Building a DIY Music Server

[MarcelNL]

CPU temp verified? Ran a memory check?

 

[MarcelNL]

I have used the Asus ROG Crosshair VII Extreme for a bit, also drawn to it by it;s power capability!

It sounded a bit better then the Gigabyte Aorus Extreme I'm now again using, but the issue was that the archtecture was not designed for audio server but gaming (one SSD direct to CPU slot and too few PCIe slots)

It lost its magical smoke due to a short, I'm lucky that the second Taiko ATX survived!

 

It's perfectly possible to use more powerful AMD Ryzen CPU's, I'm listening to a 16 core version right now using a 300W heatsink.

 

meanwhile the small test drive I gave the dual Xeon Asus C632 Sage MB learned me there is a Nirvana...even with fans blaring loud it sounded so much more refined in the mid range....

 

Waiting for both of the 300 Watt heatsinks to be routed...

 

 

 

[MarcelNL]

7 hours ago, Mops911 said:

Can we also record defeats here ;-) I think I just killed a i9-13900T deliding it...

 

 

It is for science....

We can start a separate thread 'Server builds KIA memorial'

 

I'll list ma Ryzen 5 a 7 and the Asus ROG Crosshair VII Extreme as casualties

[MarcelNL]

33 minutes ago, Nenon said:

 

I can't believe it has been almost fours years since these experiments. Time goes by so fast! Nostalgic indeed. 

 

Obviously, I stopped these experiments after buying my Taiko Extreme. But I haven't stopped poking around with various audio experiments. In fact, that freed up some time for a lot more DIY. I've actually never slowed down on my experiments. I have a new room - done a lot of measurements and a lot of treatment. I have built new speakers - a 1500 lbs. horn based speakers system based on commercial product but with a lot of DIY. With that, I did some custom subwoofer preamps/amps. Did some more work on my DIY DAC. Building a killer tube-based preamp. The list (of DIY projects) goes on. I also got the Taiko switch and router. 

 

I do like sharing more about these projects as I work on them. It's just that this thread is not the right place and they are not "music server" related. I can start new threads, but I think they will go silent very quickly and lost in the Internet space. I can do a blog, but who reads blogs nowadays. I have been tempted to start a YouTube channel on a few occasions. That's the new way of sharing things. But I have been too busy and I am afraid a YouTube channel will take too much of my time. We'll see what 2024 will bring us :). 

you could start a DIY thread at WBF, there used to be/are some folks into DIY-ing serious stuff!

[MarcelNL]

quick question, how much capacitance after each choke?

Adding more after the last choke helps cutting voltage swing, other than that there hould be no issue IMO.

you need three capacitor banks, one right after the rectifier, one after the forst choke and the larger/;argest after the second choke

 

[MarcelNL]

here is the spec sheet;

https://taikoaudio.com/taiko-2020/product/taiko-audio-dc-dc-atx/?v=796834e7a283

[MarcelNL]

How much capacitance after the rectifier? you need some capacitance to make the rectifier work properly. 

Ultimately you should be looking at something like;

 

rectifier> 10.000uF (or 22.000uF) choke 47.000uF (or more) choke 200.000uF (or more) 

 

The higher the last Capacitor is the lower your voltage swing at startup (you likely won't be able to measure that peak with a DMM) 

[MarcelNL]

If you know the power rating for the heatsink you'll get an idea if it'll work, you definitely want cooler temperatures than what 'typical computer folks' are running. The audible difference between <40'C and >50'C is too large to be ignored

 

Problem I see with a hybrid approach (clever idea!) is that the electronics working in the background may well have an effect on sound as well, but that can be tested quite easily. I found the Hypex softstart module audible enough to get rid of it.

 

 

[MarcelNL]

14 hours ago, seeteeyou said:

https://www.whatsbestforum.com/threads/introducing-olympus-olympus-i-o-a-new-perspective-on-modern-music-playback.37939/

 

Maybe it's more likely to have their own custom-made motherboard this time?

