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

[numlog]

EDIT: sorry, I mean NVMe M.2 drives

 

has anyone compared the sound difference between an M.2 stick drive and SATA SSD?

 

I have only used SATA SSD for audio PC because I already owned one, It also fairly easily allowed use of external supply with them.

I had bought a cheap SATA PCI card and modified that to use external supply too, and was considering buying a hifi SATA cable to complete the optimisation of SATA storage... but it turns out that with a card adapter like this M.2 can be powered externally too which I thought wasnt possible. Aside from this critical advantage SATA had over M.2 , looking at the adapter it also appears that it natively connects to PCI which is a pretty substantial... can upgrading clocks or using better power supply for a SATA PCI controller ever beat a native PCI connection?

Better yet no cables, another huge problem area of SATA eliminated.

And then the actual question I was wondering: Could M.2 sticks actually sound better than SSDs under the same conditions?

Some have said in this thread SATA is better than m.2 with power filters etc. but what about without it? both types using the PC PSU?

 

 

 

[numlog]

29 minutes ago, sandyk said:

 

Motherboard power will almost certainly be noisier.

than SATA power cable directly from PSU? 

so no way to know which technology is actually better for audio without external power for both which I assume no one has done yet?

 logically, for the reasons mentioned above it would be a suprise if it wasnt better (and the total cost to switch to m.2 would be less than 1 fancy SATA cable).

 

in the case of going to m.2 the old SSD could be used for music storage/playback too...  do you or anyone know if its better to have 2 drives for OS and music, pretending they are both identical, than everything on one?

 

Might try the M.2 drive to settle what the best form of storage is.

[numlog]

 

7 hours ago, Iving said:

Incorrect

As posted before at CA and elsewhere, Optane PCIe AIC gives splendid results. I supply dedicated PCIe power from HDPLEX. No other drives required. All CPU/Avoids chipset.

How do you supply PCIe power from the HDPLEX? do you mean with a molex, SATA etc. to the PCI card? I ask because looking at optane PCIe cards I dont see any power inputs on them and cant understand how you would easily power it.

Is the power usage much higher to support these high speeds?

[numlog]

11 minutes ago, Iving said:

molex on mobo inc PCIe (I have 2 CPU-direct PCIe AICs - Intel Optane SSD 900P 280Gb AIC in CPU-direct slot PCIEX16_1 and Intel Converged Network Adapter X540-T2 in no-switch slot PCIEX16_3 )

I don't know about power usage/consumption for Optane PCIe AIC. It is a non-issue for me because I don't subscribe to low-power-religion/dogma in the PC/CPU. Optane PCIe AIC is one reason I don't have a NUC/AL (case size). Second reason is I hate USB/use RedNet/Dante and AL doesn't support Dante. Third reason is, whilst I'm very interested in Larry's experiments and views, I'm not sure that it's NUC per se that's accounting for your SQ advantages. Fourth reason is I am happy with what I've got and wouldn't upscale everything without some kind of paradigm shift.

I'm posting pictures so you can see the molex arrangement - and also that I manage mains noise from the front end via Isotek Titan with independent outlets feeding digits via small Topaz IT and analogue via larger Topaz IT. (And also downstream - I have a lot of Isotek in the whole system and find that through trial and error I can find an optimal permutation of isolation and shunts that gives me excellent results.)

 

Nice, didnt know that PCI power input were a thing on some mobos.

[numlog]

6 minutes ago, rickca said:

How?

by cutting PCIe power pin traces and soldering on an external supply, a sloppy solution but it works. just after finding that this guy found another way. this one doesnt require modding but uses 5V external power when M.2 are powered with 3.3V( probably for the same reason the adapter uses 12V from PCIe slot, because PC 3.3V rail is expected to be dirty), with probably a DC-DC converter onboard to drop it down, so Id still prefer modify the adapter and use my own 3.3V supply.

