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

[Johnseye]

8 minutes ago, afrancois said:

The FS105 v3 however works with the MeanWell (7.5V) that comes with the LPS-1.

 

That's odd.  It works with the SMPS that powers the LPS-1, but not the LPS-1 itself which runs at the same voltage?

[Johnseye]

1 minute ago, Superdad said:

I tested and posted about this in the SMPS grounding thread last night. 

The Mean Well that works is 7.5V, whereas the highest setting of an LPS-1 is 7V at present. 

7V sees to be just outside of the range of the wide-range synchronous buck converter chip as implemented in the Netgear switch. 

 

Not wanting to read too much into this, but you said "at present".

[Johnseye]

Just now, LTG2010 said:

Definitely Option B.

Reclocking / daisy chaining with the txUSBultra may have a positive impact, but I would add a Master clock input and sell the txUSBUltra to part fund an ocx10 clock.

 

The master clock is still available with either option A or B.  Other than incorporating a master clock with the devices, are there any reasons you would choose B over A?  The cost of an sCLK-EX vs a txUSBultra aren't that different.  I'm not too worried about the cost.

[Johnseye]

17 minutes ago, limniscate said:

@Johnseye

 

SOtM found, to their surprise, that the tXUSBExp sounds even better with a tX-USB Ultra after it.

 

One of those things where there's little info available and I'll probably have to listen for myself.  I also use an ISO Regen, which I'm sure SOtM hasn't tried.  So, with an IR in the chain is the tx-USBultra necessary?  One other thought about my option A vs B is that if I ever want to use the txUSBultra outside of this scenario, meaning a later date with other hardware as just a reclock/filter or if I want to resell it, it would be easier.  The sCLK-EX board as a stand alone needs to be paired up with something and has little value as a stand alone device.

[Johnseye]

1 hour ago, rickca said:

That's great!  Which Asus motherboard were you using previously?

 

Asus Q170T/CSM

[Johnseye]

6 hours ago, lmitche said:

My problem with the DC input boards is that you lose control over the quality of the DC to DC power supply as it's embedded in the board. If you use standard ATX boards there are a wide range of choices that are eliminated here.

 

I've have used four different DC to DC solutions here. In the end a high end atx power supply sounds best.

 

The number of DC in boards available are much more limited.  Right now I don't know what improved the SQ.  It could have been the 3M sheet, it could be the Celeron proc, it could have been the RAM, but I doubt it.  It could also have been the components (caps) of the board.

 

Testing different ATX against DC boards is easy enough.  I've been using ATX boards with various non linear PSUs for years until this past year.  None of them impacted the SQ as much as when I started using the Hynes SR7.

 

Larry, I believe you're upsampling to DSD512, correct me if I'm wrong.  If you take that factor away does 48KHz PCM sound better in a high power PSU?  Also, what exactly qualifies a "high end" ATX PSU and why?  The Seasonic Prime?

 

 

10 minutes ago, ismewor said:

That is correct, and that doesn’t matter how good your 12 or 19v LPS are, you still limited and filter by the mobo dc section.

 

Well, it actually does matter how good your LPSU is.  The power you feed that DC input of the board will definitely impact the sound quality. 

[Johnseye]

3 hours ago, R1200CL said:

http://www.mini-box.com/picoPSU-150-XT should probably widen the choice of MB. 

 

Here some examples:

http://atechfabrication.com/MC-2500_system_order_form.htm

 

(Even with Sotm PCI card)

 

 

 

HDPlex makes a better quality version of the ATX adapter.  It's going to degrade the SQ more than a direct DC connection, but it does open things up to using ATX boards.

[Johnseye]

11 minutes ago, ismewor said:

Yes, of course good quality on LPS is always matter. What I mean are the improvements is limited by the mobo dc section. 

 

Now take it a step further. The quality of the L/PSU is going to have a bigger impact on SQ than the various DC sections of  motherboards. 

 

I can still use that HDPlex ATX adapter with my LPSU and it's going to degrade my sound more than using the on board DC in.  I can also use it to power a server powerful enough to process DSD512. It's a cheap test which one day I'll probably do.

[Johnseye]

5 minutes ago, lmitche said:

Johnseye, the point of my post was merely to point out that DC input motherboards embed the DC to DC PSUs. I have chosen to use standard ATX boards to give me more choice in which DC to DC PSUs is used.

 

You seem convinced that the embedded DC to DC transformers sound better.  That's cool. I haven't tested one vs. the other. It does seem to me that an $80 purpose built hdplex is likely to sound better then a $80 DC input motherboard with an embedded DC to DC PSU,  but I can't prove it.

 

Long ago  before moving to Hqplayer, I used a low power j1900 motherboard in my primary rig. During the build three different DC to DC supplies were tested, one that was bundled with an Antec case, the second a 160 watt pico PSU, and lastly the hdplex. The hdplex sounded equivalent to the Antec version.  Each DC to DC PSU was powered by a 19 volt Sigma 11 based power supply.

