Jump to content
IGNORED

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


Message added by The Computer Audiophile

Important and useful information about this thread

Posting guidelines

History and index of useful posts

Most important: please realize this thread is about bleeding edge experimentation and discovery. No one has The Answer™. If you are not into tweaking, just know that you can have a musically satisfying system without doing any of the nutty things we do here.

Recommended Posts

On 1/27/2018 at 9:59 PM, lmitche said:

I have the same 12 pin pico converter here and it must be powering the normal 3.3, 5 and 12 volt pins on the ATX header.  Where did you learn that the board would run with just the 4 pin 12 volt header connected? Most single voltage DC powered boards use a normal DC barrell connector for power.

 

On 1/28/2018 at 12:06 AM, seeteeyou said:

From what I could tell from looking at the pictures, they're dedicating one of them (all the way to the right with black+yellow wires) for the 4-pin molex connector (12V) that's hidden under the giant heatsink.

 

Now it's somewhat to tricky to understand how they're getting things done for the ATX connector, let's just take a quick look at the pin-out first but please ignore the colors assigned to the corresponding pins

 

I got a response back from Supermicro's technical department it was'nt very detailed but basically you can power their boards from either the 12V DC 4 pin connector or ATX connector and also both connections if using an ATX supply.  When you use the ATX connector 20/24 pin the 4 pin input can be used as a 12V 'backup' supplying extra juice when the use is excessive. Therefore in Using the pico connector this is essentially the ATX route combined with the 12v 4 pin.

They couldn't give me specifics as to how the extra 12V is used or where its diverted to, when using the ATX connection.

I think for our purposes avoiding the pico connector and connecting directly to the ATX source is the best way to get clean current to any board. You will need the 3 rails (3.3v, 12v and 5 v) and T off from that to your atx connector and basically test it out. Depending on your board and uses the current can get high so you might need to parralel your lt 3045 boards and use big heatsinks. As most of us are powering hard discs, USB cards and other devices seperately the current demand will not be great especially, with a low current processor.

I guess if your using SOTM output cards directly powered then the differences between the 2 options won't be huge.

edit: @seeteeyou just seen your above post :)

Link to comment
3 minutes ago, rickca said:

All the ATX boards I've used require both the 24-pin and the 4-pin.  Of course, they are designed for more powerful CPUs than a J1900.  Is this Supermicro mini-ITX board circuit design different than a standard ATX board?  I realize it's a SoC without a separate chipset.

Its a bit different because you can power it only with the 4 pin 12V input. You can also power it with the 24pin ATX input or both although this is not stated in the manual supermicro confirmed this to me. Some of their board use the 12V input for 'extra' current when demand gets high. Whether there is some switching going on or a parrallel circuit its not clear.

Link to comment
2 minutes ago, rickca said:

OK but that doesn't answer my question.  Sure, there may be alternatives that sound as good as the Supermicro.  The question is why are some motherboards less noisy than others?

It's a good point it seems the lower powered boards sound good - less noise generated, also the single chip processors seem to sound very good- less circuitry going on, and it must also be the build quality, server boards are built for continuous use and reliability in mind, they seem to also have better quality regulation at their outputs.

Link to comment
2 hours ago, lmitche said:

The whole 24 pin connector only pulls 18 watts with Hqplayer upsampling running. The 12 volt cpu connector pulls ~70 watts or < 5% of the rated power.

Looks like the titanium supply was the weak link. From those measurements you could feed the CPU with a 6-10amp linear psu and the 24pin atx with a 3 rail linear PSU 3.3/5/12v around 2-3 amps, (you can measure each output more precisely) eg, lt3045 boards and get rid of the dc to dc converter.

Link to comment
6 minutes ago, lmitche said:

Just to be clear, we are talking about a machine capable of upsampling to dsd512 using Hqplayer. 

