Jump to content
IGNORED

Building a DIY Music Server


Recommended Posts

23 minutes ago, Marcin_gps said:

TAIKO DC-ATX + ... ? vs OPTIMO ATX?

Ideally:

 

Taiko DC-ATX + DC4 vs Optimo ATX

Taiko DC-ATX + PHSR7T vs Optimo ATX

Taiko DC-ATX + Nenon's Unregulated Linear Power Supply vs Optimo ATX

 

I have the Taiko DC-ATX in hand and will try the with19v 10A rated rail of my SR7T soon to power my i9 server (not sure if it can handle the load but still worth a shot, as I do not overclock).

 

Edit: @Nenon that PSU is a work of art. Well done.

Link to comment
50 minutes ago, RickyV said:

Looking forward to your tests.

 

To clarify:

 

I can only test single 19v in vs 12v to EPS + 19v to DC-ATX  and both compared to the unregulated supply once it is built ( to be tested on a single core i9 9900k based machine).

 

The list below was in response to @Marcin_gps's question.

 

50 minutes ago, RickyV said:

Ideally:

 

Taiko DC-ATX + DC4 vs Optimo ATX

Taiko DC-ATX + PHSR7T vs Optimo ATX

Taiko DC-ATX + Nenon's Unregulated Linear Power Supply vs Optimo ATX

 

I have the Taiko DC-ATX in hand and will try the with19v 10A rated rail of my SR7T soon to power my i9 server (not sure if it can handle the load but still worth a shot, as I do not overclock).

 

Edit: @Nenon that PSU is a work of art. Well done.

 

At this point, I do think the Taiko DC-ATX is a no brainer purchase even for those with single core setups and good LPSUs such as the ones mentioned above.

Link to comment
33 minutes ago, MarcelNL said:

As much as I don't mind spending money on quality of things that matter in life at a price point of 1200 euro or so this ATX is not a no brainer to me, some way to audition it would be nice

Absolutely. It would be great to try first then buy if the value proposition is right. I would invite you over for an audition but I am NYC based 😁. The improvement in my system has been worth the price, having purchased more expensive amps/preamps/dacs etc. 

 

Again, my situation is unorthodox as I have no cards externally powered via LPSU (most people here do) so perhaps the difference in my system is far greater than those with DC4s/DC3 or other high quality LPSU powering their network card/usb card.

 

I will get around to testing that in the future if there is interest in doing so.

Link to comment

Guys,

Have we started looking at Gen3 Xeon Scalable CPUs? What we have below looks like a newer alternative to the highly hailed Xeon Silver 4210 CPU:

 

https://www.intel.com/content/www/us/en/products/sku/215283/intel-xeon-silver-4310t-processor-15m-cache-2-30-ghz/specifications.html

 

Here is a comparison between the 4310T and 4210.

The 4310T is a 10nm design running at a slightly higher base/turbo clock speed.

 

The interesting part that we may all benefit from is a larger cache! 13.75 to 15mb.

 

The downside? TDP appears to jump to 105W but with none of us planning to upsample with these Xeon Builds we should be fine.

 

Another downside: A new socket :(

 

Back to the ol' motherboard hunt. Also, I am assuming that this CPU will sound just as good as or better than the 4210. It is never good to assume. Perhaps there are other good discoveries to encounter.

 

I suspect Asus motherboards similar to the C621 sage will be a good place to start once they become available.

 

Link to comment
10 minutes ago, Nenon said:

There are several ways to fix that, and I tried some of them, but after all this, the particular transformer I chose sounded best to my ears.

This raises an interesting question. I am almost certain my speakers sound nothing like yours. I would describe my system as neutral with moderate efficiency and moderate to low overall power in terms of amplification. Rarely do I find myself listening above 80db from seating position (ok, the occasional 85db on a Friday night).

 

My speakers aren't the most efficient or fast, but they do keep up with most complex passages (more so now with the improved source). They provide extension down to about 30hz with authority from seating position. Overall, there is a reason i've had them in use for 15 years and have made incremental upgrades to them along the way.

 

I'm not the best with describing these traits but how would you describe the character of the system used for voicing the DIY Taiko server project? What characteristics about the Toroidy did you perceive to place it in 2nd place?

 

The reason I ask is I (and others) may find that those traits are more synergistic with my particular preferences, or, it may offset deficiencies in other areas of our system.

 

Just some food for thought.