 

My current Hqplayer upsampling rig is powered by an evga ATX SMPS that is reviewed here:

 

http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=410

 

Prior to building the new rig, I tested the evga against the sigma 11 hdplex solution.  I could do this with the standard ATX motherboard.  The evga sounded much better.

 

With the same evga ATX PSU and upsampling with Hqplayer to dsd512 on a z170 based motherboard and i7-6700k processor is a great improvement from the low power rig.

 

I can't talk to 48khz pcm performance as my system is designed for dsd and pcm sounds awful.

 

That is the extent of my experience.

 

Understood.  The 48KHz reference was just an example of a frequency other than DSD256/512 and intended to address the question of what would a system sound like when you remove that requirement. It could just as well be 192 or 384 PCM.  I personally like PCM over DSD, but my DAC doesn't handle anything higher than DSD64 so I don't know what I don't know.  We're probably just looking at things from the perspective of our equipment.

 

Do you know how much noise is generated by this PSU and passed through to your server?  I still don't know what a makes an ATX PSU high end.  I wonder how the evga sounds compared to the SOtM PSU.

[Johnseye]

10 minutes ago, LTG2010 said:

Agreed both have their merits and differences but I will also add that the server or pc influences the sound, especially high frequency noise, emi, jitter etc. PCM can sound harsh in some systems and converting to DSD can alleviate this, but I found that the introduction of the SCLK - EX in my system pointed to jitter - poor clocking as the main cause. The addition of some SATA filters removed high frequency coloration, better power supply, EMI RFI shielding etc- Those same PCM tracks transformed into Music.

 

Very True

 

Sounds like using a noisy motherboard and high power switching PSU can lead to poor audio quality where one needs DSD to compensate. (that was tongue in cheek)

[Johnseye]

5 minutes ago, afrancois said:

Let's not forget that when you are talking about 'small' like for example the sMS-200 ultra, you should also say that it has a single purpose dedicated motherboard. So comparing a small PC with a small renderer is actually comparing apples with pears. I can follow Summit's reasoning when he says that a big PC could perform better than a small PC. Especially regarding EMI/RFI. It is always a good idea to have room on a motherboard to place your components. 

 

But theoretically we're only using these servers to play audio files.  If they are dual purpose PCs then any type of comparison is moot.  Our efforts here, and the entire purpose of this forum is computer audio.  There is no place for a discussion of a PC that is used as a game machine, video renderer or other use which conflicts with audio playback.  So what use is a big PC for audio purposes?  The only thing I know of that requires more processing power is DSD256/512 upsampling.  This should be the only need for a high speed multi core proc.  Power requirements to support this kind of proc are still low.  Managing the heat from that proc is more of a concern as a fan might be needed, but with a good heatsink might not be necessary.

[Johnseye]

22 minutes ago, afrancois said:

I think you misunderstood. Of course, audio streaming is single purpose, but a full-fledged PC isn't. A PC is designed to handle a magnitude of tasks and processes and all sort of stuff that has nothing to do with audio. That's why people are buying software to suppress as much of the un-needed processes. When I'm talking about dedicated and single purpose, I'm talking mainly about the hardware and not about the software. For me, a renderer has a simplified motherboard with mainly audio streaming in mind. What in fact is happening with the sCLK-EX board you are adding, is that this card takes much of the audio processing for its account and doing so by isolating itself as much as possible from the rest of the PC hardware. Is it a proven fact that a PC with the sCLK-EX is better than a server/renderer? I understand that the server/renderer solution needs, of course, some other things to be taken care of, like a good switch, galvanic separation, etc...

 

Ok, you're just pointing out that an OS or hardware designed for multiple purposes is not ideal.  Absolutely.  We're trying to minimize that by using the simplest, lowest power/noise components to a streaming PC as possible.  We're taking what's available to us today and modifying it because as of now there is no audio purpose built motherboard.  Same thing on the software side, with AO, Process Lasso, etc.  ROCK is the closest thing to an audio specific OS.

 

10 minutes ago, afrancois said:

If this is the case and I'm not saying that it isn't, it's time that SOTM starts selling PC's to the masses that are made to these principles and with their tXUSBexp PCIe card and sCLK-EX built-in.

 

 

They have, it just costs $4.5k.  And I'd guess their mobo isn't much different from what you get commercially.  No one has built a completely custom, mobo (that's not an endpoint) yet.

[Johnseye]

1 minute ago, afrancois said:

I'm anxiously awaiting your conclusions when you have your PC server/renderer ready. However, you can't compare it to the sMS-200 Ultra.

 

I have an sMS-200, but unless I have SOtM modify it, no it won't be an ultra.  If someone wants to let me borrow theirs for a comparison I'd be happy to share my opinion.

 

It will be another two weeks before I get the hardware back.  Not too long.

[Johnseye]

1 hour ago, rickca said:

OK I understand your point now.

