Building a DIY Music Server

[One and a half]

@Nenon,

"Done with the messy thermal compound paste that goes on the CPU, copper pipes, and all over you and the case!!! :)."

 

Yes, horrible stuff that ends up everywhere no matter how hard to keep it from anything other than heatsinks. Plain ol' WD40 works well to remove the gunk and clean up with IsoPropyl.

[One and a half]

6 minutes ago, Nenon said:

 

This best thing I've seen on my laptop screen this week :). 

Uh oh, CPU21 & CPU22 is approaching critical  

 

Would like to see the CPU temperatures if that's possible? 

[One and a half]

1 hour ago, OAudio said:

 

I think it would be good to see measured performance specifications provided as well.  

 

As a general observation there are some large ticket LPSs out there for audio systems but few vendors provide measured specs for products. At the price levels we are operating at, it's not unreasonable to be looking for similar measured performance to lab class LPSs from companies like R&S Tektronics and Agilent.

 

So ball park or better than:

Sub 50 uSec 60% load transient recovery times.

Sub 2mv rms ripple and noise.

Sub 2mv rail sag from 10 to 100% of rated load.

(For 6 to 10 amp rated supplies)

 

I know from developing supplies there are many layers to a good design but my experiance is that these measurements do correlate with SQ and could help better inform people's decisions.

 

 

 

 

 

 

 

Where is your ball park?

[One and a half]

22 hours ago, OAudio said:

 

I think it would be good to see measured performance specifications provided as well.  

 

As a general observation there are some large ticket LPSs out there for audio systems but few vendors provide measured specs for products. At the price levels we are operating at, it's not unreasonable to be looking for similar measured performance to lab class LPSs from companies like R&S Tektronics and Agilent.

 

So ball park or better than:

Sub 50 uSec 60% load transient recovery times.

Sub 2mv rms ripple and noise.

Sub 2mv rail sag from 10 to 100% of rated load.

(For 6 to 10 amp rated supplies)

 

I know from developing supplies there are many layers to a good design but my experiance is that these measurements do correlate with SQ and could help better inform people's decisions.

 

 

 

 

 

 

 

When it comes to linear supplies for hard disks, the tolerances are not *that* critical, but still important. 12V Linear power for micro-ATX boards need some thought, and your ball park figures will work. 

 

For (E)ATX boards, it's quite a different kettle of fish. There's 5V, 3.3V notably the 3.3V with 100mV max sag and lead compensation. Even the Power OK signal for a linear supply has specific timing so that the mobo understands what the PSU is up to.  Under short circuits or overloads, the power is limited to 240VA output for any output.

 

Intel ATX Power supply guidelines attached to this post refers. 

 

For ATX spec, the challenge for linear supplies is a daunting one (to an amateur PSU designer like me). 

atxps09_atx_pc_pow_supply.pdf

[One and a half]

On 1/14/2020 at 6:31 AM, Nenon said:

I don't know the price. It was a four rail Sean Jacobs DS3 LPS, similar to what Innuos uses. Probably best to contact Sean Jacobs for that. http://www.custom-hifi-cables.co.uk/home/power-supplies/dc3-power-supply

I asked Sean if he is willing to sell some boards to the DIY community, but he is currently away. We'll probably get an answer next week.

 

Many others have considered a balanced power feed on the AC side as having benefits of a noticeable reduction in the purging of noise. For this topic, I had thought of ways to produce a balanced DC supply for computer use. DC has the advantage that it is 0 Hz, so in theory there's nothing to cancel. However. The DC in power supplies contains garbage from other pieces of equipment connected to the AC network, coupled capacitively through linear elements such as power transistors on heatsinks, mainly from the PC's PSU.

 

A Balanced 'DC' rail is actually DC+ a small amount of AC. It's the AC that can be effectively dealt with a balanced network, any asymmetrical voltages are 'gone', leaving pure(er) DC. This is especially important in ATX supplies where high currents are involved.  There's also load noise to consider, would also benefit from a balanced supply.

 

How this would be setup, is the centre connection is at 0V earth potential, wired back in a star scheme to the earth of the wall outlet or balanced AC supply mid point.

The 5V would be +/- 2.5V, 12V +/-6, the tricky part is 3.3V, since many 3 terminal regulators have a minimum of 1.25V, although with a bit of work and thought, can be overcome. 