 

Cheers

Link to comment
5 hours ago, dctom said:

which hxplex ATX are you using is it the 800w. Can you say which usb cable you use to your as well please.

The 800W HDPLEX was used. I probably should have mentioned that. 

 

USB Cable is the Sablon 2020.

 

Euphony Stylus is what I am testing with (no Roon components involved).

 

Both streaming from Tidal and local Redbook flac. Music files currently stored on Optane 800P.

Link to comment
5 hours ago, Nenon said:

3. All my experiments were with two 85W TDP Xeon CPUs. Those are much harder to power than a single 65W TDP AMD Ryzen CPU or a 95W TDP I9 9900k. And that is where this power supply arrangement really shines. I am not sure how well it works on lower power CPUs. I am guessing on low powered CPUs a good LPS would work better. But this dual CPU computer is a different story. Those are some experiments people would need to make. There are specific applications in which we want to use the Taiko ATX. It's not a miracle device that would work best in every single application. 

 

4. I find the GaN Fet ATX to be extremely fast. Too fast actually. Adding some big capacitors in front of it makes it fuller, more dynamic, and more expansive. It also slows it down a little, although it is still faster than a LPS. Seems like this is one of the things @Exocer ran into. 

 

@Nenon, thanks.  Do you have any interest in testing the unregulated supply + Taiko solution with your i9 rig (if its still around)? It would be an interesting data point for those who plan on sticking with single CPU builds for a while. 

 

In the future I will test whether the unregulated linear supply + Taiko Dc-ATX will surpass the two high quality PH SR7T rails. That would not be an apples to apples comparison since my SR7T only outputs 19V DC, so considering the impact voltage has on the Taiko DC-ATX it would only serve as a way to compare what most would have on-hand (at least a single high quality 19V regulated DC rail) to a dedicated unregulated supply voiced to work with the Taiko DC-ATX/Asus Sage C621/Dual Xeon Scalable CPUs. This is still a worthwhile test to conduct for the community.

 

Another test to conduct would be powering just the CPU from the PHSR7T and ATX from the Unregulated supply + Taiko. I would think this would get a bit tricky and may introduce grounding issues if not careful. The more tests we conduct, the better.

 

With the right EPS cable perhaps Solution A (Taiko powering everything) will work better for me. I am very picky about the high frequencies. Maybe an issue elsewhere in my chain is revealed by the Taiko. There are so many variables to keep in check and to take into proper context when testing and making comparisons.

 

I have switched to Qobuz today and I will be conducting a similar test to the previous one in the near future. I will probably keep both Tidal and Qobuz since Qobuz did not have some of my favs. 

 

Cheers,

-Rob

 

 

Link to comment
3 hours ago, dctom said:

It would be good if you could compare the PH SR7t to the Taiko PSU, maybe powering a single CPU would give different results? After waiting so long for the PH7 seems a shame to give it up!

Sure. I will conduct this test. I do not think you will need to give it up if you are sticking with a single CPU. The Taiko DC-ATX + PH SR7T is fantastic. You have the option to keep it and I think you will be pleased going that route. I think the Unregulated supply is more of a necessity for the Dual CPU build per Nenons tests. Will keep you posted.

 

  

3 hours ago, dctom said:

Have you compared local files and Qobuz using the Taiko?

No, I have not. I can test and report back my findings. Porting over my favorites to Qobuz as we speak :)

Link to comment
9 minutes ago, Downtheline said:
1 hour ago, seeteeyou said:

Supermicro X11SPL-F

I have this motherboard and the silver xeon 4210. I'm also waiting for a taiko dc/atx and I have a 2 rail ph sr7t. 

Would be interesting if you tried it with the Xeon Silver. You'll have less cores overall but there may still be benefits to doing what@seeteeyou suggested. This would be a great data point.

Link to comment
1 minute ago, Darryl R said:

Just to clarify, the only difference in DC cabling is the EPS run in Ghent, and the rest are all Hynes silver in A and B?

That is correct. I am hoping to confirm if the EPS cable had a major influence by testing again with a DIY version made with better wire.

 

4 minutes ago, Darryl R said:

 

So many variables to evaluate.  I'm also interested in the types of music involved.  In my experience for instance, orchestral and acoustic instrument, electric instrument, and vocal detail can vary.

Yes. Lots of Jazz (New and Old), and a little bit of everything else including Pop, Rock, Orchestral. I do not have the exact albums handy at the moment.

