The ultimate Power Supply Units for music servers (and other devices for cleaner power source)

[Nenon]

So, this is about transformers design rather than power supplies? R-Core vs Toroidal? I think we would end up with a discussion similar to vinyl vs. digital, tube vs. solid state, etc.

 

The guys at Uptone, who I respect and trust a lot strongly believe that the R-Core is better. Yet, the best power supplies on the market (i.e. SR-7) that are considered better than the Uptone JS-2 use toroidal transformers. But in a power supply the transformer is only one of the major ingredients. Would an SR-7 with an R-core sound even better? It is considered that Paul Hynes has perfected every component in his reference power supply over the years, so probably not. 

 

As mentioned the R-Core transformers emit EMI directionally, so they can be rotated in your chassis to reduce the EMI affecting other components. Toroidals do emit 360 degrees, but things like EMI shielding, GOSS band, covers, mumetal, etc. help reducing the EMI. And that's not the only thing that matters in a transformer. Materials and quality matter too. Take Audio Note and Kondo for example - they make their own transformers with pure silver wire. They are ridiculously expensive. So how would a toroidal with silver wire compare to a R-Core with poor quality copper wire? Or how will a R-Core with top quality silver wire compare to a toroidal with crappy copper wire? To put this into a prospective, you change a fuse that has a 2'' conductor and that makes a significant difference. Do you realize how much more wire is in a transformer?

 

I have vinyl records in my collection that sound a lot better than digital. I also have digital recordings that sound a lot better than the vinyl. Like almost everything else I guess the overall quality matters more than the type. And it's hard to generalize. 

 

[Nenon]

8 hours ago, Ralf11 said:

Isolate the music server from the DAC and don't worry about a PS for the music server.

 

For the DAC, you first want to buy a high quality one with a top notch analog stage - it may have multiple power supplies inside it.  A battery would be the ultimate PS for a DAC.

 

That would be my first strong disagreement on this forum, but we can't all agree. 

[Nenon]

9 hours ago, k27R said:

I haven't looked inside though, too scared to see what's in there

 

Posting a photo when you do would be much appreciated. 

I have never bought expensive equipment without knowing what's inside, at least at a high level / design concept.

 

[Nenon]

16 minutes ago, R1200CL said:

Can someone invent a way to parallel two UltraCaps ?

 

It must be possible adding some circuits after them...

 

You may want to look at those posts... assuming you are talking about LPS-1.2... if not, sorry about the spam. 

 

On 11/22/2018 at 10:31 PM, gstew said:

The 2 LPS-1.2s have to be carefully matched during setup by Alex to ensure they actually share the load & truly parallel. Without that one takes the load until it shuts off, then the other until it shuts off... You get the picture. 

 

I know he did the matching at least once & as far as I know swore to never do it again. I had those units & truthfully, I preferred a single unit to the 'twins'when used within their current limits. 

 

But now I use several configurations with 2 LPS-1.2s paralleled via a dual regulator board similar to the one I linked. I generally like that better than a single LPS-1.2.

 

On 11/23/2018 at 2:05 AM, JohnSwenson said:

The LPS-1.2 already has two LT3045 regulators in parallel. It is theoretically possible to use the same techniques to parallel 2 boxes together, BUT that is a different design. It takes a different board and several connections between boxes, it is NOT trivial.

 

The existing LPS-1.2 design is VERY carefully tuned, doing anything to the design is guaranteed to break SOMETHING, it would take several prototypes and months of development work to get right.

 

I do NOT want to do that. I want to focus on the new developments we are working on now, any modifications to existing designs are just going to slow down the new developments and I really don't want to do that.

  

John S.

 

 

[Nenon]

What if the source is Tidal/Qobuz? 

[Nenon]

43 minutes ago, Chopin75 said:

I don't know, I don't do internet streaming but that would take into account of how they play the music e.g how does Qobuz play the music, and how do they transmit the music (compressed data or full 24 bit etc) so I guess it could be a combination of 1 + 2 I mentioned.

I suppose even the internet service matters, of it is efficient in transferring the data, are the optical cables they use of high quality ??

(for e.g. my comcast give shxty looking TV services, you can see jitter on the screen, though occasionally when I listen to internet radio it sounds fine, I guess there is much less data)

If you stream large files like DSD (if that is possible ?) then there will be other issues of difficulty sending huge files, with signal loss etc I would presume ??  I don't think tidal/Qobuz do DSD anyway, Their upper limit is 24/192 correct ?

 

 

I was just pointing out that you don't always have access to the source to fix it. I guess in those cases good recklocking of the signal helps. I don't stream, because the quality can't compare to my local files, but I have noticed that as I improve my network and server streaming Qobuz sounds better and better. Maybe one day streaming quality would get close enough that convenience would win :). 

