nenon]

50 minutes ago, elan120 said:

Looking forward to your test result when you get the chance to install the second choke.  I am considering trying Lundahl LL2773 and see if that will be an improvement.

 

Per Emile: "The ll2773 with 66.000uF will kill the dc-dc atx on powerup".

[MarcelNL]

I'm intrigued by the Saligny rectifier, easy enough to compare with my goto schottky's so I ordered one.

[MarcelNL]

3 minutes ago, Nenon said:

 

Per Emile: "The ll2773 with 66.000uF will kill the dc-dc atx on powerup".

 

what if you put a switch after the LCL network and powerup that part via a soft start? Did he say why the ATX is killed?

 

[Mags]

13 minutes ago, MarcelNL said:

 

what if you put a switch after the LCL network and powerup that part via a soft start? Did he say why the ATX is killed?

 

Put a bleeder resistor across the cap to ensure the choke input supply always draws more than the critical load current, thus remaining in choke input mode.

 

http://www.r-type.org/articles/art-144.htm

[MarcelNL]

For now I can add to rhe confusion on two fronts;

1-taking the MB out of the metal case and putting it on some non specified wood chip board (underlayment) opens the sound

2-Using a Corning LCLC (no filter ;-) ) fiberoptic cable of the same length as my previous (probably) plastic core optical cable improves detail and nuance in the sound when streaming music through it,

[guiltyboxswapper]

Time to let the cat out of the bag. 

 

For a while I've been using some audio grade SMPS from Micro Audio.  Prompted by Nenon's reference to Mundorf AG caps, back in December 2020 I purchased some Mundorf AG 80v 10,000uf caps, and replaced the ELNA AG caps that were on the provided capbank, using the same PSU and setup in my amps (Purifi IET400A's using Sparkos SS2590 Pros).  To my surprise I experienced one of the biggest (and cheapest!) upgrades that I'd heard in a long, long time.

 

Then came the Taiko DC DC allowing all sorts of PSUs.  So I tried his off the shelf "class A SMPS" with Elna first, and ... blew it up by accident (mea culpa, I shorted it) taking out the RCD tripper and creating some fireworks in the process (see below!).

 

 

In this moment of madness, instead of asking for a replacement I asked if he could do a few small refinements namely move the CapBank PCB onto the same PCB that accepted 6x 30mm caps, and ensure we have nice juicy PCB traces for ultra low impedance from the same unregulated SMPS circuit.

 

That revision populated with 6x 10000uf 63v Mundorf AGs was an astounding success, leaving behind my TDK Lambda HWS 1500 (a regulated industrial SMPS capable of 70a peak) no problem with a SQ upgrade from the first 30mins.  In other words, Nenon and co are right, GTFO off regulated PSUs when it comes to very high current demand systems!!

 

Being a zero current switcher even with my T+A SDV HV 3100 asking for DSD1024 from the Ryzen 5900x, it stayed as cool as a cucumber under load too (wall draw of 110w).  It nicely fits into a PC chassis too.

 

 

Amazed with the results, I then asked what if we did the following:

- more output caps, from 60,000uf to 165,000uf

- lowered switching frequency to 55khz

- changed the AC filtering caps to Mundorf AG too (1360uf in total)

- added bypass cap to the output (I chose V-CAP, but there's some tiny ceramic ones mounted on the PCB too)

- gold plated PCB

 

Tbh I've forgot what else we change....

 

So here comes revision 3 (as pictured below) and once again within 30mins, it blew the doors off revision 2 (above) straight from the start.  Same compact size, same cool running etc.  

 

 

Will come back to yourselves in the next week or so once revision 3 is broken in, but this to me has been a massive revelation.

[MarcelNL]

Interesting development! My main question is; did you compare to a linear PSU and at what cost level did you arrive with V3, I suppose all that goodness cannot come cheap ? 

[NEMO777]

15 hours ago, Nenon said:

 

По Эмилю: "ll2773 с 66 000 мкФ убьет atx постоянного тока при включении питания".

Can you explain why LL 2773 will kill ATX? After all, Taiko himself, judging by the photographs of his unregulated power source, does not use any current-limiting components of the ATX relay. Apparently, Taiko himself uses LL2773 made according to his specification. I compared the original LL2773 and a photograph of the Taiko TA 350 throttle printed in the original size. I did not find any differences in the dimensions and cross-section of the wire and the number of turns.

[guiltyboxswapper]

54 minutes ago, MarcelNL said:

Interesting development! My main question is; did you compare to a linear PSU and at what cost level did you arrive with V3, I suppose all that goodness cannot come cheap ? 

 

Sub $550 or less with a group buy of kinds, as you correctly pointed out some of those components are not cheap... 

 

You can ofcourse use HDPlex 800w or Taiko DC DC (I've been testing purely with Taiko tbh).

 

As Nenon has said short cables make a massive difference, but being so compact you can place it right next to the DC DC and stay under 10cm cables between the two.

[sakso136]

Very very interesting,specially if you can place it next to the taiko atx- dc.

Waiting for your findings!

[guiltyboxswapper]

15 minutes ago, sakso136 said:

Very very interesting,specially if you can place it next to the taiko atx- dc.

Agreed, I was getting fed up with the ridiculous amount of boxes and wires I was accumulating to make this "happen" hence this development to begin with. 

 

In the end mind, I ended up purchasing a network attached DAC (https://www.ta-hifi.de/en/audiosystems/hv-series/sdv-3100-reference-streaming-dac-preamp/) and placing the server 30m+ away from the hifi itself (which as you can imagine brought improvements alone isolating it from the energy from the speakers...)

[sakso136]

2 minutes ago, guiltyboxswapper said:

Agreed, I was getting fed up with the ridiculous amount of boxes and wires I was accumulating to make this "happen" hence this development to begin with. 

 

In the end mind, I ended up purchasing a network attached DAC (https://www.ta-hifi.de/en/audiosystems/hv-series/sdv-3100-reference-streaming-dac-preamp/) and placing the server 30m+ away from the hifi itself (which as you can imagine brought improvements alone isolating it from the energy from the speakers...)

Very nice.

I would love to know your results after burn in and it compares to nenon v2  model.

My taiko atx-dc is fed by keces p8 19 v now,and it is an upgrade from the hd800.

Regards

[guiltyboxswapper]

3 minutes ago, sakso136 said:

I would love to know your results after burn in and it compares to nenon v2  model.

 

Me too, but I have no way of getting a hold of v2 nenon at this point - which is why I started all this to begin with. 

 

Plus didnt want yet another box/chassis and all the cables associated with that.  I did plan to compare with my own unregulated linear build, but the performance from this has been so ... surprising ... that I may not bother now.  

 

Yes that Taiko DC DC whilst seems pricey, is strong VFM IMHO too.

[sakso136]

1 minute ago, guiltyboxswapper said:

 

Me too, but I have no way of getting a hold of v2 nenon at this point - which is why I started all this to begin with.

 

Yes that Taiko DC DC whilst seems pricey, is strong VFM IMHO too.

Good job you ve done! Congrats!

I probably missed the specs of that ps,can you state again how much is the voltage output and amperes?

Thks