 

Linear supplies would work well, and SMPS, but finding 6V or 2.5V off the shelf are not so easy to source, unless custom built like from Acopian. Like with a linear supply the fixed switching frequencies on a balanced supply are asymmetric from an SMPS, so the noise factor could reduce nicely as well.

 

I'm just putting this out there as a point of discussion as an improvement to remove noise. 

[One and a half]

14 hours ago, basillus said:

Interesting, thanks for your input. You don’t have a picture/schematic about the balanced dc-rail setup, because I am not 100% sure that I have understand it correctly. 

Here's a low noise design, but it's only a guide to illustrate the purpose, the voltages and currents are incorrect for ATX use. 30,000uF is substantial for power delivery, must say, not much would get through that.

 

When the load is connected across +VA and -VA, the voltage is 26V (not zero). +13V is measured at the + and PGND1 and -13V also at PGND1.

PGND1 is connected further back at the transformer to earth (ground) PE conductor (AC 12V C.T.) 

Depending on cable lengths, say 300mm or more, the load side PGND1 needs a separate wire back to the transformer earth point.

To ensure the noise cancellation works effectively, the + and - voltages need to be trimmed so they are they the same. I would expect 1mV or less would be achievable especially with lead sensing. AC transformers can balance to within 2% and work effectively at these parameters.

[One and a half]

The 0V from the PSU is at earth potential. The -ve rails cannot be grounded since it will cause a short circuit.

 

The 0V in a computer power supply is often earthed, the two PC's I have are out of the box.  Grounding one end to earth of the 0V doesn't guarantee that it will be the same at the load end, since noise pickup will create potential differences.

By using a balanced supply and routing the 0V can effective noise cancellation occur.

 

[One and a half]

15 minutes ago, exeric2 said:

This is an interesting topic about transformer power. I'm interested in  the  Taiko ATX power supply.  It seems like the minimum requirement for transformer power rating would "not" be what is required for peak dissipation but for steady state power during the playing of music. Capacitance can supply peak loads. But that steady state power can change drastically depending on what voltage outputs of the ATX you are availing yourself of. If you are using it to power all the internal cards it would be significantly different than if you are powering the cards from an independent LPS.

 

What I'm getting at is this: I won't be getting the Asus Sage MB dual cpu solution for financial and available space reasons. I'll be sticking with an Asus Ryzen Crosshair 8 for now with a AMD 3950x processor.  I also plan on powering the USB and NIC cards from an outboard LPS (Paul Hynes SR5). I'm wondering if there are some smart people out there who would estimate the average power dissipation that would be the minimum required for a transformer powering just the ATX connection and a seperate LPS for the EPS connection, ( there are separate connectors for the Crosshair MB). Am also wondering what the total dissipation would be using the 12V coming from the Taiko ATX power when it is powering the CPU also.

 

Of course the implicit assumption is that Nenon using the minimum steady state transformer power dissipation for required for good sound might be scalable to other platforms with different power consumption.  That might be a totally bad assumption but would be really good to know one way or another.

The size of the power transformer is a trivial exercise compared with the requirements of the ATX power supply requirements.  The spec for that is available on the Web and lays down rules that the power supply must adhere to factors such as tolerance of voltage, delivery of current, and failure shut-down systems for the mobo in case the psu fails.

More than one 3 3V supplies for the CPU need to be connected in parallel and load share correctly, when some CPUs can take 45A this is somewhat not easy for a linear supply that also needs to have a ultra noise footprint for audio purposes. 

There's a trade-off for sure, where that ends up is not that yet defined.

[One and a half]

2 hours ago, Nenon said:

I also heard a Lumin U1 while picking up a Gryphon Essence amplifier to try at home. The U1 did not sound bad at all, and the thought about trying a better supply on it occurred to me right away. But it was in another system and hard to tell if it was good or just the rest of the system was pretty good. 

The Lumin X1 power supply can also be used on the U1 with a cable supplied to suit.  Any Lumin is heavily dependent on the source that's feeding it, that includes cables, and the music server. If I make changes to the source, the effect is noticeable right away on the U1.

[One and a half]

1 hour ago, Nenon said:

Yeah, I heard it with the X1. But I wondered how it would sound with something significantly better than the X1. 

Only limited by $. I measured the supply, well tried to, I think it's a multiple of 12V rails.