But that seems to be offtopic. 

 

BTW, on the topic of power supplies, I just noticed some good power supplies for sale in the "Buy & Sell" forum - JS-2, HDPlex 400W, and a Sean Jacobs LPS. We have quite a few choices like never before. 

[Nenon]

10 hours ago, Jud said:

Regarding simplicity, that can be but isn’t always the way to get best sound.

 

Two things are generally true in the usual computer audio setup:

 

(1) Power from a normal desktop or laptop will not be terribly clean.

 

(2) Clocking from the source can’t physically be as precise as clocking at the DAC.

 

The second of these means that all else being equal, we’d want async USB input for the DAC.

 

The first means we probably don’t want the desktop or laptop electrically connected to the same system as the DAC.

 

A streaming device with clean power and good clocking seems to fill both requirements.  It may also have the advantage of requiring a smaller capacity and thus less costly power supply.

 

I completely agree with both of those things. Clean power is the most important thing in digital audio. Excellent clocking sits right next to it. The clock depends on clean power. How clean the power is does not depend on the clocking, though. 

 

However, when you not only pass signal from one stage to another but also do software processing, clean power and clocking are not the only things that matter. And this I believe is where we'd disagree. As it turns out more powerful processors are much better than slow processors. To me that means that processing latency is crucial, and in some cases it might be more important than clocking  (given at least a decent clock implementation of course). 

I always found the microRendu and ultraRendu mid-level devices. They are not bad for the money but never impressed me. I kept my ultraRendu for as long as I could, and tried to like it as hard as I could in my system but ended up selling it after a while. I believe the main problem with them is their processor cannot compare to my i9-9900K. I've never heard the layers of depth and information, the explosive dynamics, and the wide soundstage with my micro/ultraRendus that I am hearing now. Actually the Rendus were good at depth but not at dynamics and did not have the wide soundstage I get now (going beyond the walls of my room), and a few other things I won't bother trying to explain. 

 

Also, from my experience (and John Swenson seems to be confirming that) the clocks at the source matter. They always made a difference in my system, no matter how good your DAC is. An owner of a MSB Select DAC with their Femto 33 clock is confirming all my findings. It's hard to beat the clock and implementation on that MSB DAC, so that must mean something. At this point I have absolutely no doubts that the clocks before the DAC matter. If anyone thinks they don't, you are on the wrong track my friend... 

 

So to put things into prospective, since the topic about simplicity was touched by the thread starter @Chopin75 , some extra reclocking, if done right, and with clean power, might be beneficial at the digital source before the DAC. Multiple good clocks in series can diminish the negative impact of a bad upstream clock. Again, if done right, as they can easily do more damage.  Of course nothing beats a good clock at the source and a minimalistic approach with as little devices as possible. But sometime it is not possible.

 

BTW, I do believe SOTM when the say that two of their switches in series sound better than one. I am sure that's the case for people streaming Qobuz/Tidal. The signal passes through more clocks, and that helps removing the fingerprints of bad upstream clocks. Here is a video where Hans Beekhuyzen puts two devices that clean up his USB signal, and adding a second device is an improvement - https://www.youtube.com/watch?v=U4RYineYsF0.

 

 

 

 

[Nenon]

11 hours ago, charlesphoto said:

 

Looks interesting (as well as the Sean Jacob units), but I think I'm going to stick to products made on this side of the pond. If something does go wrong, then shipping for either exchange or repair overseas is nothing but a pain. 

 

Interesting perspective. To me, sound quality takes priority over any other factors. And even if I need to order parts from another planet, I would not care, as long as they sound better, and I can afford them. I would leave the worries about warranty and repairs for my mass production appliances. Those I would not buy from oversea, but my audio stuff is coming from where I think they make the best I can afford.

Funny enough - overnight shipping from the Texas based soniccraft to my Chicago area house costs about twice more than the overnight shipping from the UK based hificollective. Go figure!

Now back to power supplies - it just happens that our US-based manufacturers do not make power supplies on par with the sound quality of some overseas manufacturers. Ironically, the overseas manufacturers that make excellent LPS's use a lot of parts designed in the US . 

[Nenon]

3 minutes ago, Chopin75 said:

Totally agree, would you say that in general it is better to have a better clock downstream than upstream (of course best is to have clocks equally good all the way) ? So the downstream clock would not contaminate the audio signal ?  I would imagine an inferior clock downstream would be counter productive and would defeat the purpose of an excellent clock upstream.  

 

Absolutely! The closer to the analog conversion stage, the more important it is. The very last clock in the chain is the most important one. The further upstream the clock is, the less impact it makes. 

Many engineers believe that the last clock is the only clock that matters. I don't. 

