Building a DIY Music Server

[Soul Analogue]

23 hours ago, Tatomek7 said:

 

Hi @Nenon, Do you know what is power requirement to boot C621 MB? Hammond choke is rated with 5A, do you know, what current is taken from PSU by Taiko ATX board? In the meantime ( waiting for final information ) I'm considering to base rectifier on diodes, do you know what amperage will be needed?

 

I saw Tatomek7 quote on Nenon's post dated 16 May 

 

"The second version (aka v2) consists of: IEC inlet --> Fuse --> Soft start --> Transformer --> Mosfet rectifier --> Hammond choke --> 3 x Mundorf HC 22,000 uF (each bypassed with VCAP ODAM 0.1 uF film cap) --> Output connector. This is an improvement over v1 in my opinion."

 

this is a choke input (LC) filter 

 

and this thread developed with Hammond Chokes to be used in this kind of topology

Thats why I was asking if anyone had tried that

[Dev]

6 minutes ago, Soul Analogue said:

 

I saw Tatomek7 quote on Nenon's post dated 16 May 

 

"The second version (aka v2) consists of: IEC inlet --> Fuse --> Soft start --> Transformer --> Mosfet rectifier --> Hammond choke --> 3 x Mundorf HC 22,000 uF (each bypassed with VCAP ODAM 0.1 uF film cap) --> Output connector. This is an improvement over v1 in my opinion."

 

this is a choke input (LC) filter 

 

and this thread developed with Hammond Chokes to be used in this kind of topology

Thats why I was asking if had tried that

 

Ok, you have a point. I have been thinking CLC filter in my head all along and that is how I am planning to implement it with the Taiko DC-atx. I will let Nenon clarify. But even with CLC filter, the question still remains weather a 10mH/5A choke be sufficient ?

[Soul Analogue]

23 minutes ago, Dev said:

 

Ok, you have a point. I have been thinking CLC filter in my head all along and that is how I am planning to implement it with the Taiko DC-atx. I will let Nenon clarify. But even with CLC filter, the question still remains weather a 10mH/5A choke be sufficient ?

It depends on what you mean "sufficient"....even if the power drawn by the computer is higher than 5A...the choke core saturates and gets a hot ..... the ripple current cannot be controlled ..... but the PSU will still work if the voltage still within the operation voltage of the DC-ATX...

if you are going to power the full CAS machine, I would definitely suggest at least a 10A choke...

the lower DCR of higher current choke will give lower PSU impedance and hence better dynamics

[Nenon]

Don't forget this design is super efficient - we are using active mosfet rectification and GaN regulators. If it wasn't the transformer this whole thing would be 99%+ efficient. The transformer is the most inefficient component here. This choke (which btw is 10mH not 5mH) was not a good fit in a fully regulated LPS, but here is not a problem with my dual Xeon set up. But don't look at just that individual part. It's a well thought out design, carefully tested, and tweaked with the components I like over many listening tests. 

5A may seem like very little... But at 99% efficiency is not... and it works very good in practice. It's easy to implement, it sounds pretty good, and it's cheap to try (~$30).

 

With high current power supplies, we benefit from very low ripple. We have 5mV ripple with v2 and 5uV with v3. It's very illuminating to listen to v2 vs. v3 to understand exactly what the lower ripple does to the sound quality in a computer source. Everyone who has a chance should listen to that. V1 does not have a choke to limit the current. But I do prefer the sound of v2 and v3. 

 

A 10A choke would require a lot more capacitance not to overshoot. We need to be careful not to kill the Taiko ATX. I already killed one and have been super careful since that. I know Emile added protection after my incident, so maybe you won't kill the production units, but you would not have a reliable PSU unless you add that extra capacitance with a 10A choke. 

 

Lastly, just to clarify, I did not suggest a LC configuration. In the "Transformer --> Mosfet rectifier --> Hammond choke --> 3 x Mundorf HC 22,000 uF" version, there is a 22,000 uF Mundorf AG capacitor on the Mosfet rectifier board. Should have clarified that. 

[prinz]

component on the TAIKO, next to Mundorf Mlytic is not a choke. It is most probably a Mundorf EVO foil capacitor mounted into wooden holder, made from panzerholz.

 

Guys, do yourself a favor and learn how to use free tool called PSU Designer. It's easy, widely available and it will allow you to simulate the basic performance of your power supply. You can try different values, different schematics etc.

 

If someone wants to use LC type of filtering for whatever reason, the basic formula to calculate choke inductance is:

L = V (dc) / I (mA)

For example, if you have 30VDC after your bridge rectifier and your maximum load is 8A:

30V /8000mA = 3,75mH --> this is the minimum required choke value

The higher the load, the less inductance is needed.