[numlog]

 my Mobo has a M.2 Slot and using this slot is really the only way for me to use full speed m.2, because running both x16 pcie slots  at  x4 or more disables x1 slots, meaning I cant use USB card (and that would also mean downgrading to previous USB3.0 card cos the USB3.1 card uses a  x16 slot).

But the M.2 slot on the MoBo turns out to be best way. right beside it is a big 3.3v linear regulator, its nice the MoBo is already cleaning up the power a bit but it means there is an easy way to disconnect the MoBos power and connect external supply to the M.2 card

[numlog]

14 minutes ago, Johnseye said:

Has anyone compared using the USB 3 ports direct to the NUC board vs the optional USB 2 ports?  Or has anyone compared USB v2 to v3 in general?

I compared the integrated USB2 ,3.0 and 3.1 on a desktop Mobo and later 3.0 and 3.1 ports in form of PCIe cards with low noise power supply.  Even for USB2 devices the 3.0 and 3.1 do , to my ears at least, provide better sound quality in all cases, 3.1 being the best. 

 

[numlog]

On 1/16/2019 at 12:41 AM, Quadman said:

 

The optane M2SSD direct into the first M2 slot had perhaps a bit sharper leading edge, think square wave with a true 90 degree angle at the top and bottom, transients were very well defined, sharp without being harsh.  Startling in many ways.  Of course when this effect occurs you get the impression of increased pace and drive.  Bass was perhaps a bit cleaner, sharper leading edges, but maybe missing a bit of body and a touch of bloom.  Backgrounds were maybe a bit blacker, quieter.  Performers seemed a touch more separate but more individual meaning they seemed a touch less coherent as a group as a whole.  It was very involving and as I said I enjoyed listening to this set up a lot.

 

My old Intel SATA 2 SSD with external LPSU for power and a short 6" silver coboc stata cable seemed just a touch rounder on the leading edges, maybe a touch less sharp on transients, backgrounds maybe not quite as black but the body and palpable presence of the performers so real, so there, so human.  Very coherent performance.  Bass maybe a touch less tight but it had a touch more body and bloom, the resonance off a cello string is stunning.  I just cannot stop listening to this set up it is so human as I call it.  If I had to live with one server over the other I would choose mine with the sata 2 SSD.  Both systems have a very wide (beyond the speaker edges) deep (past my back wall), and high (above speakers) sound stage.  Both are very involving and as I started with, non-audiophiles would not hear the differences.

this mirrors my impression of an M.2 NVMe card vs externally powered 850 Evo SSD as OS drives, its hard to choose between them but with the M.2 powered by a basic LPS it is the preferred choice, with a slightly more dynamic and less grainy sound.

 

It turns out for music storage SSDs (or just the 850 Evo) may not be the best choice. using a 5400 rpm 2.5'' HDD as storage sounds better than the SSD,  the SSD sounds much colder/harder but there is no noticeable advantage in resolution or realism, it might be worse if anything. These small slow harddrives dont make much noise and dont need much power, it can replace the SATA SSD and use the same low noise power supply with no issues, the old OS is still installed on this SSD  so its possible a format could help with sound issues.

Will have to reevaluate this after some more improvements are made to the SATA interface

[numlog]

14 minutes ago, Quadman said:

 

I also use 5400 RPM server grade HDD's for music storage, wouldn't dream of using SSD's for music.

Is there an explanation to this or just something you discovered yourself? 

most impressions online give the ''no moving parts, electrically quiet'' explanation as to why SSDs are best.

[numlog]

 

if there is an option for it, have you tried changing voltage settings? 

Setting a  manual core voltage has had a big influence on SQ , the effect is different to clock speed changes.

there is limit to how low you can go before it becomes unstable, the limit is determined by the clock speed.