[Nenon]

7 hours ago, R1200CL said:

Is this a NUC ?

Are you using HQPlayer since you need such powerful processor?

Not a NUC. A server I built. And no, I don't use HQPlayer and don't upsample. I use Euphony / Stylus. The resolution does not matter in my system. The recording does. Some 44.1k files sound better than 192k and vice versa. It's all about the recording process and how much care was taken to preserve the original recording quality.

It just turns out that in my system the faster the CPU is, the better it sounds. 

 

4 hours ago, vortecjr said:

Each Sonore Rendu with USB output has a dual-core 1GHz processor which I would not necessarily categorize as low power.

A dual-core 1GHz iMX6 in my book is exactly what I mean by slow and low powered CPU. I went from a quad core 2.5Ghz CPU to an Intel i9-9900K and the difference was stunning. It's not about whether the CPU has enough power to do the job. It's something else, which I don't understand. 

My CPU barely has to do anything at all. This is how much it is utilized while playing. But it sounds amazing!

 

Obviously you need a really good power supply with this CPU. It can sound bright and harsh as hell if you don't have clean DC power for it. A lot of that is discussed in the novel thread. 

[Nenon]

4 hours ago, vortecjr said:

John's research doesn't suggest that you need to use multiple $1000 switches in series.

 

It does not directly suggest that. But it does suggest that multiple better clocks after a bad clock can reduce the negative fingerprints of the bad clock. And adding two SOTM switches does exactly that. So until someone comes up with a product on the market that somehow reproduces the positive impact of multiple good clocks, we are stuck with multiple SOTM switches as one of the best solutions.

The EtherREGEN supposedly does that, but that product does not exist on the market yet. I would be one of the first people to get one when it is released, and.... I am not that crazy to stack multiple SOTM switches... while I do believe in the sound benefits of doing that, I have better things to spend my money on.  

[Nenon]

51 minutes ago, misterspense said:

Again the product (etherregen) already turn legendary before the first ones are actually sold. 

 

Because of some bold claims. But nobody knows if they are true. I am looking forward to get my hands on one and find out. And I really hope it turns out to be as good as advertised.

 

53 minutes ago, misterspense said:

I quit well remember the first usb regen, which would clean up a usb signal so good, that everything upstream would not matter anymore.....

 

Did it? Not in my experience.

 

1 hour ago, vortecjr said:

you are just mis-informed.  

 

Quite possibly. I don’t know a lot of things for sure. Perhaps you can elaborate? 

[Nenon]

I might have offended some Sonore fanboys. Sorry if that is the case. I've had two Sonore products. A microRendu and an ultraRendu. I owned the ultraRendu for quite some time and sold it just recently. I've tried it with several different power supplies - LPS1.2, JS-2, sBooster, and Sean Jacobs DC3. The DC3 is not even on the list of recommended power supplies from Sonore and it sounded better than the rest. I think I gave over a year to the ultraRendu to shine in my system. But it never did. At that point the difference between my vinyl rig and digital was so big, that I was not even taking my digital rig that seriously (it was playing a lot but mainly as background music). I did all kinds of network tweaks - filters, transformers, modded fiber media convertors, enterprise class Cisco switches (since I have access to a lot of them), SOTM switch, nice LPS on everything network related, vibration isolation, expensive ethernet cables, etc. etc. Some of those things definitely helped, but again I was never excited about my digital rig.

Then I got an Innuos Zenith MK3. I would like to be careful with the words I am using but to say that the Innuos trashed the ultraRendu would be an understatement.  This opened my eyes and got me excited about digital again. I wanted to build my own server. I tried different motherboards and different CPUs. In the beginning I was a big fan of the Innuos philosophy - low powered CPUs and motherboard, clean power, minimalistic approach, etc. This approach did not get me passed the quality of the Innuos Zenith MK3 server. Maybe I got close, maybe I exceeded the Innuos quality in some areas but not others.

Then, a new trend was started by a couple companies and some people here wrote about the positive impact of a more powerful CPU. A little sceptical but completely open minded, I decided to give it a try. Got one of the gaming motherboards that was recommended and a i9-9900K CPU. Powered it up with a distributed linear power supply from Sean Jacobs the same way Innuos did on their $14K Statement. But I actually did more than that...two 400VA transformers in different chassis, upgraded parts on the power supply, super short 15.5AWG Mundorf silver/gold wires (2'' average length) from the regulators to the ATX and EPS connectors shielded in JSSG360 were some of the little things I did. Some sparse EMI treatment, Furutech Nano Liquid on some parts and wires, good vibrations isolation, multiple chassis, etc. etc. When I say attention to every detail was paid, I really meant it. When that computer was ready and I plugged it in, I was not very impressed... but it needed some time. A month later (running non-stop) it exceeded every expectation I have ever had. It's on a completely different level than the Innuos Zenith MK3 (the thing that completely trashed my ultraRendu if you still follow). No comparison at all. For the first time I really enjoy digital and on some recording it exceeds my vinyl (never thought I would have said that). My vinyl rig is not cheap and was very carefully built after about 5 years of carefully listening and testing - $10K cartridge, $7K tonearm, $15K phono stage, I don't even know how much my turntable costs days, but it is not cheap. In other words it is hard to beat. 

