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

[Superdad]

1 hour ago, davide256 said:

That its a 19V DC power input, even better... looks like I need to buy a NUC since my existing PC doesn't have UEFI which is needed for AL to boot. @lmitche  what NUC config

are you liking for AL and Roon server with  DSP use?

 

BTW, @lmitche very recently made a pretty detailed report about powering his NUC with the UltraCap LPS-1.2 at 12V:

 

Of course folks have been running NUCs--and Roon Nucleus, etc.--at 12V for a long time, with JS-2s or whatever else, but AFAIK Larry is the first to have powered a NUC from our little 1.1A toy.  

 

Anyway, my reason for posting was just to point out that 19V is far from a necessity for NUCs.  Despite what the back panel has etched, Intel's full spec documentation for virtually all NUC models of the past three years shows input range of 12-19V (and even a few 12-24V). It all gets dropped WAY down with DC-DC switching regulators on the logic boards anyway.

 

[Superdad]

3 hours ago, austinpop said:

Last night, to decompress from a Diwali party - I'm not a party guy!

 

Questionable...

a) Diwali was 10 days ago;

b) I bet you are a blast at parties!  

[Superdad]

6 hours ago, Ralf11 said:

maybe it was an Alternative Cowboy Diwali

 

Right. He does live in Austin after all. 

[Superdad]

1 hour ago, Johnseye said:

Will two lps-1.2's connected with a Y cable work?  No issues by combining two different power sources into one feed?  How many amps does the NUC require...kinda asking that to myself before I go look it up.

 

Both @lmitche and I are happily powering his AL-NUCs with a single UltraCap LPS-1.2 set to 12V. 

Nothing in these NUCs actually runs from even 12V—it all gets stepped down, by multiple DC-DC switching regulators, to 3.3V, 1.8V, 1.1V, etc.  No advantage to running from 19V.

 

[Hope this time I have phrased these facts in such a way that avoids @seeteeyou from shouting “shameless promotion.”  Larry made a nice report in this very thread a couple of weeks ago, but it quickly got buried by one of Roy’s long posts and responses to same. Clearly it got overlooked.]

[Superdad]

9 minutes ago, elan120 said:

I plan on trying  both LPS-1.2 and JS-2 set at 12VDC and see if there are differences...hopefully next week.

 

I’ve been going back and forth between LPS-1.2 and JS-2 for a couple of days. Both sound great. Will be very interested in your take between the two.  

[Superdad]

11 minutes ago, Dev said:

 

The Celeron NUC should be fine but not sure about i7 NUC. Its about 28watts (configurable down to 20watts). Moreover, earlier reports suggest that a beefier power supply on these NUCs sound better, eg. PH SR4 to SR7.

 

Of course. And the very available JS-2 can produce 7.2A at 12V, so many users have been using it for their i7 NUCs these past few years.  With wattage to spare for some other device on second output.

[Superdad]

1 hour ago, lmitche said:

Roy observed that the NUC sounds better at 19 volts. I aim to hear the benefit of the additional 7 volts.

 

I recall you saying that you though Roy was not able to do an apples-to-apples comparison of 12V versus 19V because his SR7 was only 19V and the 12V supply he compared to was either a different Hynes model or some other supply.

 

In any case, we look forward to hearing about the results of your experiment with a serial pair of UltraCap LPS-1.2 units, though I guess it will take some work to make a serial-wired 'Y' from the terrific Gotham wire you (and I now, thanks) use.  

[Superdad]

1 hour ago, lmitche said:

Cool, how about a 19 volt JS-2?

 

As mentioned on the phone, this is trivial to do and I have voltage set resistors arriving to do max. current tests at both 15V and 19V.  

The JS-2 won't be able to offer 7+ amps at 19V (as it does for 12V), but I'm hoping it will make it to 4.5A easily. Could simply substitute the set resistor on one of the four settings (presently 5/7/9/12V) on one of the output jacks.  19V/4.5A is still 86W.  Plenty for this class of NUCs.