Link to comment
32 minutes ago, bit01 said:

Thank you for your observations. The Cons of A you mention are of concern to me too. I should be getting one Taiko DC DC-ATX in the batch shipping 4/30 - I will pay special attention to that and hopefully report back sooner than later.

Looking forward to your feedback as well! Solution A is by no means unlistenable. I will test again with a new EPS cable in the near future.

 

32 minutes ago, bit01 said:

That Gotham GAC4 mic cable heavily rejects RF - that will not be the case with the PH cables I suspect. I don't have it for EPS though so cannot comment specifically on that (I have the Neotech 7N UPOCC one). As you well know it is difficult to reach isolated conclusions in a non controlled test environment due to the permutations and combinations involved.

 

I do not think it is the GAC4 Mic cable in my case. This is the cable: https://www.ghentaudio.com/pc/gt02.html The PH cables are arranged in a way to reject RF but I do not know how effective they are at doing so. Either way, I do not hear more "noise" with the PH cables going from PH SR7T 12V out to EPS in so I am not inclined to believe that is influencing the sound much. As I said in another thread, we are comparing a single shared 10A 19v rail to the use of both a 19v 10a and 12v 12a rails. This should definitely be taken into consideration and may account for some of the observed differences in addition to the EPS cable.

 

Here is a response I wrote to another concerned user:

"For those with a 19v rail (For output into DC-ATX) + 12v rail (for EPS) that are not looking to go full linear for every rail, the Taiko DC-ATX is a major improvement over the HDPlex 800W. If you are using a single 19v rail you may want to invest in the higher voltage Unregulated LPS which was designed to partner with the Taiko DC-ATX as this is the optimal configuration for the Taiko DC-ATX. Or, you may want to go full linear per Marcin's Optimo ATX (I have not heard this unit yet but I trust Marcin's ears). Or you may still go with the Taiko DC-ATX with a single high quality 19V rail. In my tests, I can only say the SR7T 19v input per the synergy with my system did not work as well as having EPS powered by a direct 12v rail and that could very well boil down to a particular cable I have never been fond of in my system that I was forced to use for Solution A testing, or perhaps there was less current for the CPU available, too many variables changed from Solution A-B to rule out the Taiko as the issue. Soon I will conduct a re-test of Solution A with a different EPS cable and report back."

Link to comment
36 minutes ago, Darryl R said:

Good news.  Thanks Rob.  This is all I'm going to have time to do as well.  Could you refresh our memory on your DC input cable details?

Sure.

  • Gotham OFC 18AWG 8P-8P(CPU EPS) Power Cable(JSSG360)
  • Paul Hynes DC18FSXL-0.75m-M6: 
    • Length: 0.75m
    • This cable uses 3 sets of DC6 cable (hence the designation of DC18)
    • PSU connector: Jaeger
    • Device connector: Molex 6-pin (hence M6 designation)
    • I use a single 19V rail on my SR7T to connect this to the power input of the Taiko DC-ATX Adapter

  

37 minutes ago, Darryl R said:

Last week I installed the new Shunyata Sigma V2 speaker cable, a copper-silver hybrid, and it is the most significant upgrade I've made to my system since my DAC.  Just like Caelin says, Sigma 2 brings in the positives from each material without the negatives.  Amazing cable.

 

Awesome info. Here I am using this to good effect: https://www.partsconnexion.com/DUELUND-87080.html

 

What were you upgrading from?

 

 

Link to comment
11 minutes ago, Darryl R said:

Thanks Rob.  I went with the Neotech JSSG360 from Ghent, but I'll have to replace the DC input because it's not long enough.  So if I understand correctly your only using one of the DC6 runs now since everything is going thru the Taiko?

No, one cable consists of 3 runs of DC6 to create the DC18. I use the DC18 from PH SR7T 19V to the Taiko DC Input + the Ghent cable from Taiko to EPS.

 

Not in the mood to move anything but you can make out the 3 DC6's combined into a DC18 in this pic:

 

image.png.228af14d6adfe7c092da785fc75bfcee.png

 

 

 

Link to comment
8 minutes ago, dctom said:

 

I have been running my Taiko ATX for a few days powered by my 12v 19v SR7T.

so far I have preferred using the 19v to Taiko and 12v rail to EPS. Admittedly the Taiko to EPS is via the bog standard one that came with the Hdplex.

My advice would be to get away from that bog standard cable ASAP. That is most definitely your bottleneck. Also, my Taiko took more than a few days of continuous use to break in so definitely check back in the future with a better cable.

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...