 

 

 

 

 

[prinz]

This is my LPS prototype

for second choke I have used regular inductor for speaker crossover. I had 1.8mH on hand which fits and simulates very well.

 

[atxkyle]

RAM Recommendation and a Question

 

Did some server tweaking and listening tests this weekend. For folks that are limited to non-ECC RAM, one noteworthy outcome was that I REALLY like the sound of this RAM from ATP Electronics. Very natural and organic sounding, I literally felt tension release from my head and shoulders when I swapped to it. 

 

I’d been comparing it with this Apacer RAM which is CL19, although not sure how it compares to the most lauded Apacer RAM model numbers. But against this one it was not even close. The ATP sounded light years better. 

 

Unrelated Question: is everyone planning to have music storage onboard your server/player builds vs. streaming from a NAS?  And if so, has the community discovered ways for onboard storage to not destroy the sound quality?  

 

I bought a 1TB Samsung 980 Pro SSD (m.2 pcie) and adding it to my server did bad, filthy, naughty things to SQ. Especially when I had it internally, but even when attached externally via USB adapter. So bad it makes me very curious about how the Extreme and other servers have managed to neutralize the impact. 

[acg]

On 5/22/2021 at 12:43 AM, Nenon said:

A 10A choke would require a lot more capacitance not to overshoot. We need to be careful not to kill the Taiko ATX. I already killed one and have been super careful since that. I know Emile added protection after my incident, so maybe you won't kill the production units, but you would not have a reliable PSU unless you add that extra capacitance with a 10A choke. 

 

LL2773/5A may be a better choke option than the Hammond.  Much larger and much more inductance and depending on how you choose to wire the windings can be set at about 5A/65mH series or 10A/16mH parallel.  I'm not convinced that 5A is enough current...5A x 30V?? = 150watts before saturation for the Hammond.  At least the Lundahl would not saturate until at least 8A which would probably handle the poweron and off requirements.

[auricgoldfinger]

On 5/23/2021 at 10:52 PM, atxkyle said:

I bought a 1TB Samsung 980 Pro SSD (m.2 pcie) and adding it to my server did bad, filthy, naughty things to SQ. Especially when I had it internally, but even when attached externally via USB adapter. So bad it makes me very curious about how the Extreme and other servers have managed to neutralize the impact. 

 

How were you powering the external SSD?

[atxkyle]

3 minutes ago, auricgoldfinger said:

 

How were you powering the external SSD?

 

The entire NUC (including the SSD) is powered with a SR7T 19V.

When using it externally, it was just powered by the USB bus. I don't have an adapter that allows for external power right now, but woudl be open to this approach.  

[auricgoldfinger]

2 minutes ago, atxkyle said:

 

The entire NUC (including the SSD) is powered with a SR7T 19V

 

There was a time when some people were using external disk enclosures powered by a separate LPSU.  IIRC, the external enclosure was connected by USB.  Perhaps the second LPSU powering the external disk is key here.

[atxkyle]

57 minutes ago, auricgoldfinger said:

 

There was a time when some people were using external disk enclosures powered by a separate LPSU.  IIRC, the external enclosure was connected by USB.  Perhaps the second LPSU powering the external disk is key here.

 

Hadn't considered that, but I bet it makes a big difference. Does anyone know if most of the top-shelf servers are providing dedicated LPS rails for internal storage?

[MarcelNL]

6 minutes ago, atxkyle said:

 

Hadn't considered that, but I bet it makes a big difference. Does anyone know if most of the top-shelf servers are providing dedicated LPS rails for internal storage?

they do, using a separate SSD power supply did miracles for SQ in my NUC. In my current build the same applied to going from a so so external PSU to a proper LPS for the SSD, major improvement!

[RickyV]

Useful YouTube video:

 

 

[MarcelNL]

what is that gunk that is applied? (the computer I'm using right now has no sound)

[RickyV]

28 minutes ago, MarcelNL said:

what is that gunk that is applied? (the computer I'm using right now has no sound)

From google:

Flux is a chemical cleaning agent used before and during the soldering process of electronic components onto circuit boards. ... The flux also protects the metal surfaces from re-oxidation during soldering and helps the soldering process by altering the surface tension of the molten solder.

[MarcelNL]

thought it was flux, but was unable to listen to any explanation there might be

 

[bit01]

If you watch it on youtube there is a list of the materials/equipment used

[Gurkel]

On 5/2/2021 at 5:43 AM, auricgoldfinger said:

It seems to me this thread went completely off the rails about 3 pages ago.  I suggest deleting them would keep the thread properly focused.  This discussion rapidly devolved into the usual objective/subjective arguments.

