CORE AUDIO: Music server upgrade from "Barebone Entry" to DAIDO REFERENCE I9 Series

[PhilG]

Hi everybody! 

A few months ago I presented on this forum a DIY music server I had built around the "Barebone Entry" solution by CORE AUDIO  (see my previous post:

 <https://audiophilestyle.com/forums/topic/57190-core-audio-daido-barebone-entry-music-server/>

This setup worked very well. However, for the sake of a better overall balance (and also, I must admit, because of the "Audiophila" virus (harmless this one and highly friendly !), time had come to upgrade this system.

I was finally able to do this as I had the opportunity to purchase from CORE AUDIO, the Hungarian company of which I was already a customer, one of the test equipments that had been used in the development of their new DAIDO Reference i9 series music server.

To summarize it, it consists of the ASUS MAXIMUS XI GENE mATX motherboard reclocked by means of the new BiClock MasterClock (BCMC) developed by CORE AUDIO.

Here are some more details about this setup:

 

Hardwire:

 

Power supply / Enclosure :  Audio Linear ATX power supply (8 circuit full analogue low noise power supply), installed in a  Streacom FC-10 case.  It's only what I kept from my previous build (exit the old motherboard and CPU).

I had already modified the back plate of the FC-10, as explained in my previous post, so that I can install my JCAT FEMTO USB card vertically and so get rid of a riser cable.  The next pcie expansion slot near the USB card will receive soon the JCAT FEMTO NET card.

 

Transformers:  considering  the features of the new ASUS Maximus XI Gene motherboard and the limited space available in the FC-10 case, it's installation required the replacement of the initial block of 2 transformers for a set of 3 (or 4) transformers (300Va or 400V, at choice).  I opted for the 3 transformers version which proved to be powerful enough and quite satisfying SQ wise (fortunately I would say, as their installation being rather tricky because of the limited space inside the FC-10, I'm not really in a hurry to have to undo everything to try something else ). 

 

Motherboard:  ASUS MAXIMUS XI GENE (mATX)

 

Masterclock:  A few words about the BiClock MasterClock developed by CORE AUDIO, which is the cornerstone of this setup:

This new version comes with 2 pcs ultra femto TCXO (24 and 25 MHz) with 6 outputs (SIN and HCMOS).

These two TCXO are PLL controlled with 10MHz reference OCXO clock on the same board solidly   attached to a thick aluminium plate (it fits the board like a glove and ensures a total immunity against vibrations).

This solution is ideally suited to the GENE which has 2 distinct clocks, 1pc 24 MHz for motherboard and 1pc 25MHz for ethernet port.

It allows  a real flexibility if in the future one want to change his motherboard. This BCMC has also a sufficient number of outputs for each frequency, so that a reclocked USB card can be connected to it, for instance.

As for the 10MHz reference clock, it can receive the signal coming from the 10 MHz Ref clock of an other source component of CORE AUDIO's Reference series (the Karuna DDC) as well as the signal from an external 10MHz Reference Masterclock.

 

PSU for BCMC :  It's a tiny one circuit ultra low noise 12v PSU (also found on the KARUNA DDC). Noise is only 15 microV.

 

CPU:  I finaly went for the i9-9900T (35 w tdp) CPU.  Admittedly it's less powerful and is working at lower frequency than its big brothers but it still has the same amount of L3 cache at 16MB and honestly it works very well. Last but not least it is easier to cool down. I still think that the CPU cooling system of the FC-10, without being bad, is not it's forte (Streacom could (IMHO) have advantageously taken inspiration from what HDPLEX did on the H5 case with its 8 heatpipes system).

 As it is, in the FC-10 the temp of CPU with  this "T" version barely exceeds 38°- 40°C after 2 hours of listening. It's rather a good thing, all the embedded electronics in a fanless enclosure like the FC-10 already generates enough heat. No need to add more if you can avoid it.

 The other reason for this choice is obviously related to the software used for my server. No power-hungry software here (for instance no HQPlayer).

 

RAM:  As an alternative for the G Skill 16 Go RAM I have recently installed the newest Industrial APACER DDR4 RAM (8 Go). Yes it's ECC and therefore not supposed to work with anything other than a Xeon or AMD platform, but as Nenon has already noticed it in an other thread, it works (as non ECC) and it works well ;-)  As it is, I found better results (perhaps not quite night and day difference but noticeably better SQ nonetheless).