Still, I am not yet convinced of any benefit to 19V.  We have loads of happy NUC and Roon Nucleus users at 12V. 

[Superdad]

56 minutes ago, rickca said:

Why does it take 3-4 attempts?  Is the 1.1A from the LPS-1.2 just barely enough juice?

 

Well I have the LarryNUC ()--with 8GB of RAM I think--and for mine it does not take multiple attempts to boot with UltraCap LPS-1.2.  In fact, I am able to boot with a wired Apple USB keyboard and an HDMI monitor attached--though I did disconnect them before starting playback.  Larry set mine up to come up as an NAA (no Roon for me yet).

 

I have not yet put my ammeter in-line to check actual current draw, but I've gotten pretty good at judging that by LPS-1.2 case temp (after hours in a give state.   At idle I think this NUC is drawing less than 0.5A because the LPS-1.2 case stays at the cool-to-very-mildly-warm temp indicative of its low-charge-current-mode (less than 0.5A load).  I was running a playlist on repeat (again as a NUC) last night and this morning the LPS-1.2 case temp has me guessing 0.75-0.8A playing as NAA.

 

When I have a chance I'll measure start-up load and load in other modes.  Having trouble booting with a keyboard or other devices is likely just an indication that the machine draws 0.8~0.9 during boot and that whatever extra loads are attached--thumb drive, keyboard, etc. are enough to go over a typical LPS-1.2's 1.14~1.18A trip threshold.

 

Oh, and we can not forget that whatever load the attached USB DAC or cleaner presents gets added continuously.  In my case it is an ISO REGEN, which adds 20mA (0.02A) 5VBUS load.  BTW, ISO REGEN is fantastic with this NUC.  Anyone here with an AL/NUC who has not tried it in a while or who does not have an ISO REGEN is welcome to contact me for a loan.  I admit that I power mine with its own LPS-1.2 and use the wonderful JSSG360 Gotham cables that I got from Larry (those are really something!).

[Superdad]

@lmitche So is my NUC (I think same as what is being spoken of) just a quad-core Pentium?  

If cores are the thing--even just for use as NAA or Roon endpoint--then I makes me wonder about the potential of Apple's new 2018 8-core mini.  I think someone here has booted AudioLinux on a Mac (I know Linux on a Mac is not difficult).  Rather pricey though.  And until we get one and sit down to figure out space and Apple's new mobo power connector, our Mac mini DC-conversion/Linear Fan Controller Kit (MMK) will not work with the 2018 minis. 

[Superdad]

16 minutes ago, austinpop said:

So Alex, how does it sound?

 

It sounds fantastic! 

 

Quote

Compared to what it replaced?

 

I have been so busy with work (partially thanks to you for making JS-2 sales skyrocket this fall ) that I just stuck the NUC into my system after several weeks of not listening at all.  I put it in on Sunday just to see if it would work right, because Larry was going to be available to assist and during a weekday would be impossible.  But the thing just worked right away and I have been grooving so much and getting used to it, that I have not unplugged it to put my old microRendu (V1.3) back in to compare (and I should warm that piece up to be fair).

So far the only comparing I have done is between the NUC powered by JS-2 versus UltraCap LPS-1.2.  And I was doing that on the first night before I was even used to the SQ in general.  So nothing definitive to state about that yet either.

 

I will say--and I told this to Larry the very first night--that this NUC makes much obvious things like HQ Player SRC filters and dither options.  And now I absolutely can not stand HQP PCM>DSD conversion.  For reference, folks should know I have a heavily modified Singxer SU-1 feeding I2S to a Holo Spring Level 3.  And for these sorts of tests I use really good acoustic recordings done in a real space, e.g. Louis Armstrong’s St. James Infirmary (from Satchmo Plays King Oliver), so lots of coronet dynamics, clarinet weaving, tall vocal and throat, piano, subtle cymbals.  Excellent for this sort of work. 