 

1st / 2nd Gen Xeon Scalable + WS C621E SAGE might be considered a bit long in the tooth as we're approaching 2024, while the latest products from both AMD and Intel would naturally result in better latency if they're compared to previous generations.

 

https://www.whatsbestforum.com/threads/qsa-lanedri-series.36078/page-25#post-918081

 

Even though we might get a little help from that kinda flagship PSU mentioned above, it's hard to tell if any DIY products were even comparable to the performance of Olympus + Olympus I/O.

even more so sonce nobody has heard the thing...

 

I recal Taiko tinkered with DIY design MB but left that track for the C621 sage platform for sound quality, I'd be highly surprised if they now made their own MB but we'll see.

 

 

[MarcelNL]

I;d say let's wait and hear...

 

All this faff (my words) about QSA does nothing for me, the idea that wors copper sounds better after a magical treatment than copper with less impurities makes ME think it's at best a symptom treatment and not a cure. Statements like 'air insulators work best with the treatment are IMO unrelated to the treatment.

Moreover, a magical treatment of a conductor using a box can encompass what; heat? High Frequency, High Voltage, High current or a combination of any of the previous, I may miss a few here...but it really sounds like that magical plug you need to put in your car and save 30% fuel instantly.

 

Metal Active Treatment by Yan Tsick Man is not exactly well documented other than in some raves, what I found on YT does not instill a sense of 'want' in me as what I'm hearing is the sort of glorified HIFI you typically hear on audio shows demo-ing music that always sounds impressive but not informative (selecting stuff with huge contrast between various instruments and little emotion).

 

Not that I'm not interested in what the process does, heck we've been doing far more crazy stuff with cables...

 

For now I'm hapy to stick with the Sage and Xeon Silvers, they will last looooong even if they are not available new there are plenty around.

 

Tinkering with a Motherboard? Its multilayer and quite well sorted as it is. My experiments with X570 MB's showed me that more phases are better, yet once I turned on the Sage (without ANY OS tinkering or load balancing) that idea went largely down the drain and I currently think that carefully balancing concepts, ideas and improvements is what is needed, and that takes loads of time and lots of listening.

 

Nothing like XDMI NSM? I haven't heard it either. I did hear the the LMS based predecessor on an Extreme and that did not make me think it had the edge over Linux/Daphile (also LMS based) on my Ryzen 16 core build at that time (granted at a show sound rarely, well rather never, is at it's best).

 

 

 

[MarcelNL]

for reference, Taiko just released pricing of their new products, XDMI is as I understand it what used to be TACDA (the output card for the Extreme that might have been released for a.o. us DIY-ers.

What they did is interesting, they concluded that putting in out components on a separate (rechargable battery) power supply the results were better to a point where the PSU for the MB almost did not matter. So you end up with a server and a separate I/O module with the output of choice at two levels 

 

 

I think it's safe to say prices are in another league;

(Ex VAT/BTW/MwSt)

https://www.whatsbestforum.com/threads/introducing-olympus-olympus-i-o-a-new-perspective-on-modern-music-playback.37939/page-13

Export pricing means exclusive sales taxes/shipping where applicable.

 

Olympus Export Pricing:

Olympus server USB: € 52.000

Olympus I/O USB: € 24.000

Olympus server + Olympus I/O USB : € 72.800

Olympus server XDMI: € 65.400 / introduction discount € 5.000 -> € 60.400

Olympus I/O XDMI: € 37.400 / introduction discount € 5.000 -> € 32.400

Olympus server + Olympus I/O XDMI : € 86.200 / introduction discount € 5.000 -> € 81.200

Olympus I/O to add to an Extreme server: € 20.800 (return your old USB and Network card)

Olympus I/O XDMI to add to an Extreme server: € 35.800 / introduction discount € 5.000 -> € 30.800 (return your old USB and Network card)

 

 

 

[MarcelNL]

11 minutes ago, RickyV said:

Hi, I did some inquiry about the SOTM motherboard to a local dutch shop and this is the answer i got back.

