A novel way to massively improve the SQ of computer audio streaming

[Nenon]

12 minutes ago, Chopin75 said:

hmm. so I guess I can tell him what I need - e.g what my server is, how many amp I need etc and he may be able to build one then?

Exactly. As far as I am aware his DC3 power supply with Mundorf caps is identical to what the Innuos Statement uses. If there is any difference (other than the custom transformer of course), I haven't been able to spot it.

 

14 minutes ago, Chopin75 said:

I hope he is more reliable than Paul Hynes.

Way better! But as far as I am aware he is a one man shop, and hopefully he would be able to handle higher demands if it ever gets there.

 

14 minutes ago, Chopin75 said:

From your article it takes 6 months to get the Server+ LPS !

Not sure what you are referring to.

[Nenon]

14 minutes ago, Chopin75 said:

I can't remember where I read this from but maybe it wasn't even about the  LPS ?? Does he work for Innuos or he owns Innuos ? 

 

He does not own Innuos. He just designed their power supplies. His company is called Custom HiFi Cables - http://www.custom-hifi-cables.co.uk/home/power-supplies/dc3-power-supply

 

[Nenon]

2 hours ago, LTG2010 said:

@Nenon's supply is based on the Innuos Statement, which has outbox DC regulation,  situated next to the motherboard and dual umbilical for shorter signal paths, one large transformer,

 

Let me clarify something. Sean did not do that. It was a DIY project I did. I used his regulators and recap modules and placed them the way I described (similarly to the way it's done on the Statement). I ordered a large transformer by his specs for this project. I wanted it to be massively oversized and the best build quality. 

As far as I am aware Sean does not offer what I have done. It's a tedious job to mount and wire everything. It's not that complex but takes a lot of time to do.

 

1 hour ago, rickca said:

Many of the latest gaming motherboards which people are considering due to good VRM etc. now have two 8-pin connectors because they're designed for overclocking.  How do you handle such a thing?  Do you use two 12V rails with half the amps for each connector or what?  I'm not sure whether such motherboards are stable if you only use one of the two 8-pin connectors.

 

I use one 12V rail. Although you have an 8-pin connector, four of the pins are the DC+ and the other four are the DC-. I mean they are paralleled on the motherboard and are not separate inputs. At least that's how they are on my Asrock gaming motherboard. I think the reason they use 8-pins instead of 4 is to end up with a higher gauge wire when you combine four parallel wires from the power supply to the motherboard.

 

2 hours ago, Superdad said:

By the way, I’d put the nice (and completely silent) R-core trans we use up against any toroidal any day. The difference is obvious right away.

 

I respect you and John a lot, just from the posts I have read. And I would never even try to go against anything John says about that stuff. Not to argue (don't really want to do that), but to provide another perspective. I had an argument about a R-core transformer (that looks very similar to yours on the outside) with a DAC manufacturer. And I tried his transformer. To make a long story short, his R-core transformer worked better on his DAC, but not on the computer (seperated in a different chassis like I have described earlier). One of the differences was that while his transformer was oversized, it was still much smaller than the toroidal I used. And the toroidal I used was much more expensive, had electrostatic shield, GOSS band, and every trick in the book to reduce noise (some of which might have been done on the R-core too). This does not tell ANYTHING about YOUR "R-core" transformer, but it convinced me as an end user (not a manufacturer) not to look back. Also, given that the power supply I use took about 3-4 months to completely break-in, one probably needs about 6 months of parallel running to determine the true performance of each transformer. I did not spend 6 months on this test, so take my word with a grain of salt. But for me, I am sticking to the oversized toroidal. 

 

[Nenon]

7 hours ago, rickca said:

I'd be surprised if it didn't outperform a 12V rail from an HDPLEX 400W ATX LPS.

me too

 

4 hours ago, rickca said:

You misunderstood me.  The motherboards I'm talking about have dual 8-pin EPS12V connectors or one 8-pin plus one 4-pin, not just one 8-pin connector with 4x12V pins and 4xground.  For example, ASRock Phantom Gaming 9 Z390 has both an 8-pin and a 4-pin.

Yes, sorry, I misread your message. Does it say anything in the manual what each is used for? I haven't used a motherboard like that yet. Have you checked with a multimeter if the DC+ on one of the EPS connector is connected to the DC+ on the other? It seems like you may need (benefit from) 5 rails for a motherboard like that.