Only one shortcoming maybe for those who are after high tech aesthetics is the look of this RAM: it's rather simple (deceptively simple), a green strip with no flourishes at all. This is definitely not the "Star Wars" style of the G Skill ! But after all, SQ is much more important.

 

OS Drive: Due to the lack of space inside the FC-10 I replaced my previous Samsung  850 Pro SSD for a M2 nvme SSD (Samsung 870 Evo plus). To tell the truth, I did not notice any drawback to this change, on the contrary.  I took also the opportunity to disable all SATA ports in BIOS. 

 

 

About Software:  No significant changes have been made here compared to the previous configuration: Windows Server 2019 (Total Commander as shell), Audiophile Optimizer 3.0, Fidelizer Pro 2.5, JPLAY FEMTO (Ultrastream, hibernate, lowest latency settings).

All this is well known, a fairly classical configuration therefore, which has proven its effectiveness (at least for me, in my system). 

I have recently added Minority Clean (really interesting if not essential, rather an icing on the cake, I found the last 28 version to work very very well).

 

 

Let's now go to the listening results:

 

Moving from an aging system, but whose qualities and flaws have been assimilated, to a system that benefits from recent advances and much better performance can be somewhat disturbing (in the good sense of the word).

This is exactly what happened when I discovered what this trilogy brings: ASUS Maximus XI Gene - i9-9900T - and, last but not least, CORE AUDIO's  BiClock MasterClock. it's a" match made in heaven"! I know it is an often overused expression but here it is quite apropriate.   You'd have to be completely deaf not to realize that. Even my 65 years old ears could perceive it immediately !

I already got a glimpse of it during the few very brief listenings I had during the burn-in period,  but I didn't suspect what the burn-in process would bring.  Compared to my previous setup, there is more of  everything: more dynamics, more transparency, clearer highs, more liquid sound, more layers, more timbral acuracy, more  alive, etc.  better soundstage ?  not really as sondstage  width and depth are highly influenced by the installed pieces of Software and their settings, which have not changed, it was already very good with my previous setup and it is still very good.  If it's better it's rather in the sense of a more precise image.

 As I say this, remembering the last listening i did, I realize that what comes to mind immediately is the word "Precision". It's undoubtedly the key word here, precision is present everywhere, as is transparency, and there is no doubt that the culprit being responsible for this is none other than the BiClock MasterClock.

Another thing I also noticed  (and it's not so much about the sound as it is about the music and its presentation) is that the music gains in finesse, inner detail, inner warmth, ease and a certain "humanness".

 

To make a long story short, and conclude this flood of praise, I will simply say  that what I got with this upgraded server is more than a noticeable jump in SQ , it's really amazing. Obviously  it's far better than what I did get before, much closer to the real performance.

If I had to sum it up in just one sentence, I'd say that it delivers a perfect musical image together with rich, organic, real and expressive sound.

 

In any case, there's one thing that's revealing: during the listening sessions, I've just been thoroughly enjoying everything I played, whatever the style, and I found myself coming back to recordings that, for some reason, I appreciated less and whose richness I suddenly rediscovered.

 

Closing this long review, I remember that magical moment when I was about to discover what this new piece of equipment had to offer.  You have to choose a recording but not just any recording, it has to correspond to the mood of the moment but also to be somehow worthy of it. It's a bit of a "blank page syndrome", there are so many musical files, where to start with?

Given my hesitations, I left it to chance and chance did the right thing by choosing "Hummingbird", an unusual and masterful piano and dobro duet, by Helge Lien & Knut Hem (Ozella Music), a superb album that allowed to highlight the qualities of finesse, definition and veracity of timbres that a music server such as the CORE AUDIO DAIDO Reference i9 series can offer.

[PhilG]

Thank you Viktor, at least one reply, you saved my honour ;-) 

I was very close to break the absolute record for the longest unanswered post !

[PhilG]

2 hours ago, Nenon said:

 

@PhilG - did you replace the motherboard clocks yourself, or did you send the motherboard to Core Audio? 

 

No, of course not. It's way beyond my soldering skills. I can probably manage for things that fall within the ordinary scope of a DIYer but here we're dealing with something particularly delicate to do (because of the structure of the motherboard's PCBs) and the risk of failure and therefore destruction of the motherboard is not negligible. Given the price of these motherboards I am not ready to take it.