Maybe the reason was because my data was transferred over WiFi which altered how it sounds and added a bit of a harsh sub tone. 

[ASRMichael]

RE Taiko ATX Upgrade
 

A few months ago I said to myself, I need to cut back on my audiophile spending....well my wife did! 😂 no seriously I did! 

 

Then I got a notification that the Taiko ATX was available to purchase! I hummed & hawed. Well not for long! Thinking “what happens if there’s was only one batch made & I miss out” I went and purchased the Taiko card! It arrived and I said ok how am I go to power this. Yes, without any plan! Crazy I know! 

 

Realising I’ve already overspent this year I had to come up with a cheap alternative using my existing LPS’s. 

 

What I’ve read from Nenon’s posts. I needed an unregulated LPS 24v or higher. 
 

To cut a long story short. 
 

I had 2 Sean Jacobs DC3’s, 11 rails in total

 

Rail 1 - 10a 12v rail

Rail 2/3/4 - ATX

Rail 5 - EtherRegen 

Rail 6 - Pink Faun Ultra USB

Rail 7 - JCAT Eth XE

Rail 8 - NUC (NAS)

Rail 9 - NUC (OXCO Board)

Rail 10 - Spare

Rail 11 - Spare

 

As you know all Sean’s rails are already double regulated. So what about double, double regulated? 

 

Here’s what I’ve ended up with;

 

Rail 1 - 24v - 10a feeding Taiko - Unregulated

Rail 2 - EtherRegen Double, double regulated 

Rail 3 - Pink Faun Ultra USB Double, double regulated 

Rail 4 - JCAT Eth XE Double, double regulated 

Rail 5 - NUC (NAS) double regulated 

Rail 6 - NUC (OXCO Board) Double regulated 

Rail 7 - Spare for 2nd EtherRegen - Double, double regulated

 

Effectively we used all the spare regulators on 12v CPU & ATX to double up to 4 regulators for Rail, 2,3,4,7. Added bonus.

 

I was very fortunate according to Sean that there was such a step down from 24v to 12v on the 10a rail. Having 24v at 10a feeding the Taiko was a bonus. 240w versus my 150w what I had. In addition because it’s now unregulated this means I can set my mobo psu settings for “long” at 240w and “short” to 368w (need to change the fuse to 16a to do that). That’s if I’m understanding Sean correctly. 

 

It’s going to be difficult to measure any SQ benefits from double, double regulated, but in theory it should. 
 

It’s very rare in this game to come up with a solution that doesn’t completely break the bank! Maybe just pure luck on my part! 
 

I’ve had to order new cables. Both cables & LPS’s should be up & running this weekend. 
 

Hopefully next year if funds permit I can purchase the V5 unregulated design from Nenon! Kidding, V3! 
 

I’ll feedback in 2-3 weeks once broken in! 
 

Thought I would share! 

 

[jean-michel6]

On 5/31/2021 at 11:24 AM, ASRMichael said:

Then I got a notification that the Taiko ATX was available to purchase! I hummed & hawed. Well not for long! Thinking “what happens if there’s was only one batch made & I miss out” I went and purchased the Taiko card!

Hi Michael , 

 

I did like you 😂. Very interested by your results . I have not yet installed my Taiko ATX . I will initially power it with a SJ DC3 19 V@5A . 
I am waiting on some feedback and may be I will use one of unregulated ps from Nenon . 
My server is a supermicro X10sdv-f ( Xeon 8 cores ar 2.1 MHz ) . This board is only fed through the atx there is no eps connector . I run only Daphile so no need for more horsepower . 
I have measured 4A at start up and about 3A during normal operations . 
 

Since you have an etherregen, I very very strongly recommend an external 10 MHz master clock . I did that recently and I call the result a small miracle . It does improve everything in sound quality and makes the sound so much more natural . It really worth every penny . 
I went for an afterdark Giesemann  the cost is 1700€ , I did try before an afterdark double crown loaned by a friend and it was also really excellent cost is more reasonable about 900€ 

 

Jean-Michel 

[ASRMichael]

7 hours ago, jean-michel6 said:

Hi Michael , 

 

I did like you 😂. Very interested by your results . I have not yet installed my Taiko ATX . I will initially power it with a SJ DC3 19 V@5A . 
I am waiting on some feedback and may be I will use one of unregulated ps from Nenon . 
My server is a supermicro X10sdv-f ( Xeon 8 cores ar 2.1 MHz ) . This board is only fed through the atx there is no eps connector . I run only Daphile so no need for more horsepower . 
I have measured 4A at start up and about 3A during normal operations . 
 