How does the faster CPU, more cores, and more cache impact the sound quality? I don't know. But it does, and I am not going back to the Innuos philosophy.

My current digital source is light years ahead of the Sonore ultraRendu at the moment. I have no desire to try the opticalRendu - I've read reviews and appreciate that people like it, but I did not get the impression that is light years ahead of an ultraRendu with a bunch of network tweaks I've briefly mentioned above. My common sense tells me it's a step backwards from where I am now. If I see that people massively start replacing their Innuos Zenith MK3 or Innuos Statement servers for opticalRendus, I would buy one right away. 

What I find interesting is that I share my experience with the only intention to help people improve their digital source, and I feel like I am getting attacked from different angles. I don't have time for that. I am not an audio equipment manufacturer. I don't offer consulting services. I am sharing my experience for free using my spare time during those nice summer months. I have an open minded attitude and switched tracks multiple times. But those Sonore fanboys don't seem to want to try anything else. Well, in a way I am glad for them that they are happy with the sound they are getting. I was not. And again, the Sonore products I have tried are good for what they do and have a great value for their price. They are just not the best of the best. Not even close. And I am always looking for the best :). And while looking for the best, the Sonore opticalModules are part of my attempt for best network design (posted a diagram in the novel thread about that), so I am definitely getting a couple of them (if they are willing to sell them to me after my posts . 

I can share every detail of my digital source if you want to try it. Try it and then post your findings. But don't ask me to explain why a faster CPU sounds so much better. I don't know. I have some speculations in my head, but they are too controversial to post here.

Actually, the only reason I have not shared every detail of my digital source yet is because it has been under constant development, and I still have a few things left to try. Once it's fully settled, if I am still an active member of this forum, I will post it under my profile. 

Okay, now that this is all out of the way, let's focus on power supplies. Ping me in a PM for any offtopic questions related to my posts. 

[Nenon]

7 hours ago, Chopin75 said:

Hence I wonder what would be minimal power needed to run a PC with these powerful CPU, which seems to range from 12-19V. I use 19V/10A HDPLEX 200  but apparently I need this high power to drive the the DC-ATX. Now most LPS are rated about 12-15V and 5-7A for music server. Would a more powerful PS in general be beneficial, too ? 

 

I use 4-rails.

 

--ATX connector--

3.3V - 1.5A

5V - 5A

12V - 1.5A

--EPS connector--

12V - 5A

 

More power typically helps if it is of the same quality. Typically a 1A and a 7A power supplies would not have an absolutely identical design. Typically... 

[Nenon]

24 minutes ago, luisma said:

@Nenonare you feeding the board directly from the rails or you are using caps in between?

 

 

Directly. The voltage regulator boards sit about an inch or two away from the ATX / EPS connectors. 

http://www.the-ear.net/how-to/power-supply-design-innuos-statement - that's pretty much what I have done, but with shorter JSSG360 shielded Mundorf silver/gold wires, WBT silver solder, Furutech NanoLiquid, etc. It was a very OCD-style implementation. And very time consuming. But the end result was awesome. 

[Nenon]

1 hour ago, sandyk said:

Re the attached statement from the link

 

Doing that is a definite No,No,  as the ripple injected back into the supply from the SSD/HDD consists of  "square wave" pulses with a bandwidth to >10MHz  with a typical HDD, and probably worse with an SSD

 

 

What is a "definite No,No"? 

[Nenon]

20 minutes ago, sandyk said:

 

Don't feed power to both an Oscillator and an SSD from the same supply rail without further regulation (preferably) or further filtering  of the power to the oscillator.

 

They don't do that. Perhaps you misread what they wrote. They are also saying it's not a good idea and explaining why they use 8 rails on this server. The SSD has its own separate rail.

I actually like that they also have two rails for the USB reclocker - the OCXO clock on the USB reclocker has its own power rail and the USB uses another rail. I wish more hardware manufacturers like JCAT, PinkFaun, SOTM, etc. did the same. 

 

[Nenon]

8 hours ago, sandyk said:

 

I didn't misread it. I was trying to reinforce in more depth what they were saying.

 It's not just about small voltage variations, it's about the wideband, fast rise and fall time noise from the SSD being superimposed on the supply to the Oscillator,

 

Agreed.