 

4 hours ago, LTG2010 said:

You used your own transformer?

Yes. 

 

4 hours ago, Superdad said:

Of course.  I was just referring to an apples-to-apples comparison of similar VA-rated toroid versus R-core—swapped into the same supply. I did just that during development of the JS-2 and it was not even close.

Good to know. For my next project I may try multiple R-core transformers (one dedicated for each rail). Sounds like that would be a killer LPS. We, DIY-ers, can afford one-off exotics like that from time to time. It would be more difficult in a commercial solution where so many things need considering. 

 

 

[Nenon]

I have mentioned in one of my previous posts, but I did compare an LPS-1.2 with a 1.5A Sean Jacobs LPS. I compared several 5V devices (JCAT, Allo, USB reclocker, etc.), an ultraRendu, and a NUC that was able to run from the 12V on the LPS-1.2. In most tests the LPS-1.2 was charged by sBooster. I liked the SJ a little better in every test I did. 

I consider the LPS-1.2 an amazing device with genius engineering that also sounds really really good. I am not sure why the SJ sounded better in my system, but the larger transformer and big Mundorf caps may very likely have something to do with it. It's also more expensive. 

[Nenon]

@Darryl R

 

This may be a huge overkill, but I am consider something like this in my network and will be happy to report back the results once I try (but first we need the EtherREGEN switch to be released):

If the claims Uptone Audio is making about their new switch are true, this can become a flat network with one switch only and no bridging...

I have my fingers crossed their claims are true. But I also highly doubt some of them would be completely true in practice. Hence, I added some extra optical isolation and a second EtherREGEN in the diagram above. Again, this might be an overkill. Also, I picked the Sonore opticalModule as it is regarded as the best fiber media convertor for audio applications on the market at the moment. Let me know if there is a better one. 

 

Too much speculation at this point, so here is a fallback scenario as well. 

If the EtherREGEN switch turns out to be a disappointment, then a similar solution can be implemented with a SOTM switch:

 

The part in blue may not even be necessary. Instead the second SFP port on the SOTM can be utilized. Listening tests would determine what's the best option. 

 

[Nenon]

1 hour ago, CJH said:

Hey Nenon how about lose one EtherRegen and replace with a multi SFP switch like Mikro Tik CRS305 (at $153). Both Optical modules can plug in as well as one EtheRegen feeding one RJ45 to the JCat.

 CJH

 

I was going for an ultimate network design with as little compromises as possible. My two major design goals are:

1. Keep any noisy devices isolated from my server; and

2. Best possible reclocking.

 

One thing I did not like on the diagram @elan120 posted above is that the WiFi access point is directly connected to the server with a copper connection. I consider the WiFi access point a 'noisy' device and would prefer to be isolated from my server. The NAS can also be 'noisy'. That's why I used an optical connection for the NAS in my diagram as well. 

 

Similarly, the Mikro Tik CRS305 you are suggesting might leak some noise through the RJ45 to the server. You can remove one EtherREGEN, it's probably an overkill anyway, but I would replace it with another opticalModule instead. Like this:

 

The reason I used an EtherREGEN in my original diagram is because according to John Swenson, who designed both, the clocking on the EtherREGEN is better than on the opticalModule. And that's important for the music files transfer from the NAS to my server. It's also my second design goal. 

Alternatively, you can just simplify things (and cost) and connect the NAS via RJ45 to the EtherREGEN. 

 

[Nenon]

3 hours ago, CJH said:

Not sure why you want 2 RJ45 connections to your JCat. My thought was to use the optical connection from both Sonore optical modules and an optical connection from the etherRegen into the Mikro Tik CRS305 since it has four possible optical connections (no use for the RJ45 connection of the CRS305). Then use one RJ45 connection from the etherRegen to your JCat. Total optical isolation and great clocking going into the JCat.

CJH

 

Of course that would work. But the signal from your NAS to the server would pass through a potentially noisy Mikro Tik. I am trying to avoid that. 

IMO, you would be better off putting an opticalModule instead of the Mikro Tik - like the second diagram in my post (#14518)...