Important to disable SVID if the option is there

[numlog]

6 hours ago, rickca said:

I just found out from Startech that this card (PEXUSB314A2V) uses 12V from the optional LP4 molex or SATA power connector, not 5V.  This makes sense because the card is designed to work with power from the PCIe slot only ... and PCIe supplies only 3.3V and 12V.

 

Was the goal to provide clean 5V through the USB ports or power the USB chips? you can still achieve both with that card

[numlog]

1 hour ago, rickca said:

 

How can you achieve both?  Here is what I got from Startech

 

the original idea involved cutting the PCIe power pins, but you did mention before that you wouldnt like to do that.

Instead of cutting these pins you could try putting tape over them.

the 12V input uses a switching converter to get 5V and  introduces noise to the rest of the card even if you dont use the internal 5v output  so for the best results you need to stop the 12V getting into the card at all by taping the PCIe power.

To see if the tape is working test the green fuses at the ports with multimeter to check that the internal 5V power is disconnected.

you can wire a DC jack to the fuses for the external supply.

know that the card will give a false  ''current overload'' warning if there is no 5V supply at all.

(I discovered with the 2 port version if you dont want/need a 5V supply from the card you can connect the 3.3V supply to the disconnected 5V output on the ports and it will trick the USB chips into not giving this error, the reason is the 5V output is connected to ''current sense'' pins on the chip)

 

Powering the chips is not easy, especially with the 4 port version.
The ASM1142 USB chips need 3.3v and 1.2V, the 2 port card used PCIe 3.3V directly and surprisingly there was linear regulator on board that drops that down to 1.2V.

 

 problems with 4 port version:

there are more switching voltage converters visible on the 4 port version, i guess it might use it to convert 3.3V to 1.2V, not as ideal but it would still be worth using an external 3.3V. there is one advantage, the 1.2V switching converters would draw much less current than a 1.2V linear regulator so load on the 3.3V supply is lessened.

 

There are no big 3.3V input filter caps visible on the 4 port card (unless they are on the other side) that you could solder the external supply to, you would need to scrap away some of the solder mask on 3.3V supply traces to connect it.

[numlog]

this is what I mean about connecting 3.3V supply, the trace for 3.3V ont he back of the card is very big so its not too hard. dont forget there is a single 3,3V pin on the front too.

Its not necessary to do it on ground plane for GND, the SATA and MOLEX sockets have ground, I prefered the neater wiring (also the wires I used are way too thick).

You could even connect wires from the unused 5V and 3.3V input on SATA socket to this trace and USB port fuses and have a neat an easily disconnected external sata supply.

 

also see in the pic in the previous post, most of the unlabeled pins arent connected to the card, tape over these too, the bigger surface will help the tape stay in place.

 

[numlog]

Just now, rickca said:

@numlog your expertise is impressive!  Unfortunately, I am unable to do anything requiring fine motor skills like soldering due to my shaky hands.  I really appreciate you taking the time to document your solution.  I hope another user will benefit from your guidance.

sorry to hear that 

[numlog]

With full size desktops its typical for MoBo ground to be connected to the case in many places, it could be good idea to isolate it and have the case like a floating JSSG shield.

 

PSU mounting screw, MoBo standoff screws and PCIe bracket screws are some of the main connection points.

 

 

[numlog]

@jean-michel6 what about for your OS?

 

I have noticed the same thing, HDD (if you can manage the mechanical noise) is superior as storage...  better resolution (''plankton'') and more smooth/natural at the same time, SSD only advantage might be faster sound (timing) but with the noise/glare dynamics are poor so the benefit is lost.

 

HDD for OS is not somthing that has been discussed much if at all, do we really know if SSDs are better for OS aside from practical reasons (load speed, easier to power)?

[numlog]

On 3/2/2019 at 11:11 AM, sandyk said:

However, improving the PSU area to help prevent this getting back into the main PSU will markedly improve this area.

 

...

 

 That is why I regulate my PC's +12V supply down to 2 separate, much less noisy and isolated +5V supplies for my OS and Music internal SSDs.