 

 

Hi Rik

Thank you for your email and interest in SOtM motherboard.

SOtM makes fantastic products, but the motherboard has become a loss-making item for us.

Even though this is a standard motherboard, we have received the board back 4 times with a damaged CPU
 base. 

Pins were bent and of course it no longer worked. (See photo below of the last board we got back)

We have therefore become very hesitant to sell this board.

I looked it up, but this is not a special CPU foot, it is just standard material.

We have the network card and clock boards in stock.

Some customers modify their own motherboard. This also has the advantage that you can work with newer generations of motherboards/CPUs.

 

 

 

 

 

???? Does anybody recognize this problem?

 

 

damaging a Motherboard during insertion of the CPU or mounting the cooler is not too difficult for the un-initiated I guess...luckily there are more options to purchase this MB

 

If this shop is interested in selling this product they should offer a service to a; install the CPU, or b: sell it with CPU installed, that cuts issues down to mounting cooling solutions...(it's too easy to damage the pins of a CPU when the heat conductive paste is cold and rock solid)

[MarcelNL]

Not having used this board I'd say that damaged sockets are most likely user error ...

[MarcelNL]

16 hours ago, seeteeyou said:

 

(WS C621E SAGE) 7 + 7 + 3 + 5 = 22

https://cdn.cs.1worldsync.com/95/97/9597d73d-7100-4541-8e62-3a464438230a.jpg

 

(ROG Crosshair VIII Extreme) 18 + 2 = 20

https://media.overclock3d.net/2023/10/asus-rog-x570-crosshair-viii-extreme-review_6527cbd90a5a3.jpeg

 

---

 

https://www.techpowerup.com/forums/threads/would-someone-explain-power-vrm-phases.236778/#post-3721865

 

And then we've got this similar example as an analogy

 

https://www.youtube.com/watch?v=WE9NIZShfZY&t=7m54s

 

WS C621E SAGE with dual processors = 76.16 × 56.6 × 2 mm²

 

https://en.wikichip.org/wiki/intel/cores/cascade_lake_sp

 

ROG Crosshair VIII Extreme with single processor = 40 × 40 mm²

 

https://en.wikichip.org/wiki/amd/packages/socket_am4

 

Many thanks for the good laugh since we're comparing apples to oranges

 

https://audiophilestyle.com/forums/topic/58164-building-a-diy-music-server/page/234/#comment-1263484

I'm not sure but we may well be talking about different things...the Asus Sage powers 2 CPUs. It might help me understand you better if you use more language and less quotes, I find it hard to read.

 

What my understanding about the Asus Sage MB is that it has ;

https://www.asus.com/motherboards-components/motherboards/workstation/ws-c621e-sage/techspec/

 

ASUS Digital Power Design :
- Industry leading Digital 7  + 1  Phase CPU Power Design

- Industry leading Digital 2  Phase DRAM Power Design

whether that is for each CPU or for both I don;t know, even when it is 7+1 per CPU I'm not seeing a huge advantage over the ROG Crosshair VIII Extreme, but please elaborate as I may be missing something crucial here.

 

[MarcelNL]

Anyone with hints to use the HDPlex Xeon het pipe adapter kit on the Asus Sage MB?  It seems the MB screws are too short to reach the heat pipe adapter mounting brackets....

 

 

[MarcelNL]

thanks!  Does that mean the metal botom plate needs to go? I haven't looked closely yet but it appears the threaded part sticks out of the MB CPU backing plate...or do you screw the M4 extenders on those to extend the existing thread?

