Hi Peter,

Xt tweaks are identical between the versions. I made a reading mistake, idle the Lps consumes 67 watt, with 9e running 71 watt and with 1.168 93 Watts!

This is of course with the very low LPS efficiency so a normal switching PSU will show less difference.

Anyway the only different setting I could find was Q5 that is set to 1 in 1.168. Furthermore "processor scheme 3" is chosen for 9e and "core 3,4" for 1.168.
Unfortunately the PC hung after changing Q5 because of a bad SATA connection with the OS disk. I will see this weekend if I can lower power consumption on 1.168.

Cheers, Coen

A lttle late reply but relevant for those why might be tempted to try.

Simply put I was trying to use the bios settings to lower and fix the voltages on the Mobo components (cpu, memory, chipset). This resulted in a lower average power consumption and a change in sound.

Now the only consistent result I got was a computer that hangs on one of the future (windows 7) boots or freezes during normal operation. Resetting the voltage to auto fixes the boot/freeze problem. Not a very unnerving problem but this also makes it an unworkable tweak.

I'm not diving into it any further.

cheers, Coen

Hi there,

I saw this PSU tweak popping up in another thread and I already planned to share some notes. This didn't happen until now because I I lacked time and energy and I find it hard to form a consistent opinion.

Following a resolution I made with myself to once experiment with a Linear supply I decided in April to order a HDplex supply and matching 250Watt ATX converter. A sort of hybrid solution that at least in theory could bring some benefits. Expedition was fast and the whole lot looks like a well designed piece of kit and was able to fire up my Audio PC containing two harddisks (see signature).

Like I expected the power consumption at the powerstrip is significantly higher with the LPS, rising from 50Watts to 70Watts with my Win7 setup and 9e. Oddly the XX version 1.168 uses even 10 Watts MORE with (seemingly) identical settings. Consequently these extra 20/30Watts heat up the case.
At a "warm" day (20+ degrees in the room) I found it uncomfortably hot with 1.168 and also SQ seemed to suffer. The transformer started to make a noises (humming). Fortunately my 9e setup did not overheat or overload the box. My conclusion is that your setup should be limited to a 35-40 Watts power draw from the PC.

At the first audition I liked the midrange clarity and increased contrast to use a visual metaphor. Also there was a more natural flow to the music. However highs seemed a touch rough and the lows a little bloated and ill-defined. On closer inspection of my setting some had changed over time and I started to reset all to the settings in my signature. This gave quite a good sound that was different on a few aspects form what I remembered from the ATX switching supply.

In the past few weeks I have been changing settings, SATA connections and other stuff to make it work to my taste. The surface the PSU is placed upon really matters as do the SATA connections. Lately the latter seems less relevant because of the Diskless XXHE PC developments. I think it sounds quite nice now, but feel it has more potential and there are a few more experiments to conduct.

Interestingly at a time I reconnected the Seasonic out of frustration again. The sound was a tad smoother overall with a rough edge on the highs. It obviously lacked dynamics and flow that makes sound sound alive with the lPS. Still the LF and HF lack some focus and definition with the HDPLEX LPSU in comparison with the switcher. So this box is no perfect solution yet but a different balance that is more to my liking, at present.

These are a few observations and opinions on the effect on SQ at my side. For now I prefer the LPS over the Seasonic, but I'm also performing maintenance on my speakers and power amp that affect the sound and balance. I will reconduct comparisons between the SMPS and LPS when the rest of the system is stable again.

Please share your thoughts, questions and experiences on this route to "perfect sound".

cheers, Coen


 

That's a simple instruction even I can remember. This sounds like a fantastic option!
 
I was guessing you used the Music PC as a "diskless boot server" and the XXPC as "diskless client". For making that work you also need administrator level Windows skills, especially with an unsupported version. I guess both solutions have some similarities.

regards, Coen

Hi,

My mobo allows for lowering the voltage of the CPU, memory and Chipset. I usually leave the setting at auto where I presume it will follow the selected clock/multiplier rates (lower rates: lower voltage).

Now I do not really understand what is going on here. There is the power state control, the Operating system (XT Tweaks?) and BIOS all playing a role.

In a moment of Zen I set all to the lowest voltage setting where the PC allows for a boot while maintaining the clocks/multiplier settings. This is not without a reason, but not for this thread.

