Investigating a full linear PSU for the XXHE PC and NOS1

[acg]

Mani has a linear PS on order for his PC..........

Which one?  The couple that I have seen are not that great.

[PeterSt]

Or don't work for everything and all, that I have seen. My "own" is among those. Project is officially still running, but not much news I'm afraid.

In other words, Anthony, no need to bet on one horse, especially when all run slow. You could catch up.
But you can only tell afterwards.

Regards,
Peter

[Scroobius]

Which one?  The couple that I have seen are not that great.

Mani did mention but I had my head buried in his PC at the time so do not recall - Mani?

[acg]

Well, with nothing happening on this front since I learned about the computer immunity of the NOS1a.  However, considering a few anecdotes of improvements from a computer LPS and Peters half-enthusiasm due to his workload with the NOS2 I have decided to revisit this project.  But the scope has changed a little with the technical advancements of both the Phasure system and linear regulators in general.  Ultimately, I want to find out just what it is about a linear power supply for the computer that can influence sound with our dacs, so I plan to go about things from two angles:

• An ATX LPS similar to those already on the market using the same voltage regulators, but in what is probably a better, lower noise way.
• An ATX LPS with new technology that surpasses existing supplies in terms of low noise, ripple, psrr, impedance et cetera.

Obviously this means that I will have to make at least two power supplies with different technology and see if they cause the system to sound different.  With the new technology I should be able to keep noise around 100 times lower than the tech. used in the commercially available power supplies.  Will this make a difference?  Is it the noise/ripple/psrr that makes a difference or is it simply about lowering the amount of noise fed back into the mains supply?  Is there a difference at all and is it work going all-out with the power supplies?

Anyway, I have been busy lately and am working on the module that will interface with the motherboard and control the power supplies.  I am quite confident that I can get it to work and once I do I can then play with the "good" power supplies (say 1mV RMS noise @8A and 4.5uV RMS noise @ 1A) and the "great" ones (say 15uV RMS noise @8A and 0.4uV RMS noise @ 1A).


Cheers,

Anthony

[PeterSt]

Very interested to learn about the results, Anthony.

If I can do something, of course you'll let me know ...

Best regards,
Peter

[AlainGr]

This is still something I would like to apply to my audio PC, since I believe that correcting problems right at the source may eliminate the need to apply "band aids" further down the chain...

Regards,

Alain

[toddn]

Hi Anthony,

I have 3 separate Acopian high grade power supply's with .25mv ripple and .005% line regulation, one for each MB voltage rail. (See attached spec. sheet.) I've been meaning to use your wiring & power up sequence research cheat sheet to give them a try, just haven't gotten around to it.

I'll try to get to it this weekend and report my results here. Maybe it will provide some additional useful information. I hope so anyway.

Todd

[PeterSt]

Todd, I have resized your image a bit, so it is better readable.

Thanks,
Peter

[PeterSt]

Todd, I only say this so that Anthony doesn't need to possibly embarrass someone :

Anthony talks about 14uV at 8 Amperes, while you talk about 250uV at 8 Amperes.
Also from your picture I can derive that at 8 Amperes it will merely be 1000uV.

then play with the "good" power supplies (say 1mV RMS noise @8A

1mV = 1000uV

Now that is out, please continue.

Great stuff ...
Peter


PS: And nobody claims that there will be a SQ difference.
Yet.

PPS: Getting something to run seems to be more crucial. At least at first (like in : mine didn't want to boot, no matter Anthony's sequence which I also tried).

[toddn]

Ah yes, I see now. micro volts vs. millivolts

Thank you Peter.

[PeterSt]

Seems quite unbelievable anyway, Todd.

[PeterSt]

PS:

Anthony talks about 14uV

No worries. That was a typo. Anthony said 15uV.

[acg]

Hi Anthony,

I have 3 separate Acopian high grade power supply's with .25mv ripple and .005% line regulation, one for each MB voltage rail. (See attached spec. sheet.) I've been meaning to use your wiring & power up sequence research cheat sheet to give them a try, just haven't gotten around to it.

I'll try to get to it this weekend and report my results here. Maybe it will provide some additional useful information. I hope so anyway.