I understood that, I don't believe the low power approach is the only feasable one. You might get away with less than 8 amps, really depends on the peak current draw also. Longdog audio make a very good server supply 12V /12amp it was around £600 but they've recently made an upgraded version at around £900 better all round specs. You could ask Sean Jacobs for a quote, 4 rail supply custom transformer that would really make your system sing. He has a variety of options price wise.

I appreciate the cost factor involved also, there are ready made boards for lt 3045 and lm317 for the higher current, combine that with a custom transformer and you'll have a supply for a fraction of the cost.

I'm a bit sceptical with DC/DC converters since they are still sharing the initial input and are ultimately governed by the quality of the supply input to them. In the case of Innuos they did feed that pico converter with a high quality source and the CPU it 'seems' seperately with a similar rail.

Not knocking HDPlex if their converter is as good as you say I accept that and am pleased that you got such an improvement.

I noticed that a few companies are now using medical grade switch mode power supplies such as those from TDK and upgrading components on them such as capacitors etc, the 'medical' part meaning filtration such as chokes on the ac input emi filtering lower ripple currents etc.

You could try a medical grade ac filter on the input of the titanium that might lower the noise floor a bit.

Link to comment
48 minutes ago, vortecjr said:

I can tell you from speaking with Andrew that he recommends an i5 at minimum and prefers an i7. The difference being that you want an i7 with a large music collection and with DSP. His i7 is only 11 watts so nothing to obsess over regarding noise.

My i7 7700t is 35W ( doesnt use this much with audio) not cheap bought it with Roon in mind upsampling etc, a sort of bridge between low power and the higher powered 90W processors. To make it sound good I turned off Hyper threading, stoppped it from going into turbo boost mode and ran it at a constant speed, turned off some other 'features in bios. So now its basically a celeron :D The quad core j1900 runs at 2GHz can upsample a bit and speeds along nicely. For larger libraries SSD's help but even on my slow HDD it manages well.

53 minutes ago, vortecjr said:

Also, you said that the tX-USBexp does not draw current form the motherboard. You might want to double check this with SOtM. I asked them about this a long time ago, but I can't remember with certainty what they said

Theres a switch on the rear panel to swap from internal to external power, a linear ps makes it sing.

Link to comment
5 hours ago, seeteeyou said:

Basically the "holy grail" for powering Supermicro X10SBA-L would require two different sources of power, one for 24-pin ATX connector and then another for 4-pin CPU power connector respectively

If you have this board, I would recommend using a 20 pin atx connector not 24 or simply dont connect up the last 4 pins. These are essentially to power hungry PCIE graphics cards, etc. So if your not using these in your audio setup best use, the power elswhere.

Link to comment
6 minutes ago, yellowblue said:

I am more than satisfied with the Cybershaft clock. No need for the really expensive SOTM clock.  

I would assume SOTM did their homework pricewise and are aware of the competition, they did take a while to release it. Lets wait for the reviews. ( the first one will not be from me :) ) I think you need a balanced system to reveal the benefits, mine isn't.

Link to comment
36 minutes ago, lateboomer said:

Thank you for all the suggestions.  I think I can stick with the x10sba-l mboard and also get one Startech 4-port dual channel USB 3.1 Gen2 card as Rickca and also a 1A 5v lt3045 board to power the card. Then i could use one ethernet usb nic to connect with this card to realize ethernet input and disable mboard lan. Lmitche, may i know what nic you are using?

The USB outputs and NIC ports are very good on this motherboard, USB links directly to the celeron and the NIC also directly to the Celeron PCIE interface. Why would you place a huge amount of components in this path? Is there a specific reason for the Startech card?

Edit: Ok I've seen your previous post but there is no guarantee that your new NIC card will outperform the motherboards NIC as its very good.

Link to comment
  • 2 weeks later...
26 minutes ago, rickca said:

I still wonder how much of the SOtM sound is attributable to their filtering vs the sCLK-EX.  