And, no, you don't have to use the second JCAT port, but it provides extra isolation (i.e. your home network router, wifi, etc. do not share the same switch as your NAS). We might be going offtopic here, though. Ping me in a PM if you want to discuss more. 

[Nenon]

2 minutes ago, sukyung said:

streacom fc10 + RYZEN 7 3700X + gigabyte x570 aorus xtreme
I am making an audio PC with this combination.
I want to use the pink faun SPDIF bridge.
Gold-plated Premium PCIE 16X 2U 90 degree rectangular riser card
Would it Sound quality okay?

 

 

 

 

Raiser cards and extension cables seem to degrade the sound quality a bit. You would be better off with a streacom fc9 as it does not need a riser card. 

[Nenon]

What is the best sounding AMD Ryzen motherboard people have tried here?

I am thinking about building another server with the gen3 Ryzen and doing a side by side comparison with my highly tweaked i9-9900K / AsROCK Phantom Gaming server. It would be an interesting test and will try to keep as many things constant as possible. 

From a quick search I am leaning towards the ASUS ROG Strix X470-I. Any recommendations? 

[Nenon]

11 hours ago, seeteeyou said:

though we really don't have anything else that could be compared to SR4 while delivering enough current to satisfy the power requirements of Ryzen 3000 Series or i9-9900K etc. that could consume more than 100W during heavy workload.

 

Sean Jacobs has custom products that can do that - http://www.custom-hifi-cables.co.uk 

[Nenon]

7 hours ago, seeteeyou said:

Really?

 

Dr. Jacobs seemed to offer up to 5A for each rail, did you mean combining two of them in parallel and reach 10A by any chance?

 

I've heard he offers a 10A LPS, but I don't see it listed oh his web site. 

[Nenon]

1 hour ago, George Hincapie said:

 

How do his supplies compare to the variants from MCRU?

 

I don't know what MCRU is. Never heard of them. 

But the SJ LPS is out there competing with the best. In my system, it sounds significantly better than sBooster, LPS1.2, JS-2, and HDPLEX. A trusted source prefers it over SR-4. That's as far as I can help with comparison. 

 

It would be nice if @austinpop can test a Sean Jacobs  LPS given his recent review and exposure to treasures like the SR-7. 

[Nenon]

29 minutes ago, austinpop said:

 

He designed the PSU that ships in that server. So there was nothing to compare, just that the SE has excellent SQ.

 

I used a Sean Jacobs LPS like the one used in the Innuos Zenith MKII SE on the stock Supermicro motherboard Innous uses in their MKIII series (before their customizations). The sound was very good. But then I moved the same LPS (with some small modifications) to the ASRock gaming motherboard / i9-9900K I am using now. The sound quality bump was enormous. I have both servers, but ever since I plugged in the i9-9900K, I had no desire to go back to the 4-core Supermicro server. That was a few months ago. The Supermicro was collecting dust since then.

I decided to take the Supermicro out a couple of days ago and check how it sounds. I always like to go back to my previous components after some time to double check if my initial impression was right. I let it run for 48 hours before I connected it to my system. Just listened to it, and I don't like it at all. No dynamics, no musicality, less toe tapping, smaller soundstage, etc. The i9-9900K is on a completely different level. 

 

Decided to share as this experience should be back on topic after a bunch of OT posts. 

 

[Nenon]

48 minutes ago, George Hincapie said:

 

I presume this is a server you've built yourself?

 

If so, did you document the build process or upload any images so I can have a look please?

 

Yes, it's my build.

See posts #14226 and #14229 here. And I posted a few other comments after that. No pictures yet because it's not completely finished. I have a streacom fc9 and need to move the computer components and regulators in it. But before I do that I would like to try a Gen3 AMD Ryzen with Asus motherboard just to make sure I am not missing something. Once that test is over, I will pick the best components, put them in the new chassis and finally close the top cover :). 