Sandy, both of us mentioned using fully external supplies for the SSDs and HDD and still prefered HDD.

My drive supply has its own dedicated transformer, the isolation must be very good there.

These devices will still have to share ground directly with PC so how much isolation can you really get?

 

2 hours ago, the_doc735 said:

SSD - HDD   ...with one half in favour of ssd and the other in favour of HDD, does it really matter that much? Is it just a negligible thing (nit picking)?

internal/external drive  ...does it really matter?

vbus powered or external/independent power   ....does it really matter?

SSD or HDD is the least neglible of those 3 imo. plugging a SSD or HDD into a laptop with USB adapter the HDD is still preferred.

[numlog]

1 hour ago, sandyk said:

 When using additional external PSUs you are increasing the chances of stray earth loops.

 The voltage regulators that I am talking about are very low noise types such as the 40uV from ebay or better and share the same EARTH reference in the PC/server as the they use the internal +12V supply rail...

Neither are the choice of filter capacitor types and values used in the external PSUs are necessarily a good choice, with many PSUs these days using very low ESR filter capacitors which will exaggerate HF detail.

the characterisitic sound of the SSD remained even when using various power supplies for the SSD, including  internal 5V and internal 12V regulated down to 5V like you recommend. I believe it has to be the SSD.

 

I still use SSD for OS and recently discovered the best SQ could be achieved by using seperate SATA controller for the HDD (PCIe card) and SSD (internal MoBo SATA). the SATA SSD was able to outperform NVMe SSD in this configuration, I suspect a large of advantage of NVMe drives is seperation from SATA storage drives rather than increased speeds.

 

And I should mention the regulators used are TPS7A4701 with some effort to improve filtering , low ESR tantalum caps were removed the regulator boards, relatively large aluminium polymer caps were added to outputs. it sounds  better like this but Im not sure that will have the same effect as what describe,

 

1 hour ago, Chopin75 said:

 I have never heard of HDD better than SSD

Until recently you mean, there a 2 recent posts saying so.

Most of what you will read about it is only speculation/subjective reports, just because it goes against popular opinion doesnt make it fake.

As you point out its all plays back from RAM so storage should sound identical, but most will agree it doesnt.

 

 

[numlog]

Could you explain what you mean by your OS playing back from RAM?

 

I thought RAM playback is possible with a lot of music players,

its definitely the case with XXHE which I used to come to these conclusions.

[numlog]

15 minutes ago, Chopin75 said:

What I mean is that, I am using Euphony OS which allows one to load entire track into RAM (meaning extract the entire music file from storage) and the local software player is within the same OS where euphony is stored.  You basically select 100% buffer and then the entire track is loaded. The local software player I use is Euphony's own"Styles". I can also use HQplayer embedded which also do the same RAM playback. You are correct that many players can do this, nothing special. The Euphony also allows one to retrieve music files that are not stored in the same local SSD where the OS is. This basically mean the tracks are also loaded into the local SSD wherever it is stored from an external  drive and even network drive, if that makes it clear? 

ok, thanks for explaining. XXHE also has the feature of copying to OS drive to decompress lossless or lossy files before RAM. by default it will play uncompressed files (WAV,AIFF) from their source, so using WAVs can let you test the differences in sound between pre-RAM storage. You can install XXHE on the HDD itself to have it decompress and copy to RAM from there (this sounds best). 

[numlog]

Does anyone have experience with different CPUs?

 

The previous CPU was a 6600K, finding an ''upgrade'' without wasting money on specs that might actually degrade SQ is gonna be difficult.

I think my old CPU sounded best at its normal clock speed of 3.5GHz, was forcing it to operate at a non ''native'' clock speed with underclock causing it to sound worse or is it simply that faster CPUs sound better?

 

Does hyperthreading and/or number of cores matter?

 

AMD? At least at some point AMD was known for having a large number of virtual and intel with a few powerful physical cores.