Since you have an etherregen, I very very strongly recommend an external 10 MHz master clock . I did that recently and I call the result a small miracle . It does improve everything in sound quality and makes the sound so much more natural . It really worth every penny . 
I went for an afterdark Giesemann  the cost is 1700€ , I did try before an afterdark double crown loaned by a friend and it was also really excellent cost is more reasonable about 900€ 

 

Jean-Michel 

Hopefully cables arrive this week and I’ll give early impressions.
 

I’ve decided not to spend anymore on network. I’m only using local files at present. No streaming services.

 

I had Edge X & OM before my EtherRegen. All powered by DC3 rails. I after 12 months in place I removed the Edge X and OM because it sounded better without them. This might be because my DC3 was powering 2 less devices, who knows! Also Euphony now has a feature to disconnect network. 
 

IMHO £1700 for a clock is a lot of money, could this amount of money be spent better elsewhere? 

[jean-michel6]

48 minutes ago, ASRMichael said:

Hopefully cables arrive this week and I’ll give early impressions.
 

I’ve decided not to spend anymore on network. I’m only using local files at present. No streaming services.

 

I had Edge X & OM before my EtherRegen. All powered by DC3 rails. I after 12 months in place I removed the Edge X and OM because it sounded better without them. This might be because my DC3 was powering 2 less devices, who knows! Also Euphony now has a feature to disconnect network. 
 

IMHO £1700 for a clock is a lot of money, could this amount of money be spent better elsewhere? 

In my case the music data goes this way 

local hdd (usb ) > x10sdv-f server > etherregen > ered dock player . 
I also play only local file and do not stream . The master clock is a huge progress because it delivers a very very noise free music data to the ereddock player .

In your case you go from 

Nas > etherregen > I9 server & player . 
The master clock may be not an as significant experiment as what I did . Anyhow if you have the opprtunity I recommend you to test it. 

[ASRMichael]

@Nenon whilst things are quite here at present, you posted previously you were going to post how to setup windows affinity etc on dual CPU’s. Using Asus sage board. 
 

I maybe see myself on a journey towards the dual cpu. Especially since I have the Taiko ATX card. 
 

I think users who plan to buy Taiko ATX, Unregulated PSU, Taiko case may be thinking about how to setup windows affinities etc. You posting how to do it may gather a lot more interest to go down that route. 
 

If you could do you write up that would be great. 
 

Cheers

[Nenon]

47 minutes ago, ASRMichael said:

@Nenon whilst things are quite here at present, you posted previously you were going to post how to setup windows affinity etc on dual CPU’s. Using Asus sage board. 
 

I maybe see myself on a journey towards the dual cpu. Especially since I have the Taiko ATX card. 
 

I think users who plan to buy Taiko ATX, Unregulated PSU, Taiko case may be thinking about how to setup windows affinities etc. You posting how to do it may gather a lot more interest to go down that route. 
 

If you could do you write up that would be great. 
 

Cheers

 

Yes, I was planning to do that when I get the chassis and an opportunity to build a new server. I can document step by step what I do this way. It has been a long time of tweaking on my dual CPU server, and I don't remember everything but will gather the various bits and pieces together once I start from scratch.

 

I noticed a quite significant variable while doing tests with my DIY server. Every time I turned it off to make some changes and back on as part of my A/B tests, it sounded better. It took me some time to understand why but it became clear after a few tests. The HDPlex chassis runs a bit hot after my dual CPU is powered one for several hours/days. When I turn it off and all the components cool down, my server sounds quite a bit better! I knew there was a difference but did not expect it to be that big. With that variable in mind I had to redo most of my tests. 

 

After living for several weeks with V3 of the unregulated LPS, I decided to switch back to V2. I heard exactly what I heard the last time. V3 gives you a very expansive soundstage but loses some clarity. If clarity and transparency is what you are after, go with V2. If you prefer to have soundstage way beyond the walls of your listening room, then go with V3.
I had a pretty hard time deciding what I prefer. But I picked V3 for me and my system at the end. The good thing is it’s easy to switch from V3 to V2 when I get bored :). The main point is that V3 should not be considered an upgrade to V2. They are just two differnet sounding versions. Many people may prefer V2. I like the big and expansive soundstage on V3! 

 

Asked Emile about the chassis and the transformers. 

Transformers are due end of July. Seems like the transformer company was also affected by the raw materials shortage and delayed the order by several months. 

Chassis - well, the raw aluminum shortage was a problem, but Emile thinks he has secured enough aluminum for his Extreme chassis and for our DIY chassis. The good news is that all these delays took us to the time when Taiko is receiving their new state of the art CNC machine (end of this month I'm told), so they would be able to produce the DIY chassis in-house now! 

I do believe however that this would bring this project till the end of the summer!