[Nenon]

1 hour ago, seeteeyou said:

The rationale behind choosing that ASRock gaming motherboard (instead of Supermicro) could be found here

 

https://audiophilestyle.com/forums/topic/55873-shootout-at-the-linux-corral-audiolinux-vs-euphony/page/9/?tab=comments#comment-961583

 

For X570 motherboard and Ryzen 3000 Series, this particular motherboard is gonna be fanless while beating ASRock (14 + 2 phase with 70A instead of 5 + 2 phase with 60A) handily

 

https://www.reddit.com/r/Amd/comments/bxl3rm/gigabyte_x570_xtreme_motherboard_analysis_true/

 

https://pokde.net/system/pc/motherboard/gigabyte-unveils-true-16-phase-amd-x570-motherboard-pcie-4-0-ssds/

 

 

However, its form factor is E-ATX and therefore only one fanless chassis is available so far

 

http://www.turemetal.com/product_en.html

https://item.taobao.com/item.htm?id=590791895684

https://www.fanlesspro.com/shop/product/fanless-cases/turemetal-up10-fanless-case/

 

The CPU alone could use 2 separate 12V rails (i.e. two 8-pin EPS connectors) to begin with, getting quite expensive already and then we'll also need one more rail for 24-pin ATX connector. Quality rails cost quite a bit while it also takes at least several months to build each LPSU.

 

There's also C621 AORUS XTREME with 32 phase VRM and we could even feed 4 separate rails to the CPU itself, followed by 2 more rails for a pair of 24-pin ATX connectors. Could such a beast beat that €20,000 Taiko Audio SGM Extreme?

 

That's a killer motherboard and a killer chassis. Thanks for sharing. 

Different concept than the Taiko SGM Extreme, but it would be nice to try one day. The Extreme uses dual Xeon CPUs, server-class motherboard, custom made registered RAM, a lot of vibration isolation, and they do something special with the clocks claiming it's better than the really good OCXO clocks. Hard to beat I guess. 

[Nenon]

2 hours ago, drjimwillie said:

 I understand that a bridge is a virtual switch inside your computer.

 I am currently using a bridge,  Port one in from my switch, port two out to my endpoint. 

 What I’m asking about is a little bit of out-of-the-box thinking.   

  What if you do what I have in the sketch? 

  Keep the bridged connection configured on your server and connect both ends to the switch.

  

 Is anyone doing it? 

 Does it work?

 How does it sound!

 

 

What are you trying to accomplish with this configuration? It does not make much sense to me.

 

Your endpoint is connected to your DAC. You need to take care of isolating the noise from your (noisy) Router and your Endpoint/DAC. One of the ways you can accomplish that is to configure a network bridge on your Server. But by connecting both ports on your Server to your switch you are removing that isolation (defeating the purpose) - the Endpoint and the Router share the same switch. At that point, there is no point to have a bridge. 

 

 

If you want to use a bridge with all your components, you have two options:

- Inbound from router <-> (port 1) Server (port 2) <-> Switch <-> Endpoint.

- Inbound from router <-> Switch <-> (port 1) Server (port 2) <-> Endpoint.

 

[Nenon]

2 hours ago, Tommd64 said:

Compared the prices and see whats going on. I ll say without hearing please buy a Farad3, it's also possible to order it with highgrade componends..or better dc cable. Coz he 700p for a Sean Jacobs is a lot of money for a realitively simple PS. Compared to a Farad or SotM.

 

And how did you come to that conclusion? You can tell how things sound by looking at a picture? 

 

The LPS1.2 for example has some outstanding engineering for example. But it does not sound as good the 1.5A Sean Jacobs LPS nowhere in my system. I've had some exposure to the SOTM and was not impressed. The Farad looks interesting but I've never had one to try. 

[Nenon]

34 minutes ago, drjimwillie said:

I had read somewhere regarding ethernet connections that more  Ethernet connections made for a better sound.

 

This post explains that pretty well:

On 6/4/2019 at 7:07 PM, JohnSwenson said:

The understanding of "isolation" in digital audio has been my passion for at least 10 years. There is a LOT of misunderstanding on the subject floating around in audio circles.  Here is a quick summary of my current understanding and how the current products fit in with this.

 

There seems to be TWO independent mechanisms involved: leakage current and clock phase noise. Various amounts of these two exist in any system. Different "isolation" technologies out there address one or the other, but very rarely both at the same time.  Some technologies that attenuate one actually increase the other. Thus the massively confusing information out there.

 

Leakage current is a property of power supplies. It is the leakage of AC mains frequency (50/60 Hz) into the DC output. It is usually common mode (ie exists on BOTH the + and - wires at the same time, this makes it a bit difficult to see. There seems to be two different types, one that comes from linear supplies and is fairly easy to block, and an additional type that comes from SMPS and is MUCH harder to block. An SMPS contains BOTH types. They are BOTH line frequency.