Anyway the work is not limited to just the motherboard.  Reclocking it requires precise settings on the Masterclock, tests, etc.

This is Core Audio's affair, it's their competence, that's why I left this to them, I had all interest in letting them do it.

[PhilG]

On ‎3‎/‎22‎/‎2020 at 9:00 AM, wittao said:

Nenon, thanks for your question!

 

Our JCAT card modification (only just the clock part) is in progress and of course agreed with Marcin (thanks again to him). I think the necessary time in this situation (corona) is about 1 month, we must wait for some important clock parts. If we will be ready i will post it asap. The JCAT card (USB and Ethernet Ultra Femto too) alone without any modification was the first card with what we worked with full satisfaction. We tried earlier PP (it is one of my favorite sounding and PangKaiYu is a one man army genius i think, but we have had a lot of reliability issue with the cards, so we dropped this brand), noname, SOtM (that's good). Some years ago we tried a lot of type, lot of brands, lot of chipsets and try in our music servers. We wanted hard to find what is the most important part in the USB cards. Of course the chipset is ok but the signal clock generators? Number of ports? PSU? Type of the PCB?

That was our experience: it is a package. The most important thing is a thoughtful, stable base. Thereafter we can find the good options to make a SQ better and modify them: PSU and clock control.

Our focus point was in the development phase (not for only just the USB cards and RJ45 cards, but for the whole music server) make a low noise PSU for all parts and real good reference clock generator (masterclock). We knew the change in SQ was going to be big, but not that big. Absolutely hearable changes, A-B-A tests wasn't necessary. 

We continued the development to make more improvement. Just an example: me measured the motherboard's built in clock, the phase noise was -55dB - -60dB@10Hz. The Crystec's specification (24 MHz Nr.597 TCXO for 10Hz) is -90dB! It is a very huge difference. We measured again Crystec's same clock but now built in the PCB, it was between -75-80dB@10Hz (it looks very good that the environment of the clock on the board and the card's structure is very important). Our own developed ultra femto TCXO has -105dB@10Hz built in the PCB "ALONE" without our MCMC (mono clock masterclock for one frequency) or BCMC (biclock masterclock for two frequencies) PLL with 10MHz reference signal.

(maybe it will be a question what kind of measure equipment we are using for a phase noise and jitter measurement: Rhode&Schwarz FSWP).

The other side is the PSU. There is a lot of type on the market but we must develop an own PSU what fits well to ATX standard voltages, plugs and have a very good noise parameters ( 14 - 80 uV on a different 3.3V to 12V voltages i think not too bad).

 

Otherwise after these results we wanted to move forward and connect our other source equipments to this solution. The question was: will it be better? In my opinion yes. It is important that our source equipments are running on the same clock signal with ultra low noise PSUs. That is why our Reference Series works together along the same line ( music server + USB/SPDIF converter + DAC )

 

Example1: Normal connection, without any signal clock route (of course inside the equipments are 10 MHz OCXO masterclock signals between the parts but not between the equipments):

 

Example2: 

The full reclocked version: the 10 MHz reference signal (-127dB@10Hz) comes from the KARUNA Reference and the music server's built in reference signal switched OFF and of course our dac get a better 24.576 signal too.

 

 

 

Your other question was:  inside the clock signal's cabeling is a problem or not. Not  There are a lot of industry standard to transfer clock signals without any problem between equipments and inside the equipments. Of course 5-10-25m distance is a hard question but it is an extreme situation in a home audio industry. 

 

Thank you for your open mind I hope i answered well your question.

 

 

 

It's very informative and thoughtful, Viktor.  Above all, you've lifted a corner of the veil on what should be of great  interest for JCAT card users (we know there are many of them on this forum) and more generally all those who are looking for that last bit of extra for SQ. 

Good news and small pleasures are becoming quite rare these days, so let's not deny ourselves our pleasure! 

As for me, I will wait patiently (not so easy) until that @#$%^&*corona has calmed down and activity has returned to normal, to find out more, especially about the results of the listening tests.
When one knows moreover that a JCAT dedicated PSU for FEMTO USB and NET cards should be released soon, as announced by Marcin, one can say that everything is being done so that users may be in paradise. :-)