AMD doesnt have integrated graphics, that could be a good thing since it means simpler chipset ,simpler CPU and less hardware on the MoBo.

and IME with the 6600k SQ is better using a GPU instead of integrated graphics.

[numlog]

Sandy was speaking more on the technical side of things when responding to ''SSDs are too noisy for audio use''

 

both types of drives have levels of measurable electrical noise, and if greater care is taken with the SSD it shouldnt be anymore of a problem than it is for the HDD.

 

But is this really a electrical noise problem? subjectively I would describe the SSD as ''noisy'' but I have no idea whats actually going on in the circuits (where as sandy does to a much greater extent), others are likely using the term subjectively aswell.

[numlog]

Well there is a case against HDDs, apparently they have higher jitter.

If the SSDs added noise is easily manageable like he suggests then on paper they are superior.

Since the SSD also sounded superior to him then that all ties together nicely.

 

[numlog]

12 hours ago, Chopin75 said:

I use AMD Ryzen 2700x 8-core ! so it does not need much hyperthread. According to pinkfaun, this is the chip they recommend for Music playback. It runs things very smoothly, runs DSD256 beautifully, including upsampling from PCM to DSD256 via HQplayer.. It appears Pinkfaun recommends lower hyperthread with more core is better.   I also think the lack of video card may be better if you plan to run headless, so it won't be a source of possible interference. But you may need a separate video card for setup using the screen. 

For Phasure's PC they advertise more cores as sounding better aswell. Hyperthreading is recommended by PeterSt at least for XXhighend, the software has a ''Processor Appointment Scheme'' option that is meant to optimize sound for different CPU situations including large number cores/threads ie the phasure PC.

 

It was a little disappointing to discover that its only the Ryzen 7 that lacks video, The MoBos will still all have integrated video hardware.

[numlog]

So I assume no one has switched  directly from an intel or AMD CPU to the other with otherwise mostly similar hardware

10 hours ago, sandyk said:

Yes, being mechanical devices that have their motors normally powered from an existing +12V supply rail, (especially the large capacity types) they are then highly likely to have less precise timing, and will also inject their motor current demand variations back into the +12V supply as well. This +12V supply may also be shared with other devices such as another HDD or even an Optical device.

 

There is another fly in the ointment here too. Many people augment the internal PSU with external Linear power supplies which have capacitive coupling back through the transformer to A.C. mains earth. (This is more of a problem with USB though) 

The existing would be using the 0 volts lead back to the internal PSU 's " earth plane" via fairly short leads, perhaps via a couple of 4 pin molex connectors. When you add an external linear PSU you then increase the 0 volts ( earth reference) lead length from the new PSU via it's relatively long 0 volts lead back to perhaps an earth reference derived from a 3 pin plug via perhaps an up to 1.8M long AC cable, plugged into (hopefully) an adjacent socket on the same powerboard. You may also have another variable where the 0 volts lead of the new linear PSU can be either  " floating" or connected to mains earth. This may result in additional RF/EMI pick up from external sources

 

Use an added low noise  internal voltage regulator powered by +12V from the existing PSU, then the earth resistance may then be almost the same length back to the internal PSU's 0 volts plane, or just a few inches longer than previously.

 It's ALWAYS a good idea to keep the  0 volts earth references as short as possible, which is another reason why it's best not to " daisy chain" added LT3045 (etc.) voltage regulators. This applies to both Digital and Analogue areas.

 

Alex

Cool, for my external supplies I may have inadvertently minimized that problem. toroids were used and the whole supply is inside the  PC case, with no mains earthing, the 0V path isnt too much longer with them.

 

 The PC PSU is outside case since it probably emits the most RFI/EMI, the MoBo and CPU power cables are very long anyway, this makes use of that length.You could argue keeping those power cables bundled up inside PC would minimize the intereference they pick up but I think PC PSU iteself is a bigger source.