 

The MB is real sensitive for getting all CPU pins connected! I go it to work with some active fan coolers, just to be sure everything is OK, troubeshooting compounded errors/issues gets more difficult^2  

 

 

[MarcelNL]

9 hours ago, SK8 said:

You should have some screws and stubs from Larry. For some reasons I didn't use them so I bought some original rivet nuts used in LGA3647 heatsinks from Taobao:

 

https://m.intl.taobao.com/detail/detail.html?id=644851547053&from=detail&spm=a212db.24065199

 

May be not easy for you guys to buy from there. The other option is get M4 rivet nuts like this from your local metal shops:

 

 

I did get some stuff, but it looks as if I do not have anything that fits, but I'll go through it some more and otherwise visit my trusted hardwarestore where I've seen standoffs.

 

On my Aorus MB I always ditched the screws attached to the CPU backingplate and tapped the holes to M4 and then inserted M4 threaded rods....that seems impossible here.

 

Thanks for the pointers guys, and mainly for confirming I am not doing anything obvious wrong ;-)

[MarcelNL]

think I found the M4 standoffs...I only got 4 of each length, that is likely why I did not further look at them. I'll find some more and have plenty nylon washers to make it work 

[MarcelNL]

heck I may have found something better, I was able to knock the 'bolts' out of the fan cooler...still need 4 M4 standoffs 

[MarcelNL]

2 hours ago, Exocer said:

Sorry I missed this - the m4 studs screw directly to the existing bracket. Larry sends thumb screws. The thumb screws go on top of the hex studs.

 

Here is an example of a hex stud Larry sent me that was WAY too long, a 10mm+6mm male female hex stud:

 

6mm+6mm sits right below the brace. Put one on each side and tighten all the way down. You won’t break the motherboard and you’ll make perfect contact on the cpu pins (ymmv).

thanks mate, I figured it out and ordered (and received) the extension studs.

[MarcelNL]

11 hours ago, Exocer said:

Sorry I missed this - the m4 studs screw directly to the existing bracket. Larry sends thumb screws. The thumb screws go on top of the hex studs.

 

Here is an example of a hex stud Larry sent me that was WAY too long, a 10mm+6mm male female hex stud:

 

6mm+6mm sits right below the brace. Put one on each side and tighten all the way down. You won’t break the motherboard and you’ll make perfect contact on the cpu pins (ymmv).

apparently not all M4 studs are made equal, th eones I ordered do not fit through the holes in the brackets...back to the drawing board....

 

Thinking about simply filing them down to fit through, the bracket is probably not large enough to drill larger holes and maintain strength.

[MarcelNL]

37 minutes ago, Exocer said:

I can ship you 4 of my spares. uxcell M4 Hex Brass Standoff Spacer 6mm+6mm Male-Female Threaded Pillar Screw Nut Kit for PCB Motherboard Computer Circuit Board, 20 Sets https://a.co/d/2poc7jw
 

is what you need, assuming you already have 5mm hex stud for the 4 corner positions.

thanks a lot for the very kind offer, I'll be OK after grinding down the nut part of the stud ...it's just a mm or so larger than the 4 I got from Larry with the kit, the Dremel should make short work of that.

 

see below

 

 

 

 

 

[MarcelNL]

On 1/23/2024 at 8:01 PM, abase said:

..are there any suggestions for a DDR5 Ram?
I plan to build my DIY-Server arround the "ASUS ROG MAXIMUS Z790 DARK HERO" (up sampling HQPlayer)

have a look at the Griggs audio website, they have been using DDR5 for a while now, I see they use specific G-Skill sticks

[MarcelNL]

Meanwhile I have  lift off, now the great BIOS battle begins...

 

Anyone who can tell me if I need to manually switch off the fan control or that the MB is fine with no fans connected?

(currently have the heatpipe adapter mounted with the fans loosely lying on top, heat production is not as dramatic as with the Ryzen 16 core)

 

[MarcelNL]

thanks! great news! I got a little paranoid after fighting the MB using a PCIe Videocard that did not work (MB jumper set to internal VGA), then two RAM sticks 'error training RAM" and FF on the display that I translated into fan fail without looking up the code....

 

I'll be fighting my way through the BIOS...A bit worried about the RAM sticks as those were 2 of the 4 that previously worked without any hiccup (before adding 4 more sticks)