 

 

[Superdad]

2 hours ago, octaviars said:

I measured my NUC7i5BNH it has M.2 SSD for ROCK OS and second SSD for music.

This is with regular 44.1/16 no DSP

19Vdc average 0.57A (10.83W) and peak 1.5A
15Vdc average 0.69A (10.35W) and peak 2.1A
12Vdc average 0.81A (9.72W) and peak 2.8A

Average is during play to Roon endpoint. Peak was during start up.

 

Interesting how the efficiency is greater (lower wattage) at 12V than at 19V.  

[Superdad]

1 hour ago, greenleo said:

I'm not a DIY guy and never good at soldering.  I used to buy DC cables from Ghent and now find that there are many options like star quad, upocc, ..., 

 

Please advise me which cable should I use or your experience.

 

Suggestions of other DC cable makers are welcome.  You may pm me because this post may be off topic.

 

 

I have 3 of the Gotham GAC4 cables with shields configured to JSSG360 and they are outstanding.  Mine were hand made for me by @lmitche, but a couple of months ago Ghent (who seems to watch threads here or is prompted by @R1200CL ) has begun offering the Gotham GAC4/JSSG360.

https://www.ghentaudio.com/part/dc-gac4.html

I make no claim about if Ghent are hooking up the shields properly (the GAC4 is a confusing cable! https://gothamcables.com/en/gothamcables/starquad/11301gac41ultrapro).

The owner of Ghent sent me an e-mail--offering us samples (thank you for all the referrals for DC and 'Y' cords I guess), but I have not had a chance to reply with a list.

 

[Superdad]

11 hours ago, Dev said:

Roon Core/Roon Bridge vs LMS/Squeezelite vs Roon Core/Squeezelite:

 

Hi Dev:

You mention throwing HQ Player into the mix.  Larry built me a Celeron AL/NUC that I use as an NAA endpoint.  It sounds excellent but I am very curious to know how it (NAA) might compare to its use as a Squeezelite endpoint fed by LMS on my desktop.

 

My schedule is really jammed until January so I’d enjoy reading what you might have to say if you (or others) are able to make such a direct comparison.

 

Thanks,

—Alex C.

[Superdad]

23 hours ago, Superdad said:

Hi Dev:

You mention throwing HQ Player into the mix.  Larry built me a Celeron AL/NUC that I use as an NAA endpoint.  It sounds excellent but I am very curious to know how it (NAA) might compare to its use as a Squeezelite endpoint fed by LMS on my desktop.

 

Bump @Dev.

Can you compare HQ Player Desktop>AL/NAA versus LMS>AL/Squeezelite?

[Superdad]

2 hours ago, Middy said:

My 13 month olds got chicken pox, so just had time to swap out the NUC into the Akasa case between cuddles...

 

Exhausted new dad...

Dave 

 

Congrats on being a dad Dave.  But whatever you do, avoid getting chicken pox as an adult!  A long time ago I got it from our daughter when she was 8.  It was horrible—I had boils all over my face.   And now, 15 years later I am rewarded with shingles (like being stabbed with an ice pick).

You can love and comfort your sad toddler, but be careful!

[Superdad]

1 minute ago, lmitche said:

For the record,  I did not say futzing!

 

Yeah you probably said putzing.  

[Superdad]

32 minutes ago, Middy said:

With a million audiophiles with a million NUCs we may get the works of Shakespeare the 'musical'.

 

Are you saying we are monkeys with typewriters?  I resemble that remark! 

[Though Shakespeare musicals are something I would watch...]

[Superdad]

5 hours ago, lmitche said:

Who knows, 5 years from now things could be very different.