 

Unfortunately in our modern times where essentially all computer equipment is powered by SMPS we have to deal with this situation of both leakage types coming down cables from our computer equipment. There are many devices on the market (I have designed some of them) for both USB and Ethernet, most can deal with the type from linear supplies but only a few can deal with the type from SMPS.

 

Optical connections (when the power supplies are completely isolated from each other) CAN completely block all forms of leakage, it is extremely effective. Optical takes care of leakage, but does not deal with the second mechanism.

 

Clock phase noise

 

Phase noise is a frequency measurement of "jitter", yes that term that is so completely mis-understood in audio circles that I'm not going to use it. Phase noise is a way to look at the frequency spectrum of jitter, the reason to use it is that there seems to be fairly decent correlation to sound quality. Note this has nothing to do with "pico seconds" or "femto seconds". Forget those terms, they do not directly have meaning in audio, what matters is the phase noise. Ynfortunately phase noise is shown on a graph, not a single number, so it is much harder to directly compare units. This subject is HUGE and I'm not going to go into any more detail here.

 

Different oscillators (the infamous "clocks" that get talked about) can have radically different phase noise. The level of phase noise that is very good for digital audio is very difficult to achieve and costs money. The corollary is that the cheap clocks used in most computer equipment (including network equipment) produce phase noise that is very bad for digital audio.

 

The important thing to understand is that ALL digital signals carry the "fingerprint" of the clock used to produce them. When a signal coming from a box with cheap clocks comes into a box (via Ethernet or USB etc) with a much better clock, the higher level of phase noise carried on the data signal can contaminate the phase noise of the "good" clock in the second box. Exactly how this happens is complicated, I've written about this in detail if you want to look it up and see what is going on.

 

The contamination is not complete, every time the signal gets "reclocked" by a much better clock the resulting signal carries an attenuated version of the first clock layered on top of the fingerprint of the second clock. The word "reclocked" just means the signal is regenerated by a circuit fed a different clock. It may be a better or a worse clock, reclocking doesn't always make things better!

 

As an example if you start with an Ethernet signal coming out of a cheap switch, the clock fingerprint is going to be pretty bad. If this goes into a circuit with a VERY good clock, the signal coming out contains a reduced fingerprint from the first clock layered on top of the good clock. If you feed THIS signal into another circuit with a very good clock, the fingerprint from the original clock gets reduced even further. But if you feed this signal into a box with a bad clock, you are back to a signal with a bad fingerprint.

 

The summary is that stringing together devices with GOOD clocking can dramatically attenuate the results of an upstream bad clock.

 

The latest devices form Sonore take on BOTH of these mechanisms that effect sound: optical for blocking leakage and multiple reclocking with very good clocks. The optical part should be obvious. A side benefit of the optical circuit is that is completely regenerates the signal with a VERY low phase noise clock, this is a one step reclocking. It attenuates effects from upstream circuits but does not completely get rid of them. This is where the opticalModule comes into play, if you put an opticalModule in the path to the opticalRendu you are adding another reclocking with VERY good clocking. The result is a very large attenuation of upstream effects. It's not completely zero, but it is close.

 

The fact that the opticalRendu is a one stage reclocking (which leaves some effects from upstage circuits) is why changing switches etc can still make a difference. Adding an OpticalModule between the switch and opticalRendu reduces that down to vanishingly small differences.

 

So an optical module by itself adds both leakage elimination and significant clock effects attenuation. TWO optical modules in series give you the two level reclocking .

 

An opticalRendu still has some significant advantages over say an ultraRendu fed by a single opticalModule, the circuitry inside the opticalRendu has been improved significantly over the ultraRendu. (it uses new parts that did not exist when the ultraRendu was designed). In addition the opticalRendu has the reclocking taking place a couple millimeters away from the processor which cuts out the effects of a couple connectors, transformers and cable. The result is the opticalRendu has some significant advantages.

 

An opticalModule feeding an ultraRendu does significantly improve it, but not as much as an opticalRendu. So you can start with an opticalModule, then when you can afford it add an opticalRendu, also fed by the opticalModule and get a BIG improvement.