Good or not the voltage lowering did offer a very noticeable increase in clarity.
Also, as expected power consumption dropped quite a bit.

I just wondered:
is it allowed?
Do others have found a similar effect? 
Do I fix the voltage by selecting the BIOS setting and is the affect due to less supply variations?

thanks,

regards, Coen

Hi Joachim,

You do not need a speedy processor per se, but you need one that handles threads efficiently. This is where the multiple cores AND Hyperthreading come in and loads of L2/3 cache. You do not want your data to be clogged when waiting for a free thread or freeing up memory when precise timing matters. This is why shutting off stuff counts for SQ. To my knowledge there is no cpu test available for this quality. Otoh you can find the good properties on the datasheet.

My i5 is almost as "fast" as the (sb) i7 in arithmetic, yet is sometimes even below the performance border needed for xx. I never saw anyone complan with a low clockrate i7.

My next cpu will have HT and at least 4 cores. Maybe the new k series broadwell or even a Skylake for extra low power consumption.

Regards, Coen

Because only the resitors and the oven to keep all resistors at the same value will probably will cost as much as the NOS1 now.

Alternatives exist and there are ways to do this like the MSB and Total DAC, but not 26 bits linearity.

Regards, Coen

Hello Hans,

IIRC: The HDCD coding contains explicit information to select the "best" filter and as such modifies the impulse response constantly. Are you suggesting that the HF information in MQA is there to make the filter reproduce the time domain correctly?

Regards, Coen

Yes!

Salut, Coen

Hi Peter,

I just made the cables with only the red (power) wire DISconnected. It occurred to me that with the ground wire the balanced signal floats and as such can float above the logic level 1 or below 0, so no real data are transported.

Just like the YFS I had on the visit.

The "other ground" complicates the matter, I refer to the screen. In datatransportland the screen must be connected to quiet earth to behave properly. Otherwise you are better off without it (capacitive couplings of the wires to the screen). USB cables' screens are connected to the chassis of the two devices. Any grounding current for the balanced USB signal has to go though that path....

regards, Coen

Just for the record, this win8 stops situation has been bothering me too in my win8 experiments. Amongst other things (sq, balance).

Iirc it did improve with a reboot with the  Hiberfil file disabled. No persisting solution in the end.   

That was long ago.

Regards, Coen

You van add the dirac service(s) to the 'Keep running' list.

Regards, Coen

Hi Mani,

Thanks for the inspiring post, it sounds too good to be true especially since all is on windows 8, eh ! Especially your number two caught my attention since that describes exactly what I miss there.

Is Nick intending to present/discuss the Chronos PC on this forum or is it commercially off-limits so to speak and would CA be the proper place?

One thing I am curious about is the remaining influence of the USB cable. Your post suggests such a leap in SQ it is hard to imagine the USB link still makes a significant difference...

regards, Coen

Ultra caps are an interesting idea. They look good: practically no discharge noise, 'unlimited' charging cycles and fast chargable.
 
You have to understand they are NOT batteries, energy is NOT stored chemically but by a dielectric: they are capacitors. Discharging a capacitor means that their voltage is lowered every electron that leaves it. So they might be able to load and offload superfast (destroying your charging supply in the process) yet they aren't immediately useable an an audio circuit supply where the voltage ideally is constant. Batteries can be designed for this purpose: to provide a constant-as-possible voltage during charge and discharge. Capacitors by their nature cannot.
When you want to use ultra caps, some DC-DC (=switching) power convertor must be involved or you'll have to work with an overvoltage and regulator to compensate for the unavoidable voltage loss during discharge.

So I don't see the caps as the problem but more the complexity you have to introduce to be able to use them as constant voltage power source. Probably both the DC-C switcher as the Lin regulator are more noisy than the capacitor.

They may do have a place in the regular supply filtration with their small size and large capacitance. If I'm correct they were used extensively in the linear supply days of computers. Anyway they are on my shortlist for quieting a linear computer PSU.

regards, Coen

Great post Joachim,

It thoroughly reflects my findings. Unfortunately I never heard a recent Orelo in a proper room.

For now 45 Hz suits me fine no shortage of bass quality and quantity with my tiny low df amp.

Regards, Coen