 

Or who knows, 5 years from now we could be talking again about loading tracks into RAMdisk--just as I was talking about 5 years ago:

https://www.computeraudiophile.com/forums/topic/17605-attention-current-mac-minia-users-boot-mavericks-from-an-sd-card-load-a-ramdisk-dismount-your-internal-sata-drives-and-pour-a-drink-for-the-musicians-walking-out-of-your-speakers/

 

As they say, "what goes around comes around."  

[Superdad]

7 minutes ago, austinpop said:

You had this all figured out 5 years ago?!

 

Hardly, but thanks.  (And I was partially just making a joke.)  Yet really a lot of the same folk here were part of that way-back-when journey. 

 

I've often repeated that most of computer audio front end improvement is about reducing the active processes (both h/w and s/w) of the DAC-connected computer.  For a long time the Mac mini was a good platform for doing this, but much beyond OS X Mavericks it became very difficult to slim the OS down to the level where the magic would happen.

Of course these minimalist and headless Linux installs (AL, Miska's NAA images, OrbiterOS, and others) make that much easier, and loading the whole OS to RAM is the icing on the cake. 

[Superdad]

2 hours ago, lmitche said:

Hang on, I'm dusting off my Bill Bryson British to American English dictionary!

 

Marmite is the Brit’s version of that Aussie favorite Vegemite.  Sorry, yuck to both! 

[Superdad]

31 minutes ago, rickca said:

@Superdad please comment.

 

Whatever floats peoples’ boat.  But I don’t buy it at all.  

[Superdad]

On 12/19/2018 at 8:57 AM, austinpop said:

Thanks, this is interesting.

 

One concern with Super capacitor supplies is that in the quest to deliver more current capacity, the higher number of capacitors can raise the output impedance, which has a very noticeable correlation with SQ. I notice in the writeup, they talk about the impedance of the output connector and the cable being factors, but I don't see (or did I just miss?) the output impedance of the supply itself?

 

If I recall right, the LPS-1.2 measures under 3mΩ up to 100kHz, but @Superdad can clarify, I'm sure.

 

Well that is a complicated area, and depends very much on how the supercaps (or any caps for that matter) are used.  I'll leave it to @JohnSwenson to explain in more detail some time, but simplified, but when using supercaps the way we do--where the active bank is only ever discharging (not simultaneously charging)--the ESR of the caps has no impact on output impedance.

 

When using supercaps just as a typical (though VERY large value) filter capacitor--where it is simultaneously charging and discharging--then yes, ESR have an effect on output impedance (typically at very low frequencies).

 

The FAQ of the Farad power supply page makes several extremely incorrect statements about our UltraCap units (no they don't name UpTone, but since we are the only bank-alternating supercap LPS on the market...), indicating that they do not understand our sophisticated design.  I just made a long post explaining how our product actually functions--here:

 

And a couple of posts above that one I also clarified valid question about leakage (sources, loops, and blocking).

 

Hope everyone has a wonderful holiday and New Years!

--Alex C.

 

 

 

[Superdad]

23 hours ago, seeteeyou said:

While three Connor-Winfield OH4 Series OCXO (better ones cost even more) alone might cost less than 200 bucks...

 

Actually, look at the -100 dBc/Hz phase noise @10Hz offset spec for the $50 20Mhz Connor-Winfield OH4 and you will see that better clocks (for audio applications) cost less, not more. 

Even the $9.60 Crystek CCHD-575 has lower phase noise/jitter than those C-W clocks. And that's from the published very conservative Crystek plots--actual production 575s run -108 to -112 at 10Hz (for a 25MHz part).

[Superdad]

14 hours ago, bobfa said:

The NUC is powered by the SMPS it came with.

 

Happy New Years Bob.  Why not run your NUC from your JS-2?  Many dozens of people do. 12V is just fine for those.

You’ll of course need a cable with 5.5mm x 2.5mm plug at the device end—versus 5.5mm x 2.1mm as you use with the UltraRendu.  I can fix you up with that.

 

Cheers,

—Alex C.