 

I hope this gives a little clarity to the situation.

 

John S.

 

 

 

 

 

39 minutes ago, austinpop said:

Sorry, but that's just not how Ethernet works. 

Exactly.

 

56 minutes ago, drjimwillie said:

2.  The other reason I am asking this question is because I am doing my homework to be prepared for the EtherRegen.   If there is complete isolation from one side of the switch to the other then there may not be a detriment to hooking up multiple cables. 

 

Plug your Endpoint to the B side of the EtherRegen, and everything else to the A-side. I would probably isolate the router with fiber too. 

[Nenon]

The JS comment I quoted was the best explanation I have seen about the effect of (multiple) clocks. I thought you might be referring to the same post.

 

It's not that complicated. You have electrical noise you need to worry about and you have reclocking. You want the electrical noise to be as isolated from your endpoint as possible. You also want to have the best clocks as close as possible to your Endpoint. That's why @austinpop recommended the following design for you:

---- Inbound from router <-> (port 1) Server (port 2) <-> Switch <-> Endpoint. ---

The bridge configured on the server would isolate your noisy router from your endpoint. And the (audiophile) switch would reclock the signal (with a good clock) just before the Endpoint. It's a much better solution than what you are thinking. 

 

Isolation can be achieved in multiple ways - you can use fiber optic connection, you can configure network bridge on your server, you can use a filter like the gigafoil v4, there are medical ethernet filters like the EMO EN-70HD or Baaske, etc. Some work better than others. Some kill dynamics, others work miracles. There is no consensus what's best, because it is system/environment dependent. My choice at the moment would be a pair of Sonore opticalModules with good LPS and single-mode long range 1Gbps transceivers. But once the EtherRegen is released it would be my choice for isolation. Obviously you need to maintain clean power after the isolation, so good LPS's play an important role there.

 

Reclocking is the second important factor, and that is even more complicated. The quality of the clock is impacted by the quality of the power, cables, vibrations, etc. You need to take really good care of all those factors. 

 

But those are the two major factors you need to worry about. When I looked at your design with the server bridge, first it did not make sense, because the Ethernet traffic would not flow like you thought. Also, your router and endpoint were sharing the same switch, which means there is no electrical noise isolation.

 

When the EtherRegen is released, if what they explain is true, that would be an easy solution to all your problems. And it would simplify things a lot. You put the Router and server on one end and you put the endpoint on the other end. No need for network bridging or anything complex.  I am a bit sceptical the EtherRegen would work so well and do a complete isolation plus excellent recklocking, but we would see. I really hope it does. 

 

Until we know it does for sure, the OCD in me is thinking about something like this, but that of course is a huge overkill: 

On 7/1/2019 at 12:46 PM, Nenon said:

 

If the claims Uptone Audio is making about their new switch are true, this can become a flat network with one switch only and no bridging...

[Nenon]

1 hour ago, George Hincapie said:

Can I ask generally, for those of you who are feeding your server output via ethernet direct to your streamer, and have upgraded the power supply for your music servers from SMPS to LPSU, what impact did that have? What was the scale of perceived improvement if there was any? Conversely, did any of you go through this process and fail to note any improvement?

  

Trying to decide whether to spend £1K on a LPSU for my server, and am trying myself in knots. I don't want to commit unless there's going to be a level of improvement to justify it.

 

Upgrading the stock SMPS to a good LPS is the single best thing you can do for your computer source. From all the thousands of combinations discussed in this forum, that's the one thing everyone agrees about. It is the most important thing to do for your computer after all! Whether spending £1K is worth, that's something only you can decide. To me, spending 20% on computer and 80% on LPS gives much better results than spending 80% on computer and 20% on a LPS.

 

The impact of a good LPS has been discussed many many times here. You just need to do some reading about it. 

[Nenon]

36 minutes ago, George Hincapie said:

Does that hold true regardless of how the data leaves the server to the streamer?

 

There is a short and a long answer to that. I'll go with the short answer as we need to get back on topic in this thread: YES. 

[Nenon]

Has anyone heard any updates from SOTM about the motherboard they are designing? There was some discussions earlier in this thread. I was curious if anyone knows how close they are to completing it?