Todd

Hi Todd,

Those Acopian supplies look fine, so I will be quite keen to hear how you go getting your motherboard to start and of course how everything sounds.


Cheers,

Anthony

[acg]

PS:

Anthony talks about 14uV

No worries. That was a typo. Anthony said 15uV.

Yeah 15uV which is quite an astounding noise level really for that amperage.  It is one thing to have a supply that carries that much noise at its outputs, but it is another thing to keep that low noise for the load which may be down a meter or more of cable.  A tricky part will be laying out the supplies inside the computer case so that there are either no leads (as in directly plug in the power supply to the sockets) or the leads are very short, say 10cm or thereabouts.  Of course the lower noise supplies are physically larger because of paralleled regulators and lots of capacitance so they are going to be more difficult to snuggle up next to the things that we want to power.

Ideally I will need a custom computer case to make room for these things...but we will see if it comes to that.

[PeterSt]

Hey all,

Here's once again a testimony of how complicated things are in Audio ...

So I have a 350W Teradak here at this moment. Beautifully built and it works with the ASRock MoBo.
It is completely silent (no transformer hum). Decent stuff, as it seems.

I decided to first let it work under Windows 8. I am not even sure why, but I guess I want to let work Windows 8 better because it probably *is* better, compared to Windows 10. About that in brief : In my view W8 is way more accurate BUT does not play music. The robots do that. In Windows 10 the musicians play.

OMG

Yeah, what ?

haha

Painstakingly killing. Djeezz, now THAT is bad.
S'es sharp as razors and I don't know what all was missing in the music. Everything, I'd say. Still it was not overly sharp, but it just did not make any sense to me.
I think I gave it 8 minutes, and then W10 was thrown at it.

As expected this made it somewhat better, but actually not at all for net result. Still it made no sense. However, I now could hear a bit better what was going on.
So what was going on ?

No, I couldn't really figure it out. I thought that perhaps a high dose of radiation was in order because actually the PSU was sitting quite close to the DAC and I wondered whether that could hurt.

What happened next gave me a small clue, but only in aftermath. So later more about it. But anyway, what happened is that I tried to move the unit which seemed impossible because of the too short power cord and while I really did not see anything happen in the PSU, moving the cord switched off the PC. Hmm ...

With a longer cord the PSU was moved away for a meter, an extra metal shield was placed and another 20 minutes of annoyance passed.
f*ck

Then I started to be smart and figured that while the normal switching PSU "allowed" to be without mains earth in my setup, that being for the better, nothing would guarantee that this LPSU is supposed to work well without earth connection.
And so another reboot was performed, the PSU now connected to mains earth.

This time a sort of forced rest (of mind) was introduced because after starting a first track I lost all control. Coincidence ? maybe. I had this once before, maybe a year ago. Nothing works, not even the monitor, but music played. So I let the playlist go (appeared to be ~30 minutes) and I noticed that I didn't want to pull the plug.
At the end of the playlist all started working again but by that time dinner was ready as well and I stopped the game.

My moral for this moment :
It is completely impossible to make an apples with apples comparison because I clearly had to change ground paths all over. And do notice that the no-earth for the (same) PC with switching PSU is the best (noise) measuring one; Apparantly this changes with the LPSU although I did not measure anything yesterday.

More tonight, but my hopes are not too high.


Then what I wanted to come back at :
The too short power cord appeared to be one which did not fit too well in the PSU's socket. Say it is a bit loose in there. Probably I created an "invisible" very short disconnect but now I wonder whether the PC was allowed to shut down by that. And notice that I could hear exactly when it happened, because music was playing ...

And so I started to think (later) whether the relative few "capacitor bank" isn't too low for decent operation. Actually I thought about speed and whether with such a low amount of capacitance the speed of a linear PSU is not too low for what's desired here. Uhm, whatever that is. But anyway notice that a switching PSU actually has no cap-buffers at all.
I am not looking at it at this moment of writing, but I think something like 12 or 16 of ~470uF is there for one voltage. That doesn't seem much to me, especially not for the higher current (spikes ?!) in order. And all voltages (the PCB's for it) are assembled the same (for capacitance buffer).

So far for now.
Peter