I'm struggling a bit with the SATA 2 filter, when first pluggged in I thought WOW bigger sound stage more detail. When analysed further it seemed that the upper mids were boosted and a slight loss of dynamics. When taken out sound stage flatter but the overall sound more energetic.

Link to comment
1 hour ago, ElviaCaprice said:

I don't see these far two different needs for mobo coming together here.  Low clean separate power/upgraded clocking, small as possible mobo yet capable to carry out minimum OS (Windows 10) needs without some proprietary OS, will fit my needs.  Chord's possible future mini MK2 being more important. 

It will need to have socket for user selection of processors, high or low power, also optane support both memory and ssd/ storage via PCIE interface is a good idea, I see that as the future for audio, enough PCIE lanes and sockets to cope, upgraded clocks and capacitors, a connection for direct memory power is good. The modular approach mentioned earlier is a good idea. I agree it will be difficult to please everyone and there are a large choice of boards available that can be modified, eg. the Asus board that Antipodes have chosen- maybe someone could open one up and take a peek.

Link to comment
18 minutes ago, Confused said:

Almost everything seems to be good, but to me the treble seems to be a little overemphasised, and has a slightly unpleasant 'glassy' characteristic to it. 

It sounds awful at first and needs 200hrs burn in to shine ie. leave roon radio playing whilst you are away.

Its not a 'bright' sound if compared to something like Ultrarendu. The SOTM gear has a lot of filtration especially the sPS500 and IMO it is a little too rolled off in the High frequencies, so its either burn in or a problem elsewhere eg. lush cables seem to add body and warmth to the sound and I assume you have the REF10 connected up as that will calm things down a bit.

Although the microrendu is less detailed, it has a certain musicality that is difficult to match.

Link to comment
1 hour ago, Confused said:

That said, it has spent most of this time powered up but idle, that is not actually streaming.

Definitely needs to output a signal, so components are stressed /fully warm up, capacitors charge and discharge, chips & resistors etc - those heatsinks need to get hot.

 

22 minutes ago, afrancois said:

I have a sMS-200Ultra, sPS-500 and IR. I find this combination also a bit bright when connecting the sPS-500 directly to the sMS-200Ultra. That's why I've put an LPS-1 in between. The sPS-500 is now powering two LPS-1's.

Quite surprised by your's and Mozes findings and I don't dispute them -I suppose system synergy is a must, I have both LPS1 and sPS500 and in my test LPS1 had the strongest treble which I prefered the Sps500 sounded slightly recessed, however there was a a glare and coursness in the upper mid with LPS1 compared to the sPS liquid smoothness and a much stonger bass punch and definition with the sPS500.Things change for the better with 2 LPS1's the lack of amps seem's to count so comparing 2 LPS1's with a sPS500 might be fairer.

Link to comment
27 minutes ago, welldone said:

SR4 is more expensive than LPS 1.2 actually. I'm do curious about the SPS 500 too. Which one is the best power supply among those ones (SR4, LPS 1.2 and SPS500)?

I would say it depends what they are driving and system synergy. The sPS500 is a 5 AMP supply and can be used to power a bigger variety of items eg Music servers, Dacs and can be used with a Y cable to power a couple of items. For lower powered use it 'seems'

the other 2 might have the edge, but a detailed comparison between the 3 is needed.

Link to comment
1 hour ago, Confused said:

I would not say I have a thin sounding system, it's just a touch of HF harshness and a slight digital edge that is bugging me a little.

Hopefully the burn in will have effect. As for the sPS500 I found the stock DC cable a bit 'gritty' this can definitely be improved by their OFC copper DC cable and also Ghent audio make a star quad version with Oyaide plugs at a cheaper price but it's very good, smoother tonality in both cases. I also have a Kimber mains lead and either a braded or shielded 'non bright sounding' mains cable will help also. It can also be helped further with choice of USB cables eg, Lush cables mentioned earlier. SOTM's Cat 7 network cable is very good but pricey.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...