[Nenon]

On 7/25/2019 at 11:08 AM, Nenon said:

I am thinking about building another server with the gen3 Ryzen and doing a side by side comparison with my highly tweaked i9-9900K / AsROCK Phantom Gaming server. It would be an interesting test and will try to keep as many things constant as possible. 

From a quick search I am leaning towards the ASUS ROG Strix X470-I. Any recommendations? 

 

After spending a couple of weeks with a AMD Ryzen 7  3700X / ASUS ROG Strix X470-I and comparing with the Intel i9-9900K / AsROCK Z390 Phantom Gaming-ITX/ac, here is what I think:

- If I am using the USB out of the motherboard, I prefer the Intel / AsROCK. It sounds better to my ears in my system than the AMD / ASUS. 

- I was introduced to the Pink Faun I2S card and their ultraOCXO clock. Unfortunately I could not get that card to work on the Intel server, but it worked on the AMD. I prefer by a huge margin the sound of the AMD with the PinkFaun I2S.

- I compared the PinkFaun I2S with two other I had with me - JCAT USB card and Audiobyte Hydra (USB to I2S output), all fed by the same external LPS. I liked the PinkFaun a lot better. I wish I had more USB cards to compare with. It would have been nice to compare with a PinkFaun USB with ultraOCXO clock, but I did not have one.

- One thing I learned from all those tests is the faster you process the signal, the better it sounds (assuming you have a good LPS of course). More cores, more cache, higher speed - the more the better. Looking forward to try an AMD Ryzen 9 3900X when it gets available at normal prices. 

 

There is something really good about not using USB at all. I guess the simplicity of getting the signal out of the (low latency) PCIe slot and converting it to I2S with a really low noise clock pays out. Guitar strings and piano has never sounded so good out of my digital rig. Actually, I always thought they sounded really really good with my USB output. But my reference analog rig did a good job showing me they can sound better. With the I2S they get super close to my analog rig, perhaps even better in some cases.

 

If I was using USB, I would prefer the Intel / AsROCK. For me I2S without any USB conversions just sounds better, so I am going the AMD route. I will probably try a Ryzen 9 3900X whenever it becomes available.

Now the bad news. First, the PinkFaun I2S does not work on Intel. Second, it does not support DSD. Third, it only supports up to 192 kHz. For those of you with native DSD DACs who would like to upsample, that may not be for you. For me what I hear sounds a lot better, and I have made my choice. For now at least :). 

 

BTW, since I don't need the Intel i9-9900K / AsROCK Z390 Phantom Gaming-ITX/ac, PM me if you are looking for one. 

 

[Nenon]

3 hours ago, happydance said:

hi austin @austinpop

 

thank you for your earlier positive comments regarding NUC DNBE spanning over a year as one of the best solutions for endpoint/server. I want to know if you think there are any comparable endpoints all the way from a few hundred dollars to $20,000 +?

For example, you say it is better than the SOtM? So, maybe as good as anything up to $6000, and after that small incremental sound improvements on other units if you want to pay thousands and thousands of dollars (i.e. diminishing returns) e.g. dCS ?

In the endpoint performance scale (say: top 20) where would you rank the DNBE? Number one being the best/ultimate performer.

Thanks!

 

Everybody likes charts and numbers. Hopefully you realize what you are asking is impossible. There are so many variables in a system. Sometimes it's difficult to definitely say between two components which one is better. Sometimes it is easy to say. For example, my $300 Allo DigiOne Signature does not sound nearly as good as my current computer where I have probably invested $5K+ in parts. And it would probably be a no brainer comparison in any room / any system. But there are many devices that just sound different. If they sound different, it would be a subjective opinion in the system and in the room of the reviewer. Perhaps in another system and another room, the reviewer might pick the other component. And that's with just two components. You are asking about top 20 rankings. You will never get an answer to that question.

There are almost 600 pages of super valuable information here. And there is no consensus on most things. Perhaps the only thing people agree about is that the power supply on the device connected to your DAC makes a difference. But then there are people who argue that the power supply on the streaming server does not make a difference. Read and get educated from other people's experience. Try some of the things and see what direction you want to take. 

As for the best servers, look up Taiko Extreme. That to me looks like the top server on the market. But I've never seen or heard one. Check out PinkFaun 2.16x, Innuos Statement, etc. There are plenty of options at any price level.