 

@Johnseye did you decide between your options A and B yet?  My only input is that putting the sCLK-EX in the PC minimizes the length of the clock cables to the tX-USBexp and ethernet NICs. 

 

I'm leaning towards the tx-USBultra as the sCLK-EX source.  I've asked May if my switch uses the same voltage and frequency as the motherboard ethernet.  She will not know until it's all in hand.  If it is then my dilemma is gone.  If it is not then it's back to the same choice.  I think I'd prefer having the tx-USBultra than the modified switch.  Using the GS105 switch with the shunt will already lower the noise.  If I bridge my NAS with my server then the music from the NAS will be coming directly over the modified motherboard NIC taking the switch out of the loop except for Tidal streaming.  If I ever want to add another tx-USBexp for direct USB storage I can utilize the same tap as my first one.  It will also be good to hear how an sCLK-EX modifed tx-USBexp sounds compared to the tx-USBultra.  While Roy said there was some improvement by using both, you never know.

 

@afrancois I don't know what value an endpoint will bring vs the setup I'm planning.  As has been mentioned the sMS-200ultra is just fine, but it's not going to wipe away any impact from noise generated upstream.  I think we also beat this topic to death in the other post with TopQuark and I don't want to go down that path again. The bottom line is that the more clocks you can "upgrade" from the sCLK-EX, the better.

[Johnseye]

23 minutes ago, sandyk said:

 

 Not necessarily.

A larger PC often has a larger and higher quality ATX PSU, and has the ability to use additional low noise USB cards with special low noise power. It may also use a high quality soundcard with Coax SPDIF output to a good DAC, getting around the numerous problems, and expense  associated with getting the highest quality audio from USB.

 There is also the possibility of fitting additional small PSUs internally to supply improved AND isolated power to the various components such as SSDs .

Also, the RF/EMI from a much faster processor is pushed much higher in the spectrum where it is likely to cause fewer problems.

 

I'm still searching for that high quality ATX PSU.  So far the best quality power seems to be from either an SR7 or sPS-500.  A high power ATX PSU is going to put out more noise than one of these, especially if it has a fan.  One of these two HQ PSUs can also power a motherboard with a low noise USB card.  That's exactly what I'm doing.

 

I can't comment to a high quality soundcard with SPDIF.  I think this option's been left by the wayside.  There are more limitations with SPDIF.  There's an ongoing conversation about it here.

 

Is using a faster processor to push RF/EMI into a higher spectrum theory or fact, and if fact does it outweigh eliminating the noise as much as possible by using a slower processor?

[Johnseye]

On 4/2/2017 at 10:48 PM, romaz said:

My motherboard and various PCIe cards are now en route to SOtM to have their clocks replaced with the sCLK-EX.  A total of 4 clocks will be replaced including my motherboard's main system clock, the clock on an Intel dual-port gigabit LAN PCIe card, the clock on a PCIe SATA card, and finally the clock on a PCIe USB 2.0 card.  SOtM has told me they should be able to replace all the switching regulators on these PCIe add-on cards with their ultra low noise linear regulators and also add high quality capacitors.  I also sent SOtM my newly purchased HD Plex 300W-Hi Fi-DC to ATX converter as they have suggested that they can improve this converter by replacing the polymer condenser with a better one.  They will also be constructing my ATX cabling for me using their UP-OCC grade wiring.

 

I didn't have time to do any lengthy critical comparisons but before I shipped it all out, I put everything together to make sure everything worked.  I purchased an EVGA 850W Titanium ATX PSU to compare against the HD Plex 300W DC-ATX converter which was powered by my 19V HD Plex LPSU (my 19v Paul Hynes SR7 hasn't arrived yet).  The EVGA ATX PSU has ripple noise numbers between 3-5mV at 10% load which isn't spectacular when compared against the elite lower power LPSUs but it's better than the ripple numbers posted by the HDPLex DC-ATX converter. Since my CPU will consume 11w max and the motherboard will only be powering 4GB of RAM and the 3 PCIe add-on cards I listed above, at idle, this machine is consuming about 5w and with the CPU's turbocharging and hyperthreading disabled and the CPU running at 1.1GHz, during routine playback, power consumption was generally less than 15w and so I am well below 10% load for this EVGA ATX PSU.  Based on preliminary listening, both PSUs sounded very good.  It was difficult to say that one sounded better than the other.

 

While I didn't have time to try it, I expect to dethrottle both my CPU and RAM down to 800Mhz each for even less draw.  While this practice is counter-intuitive for those who oversample with HQPlayer, with my CAD CAT, this led to a nice bump in SQ and I am hoping for similar results provided that everything runs stable.

 

I did install Windows Server 2012R2 (but not AO) onto my 2TB Samsung EVO SSD, a Toshiba 1TB 5400 SATA II 2.5 inch hard drive and my recently purchased Intel X25-E SLC SATA II SSD. For some reason, I could not successfully install Windows Server 2012R2 onto my 32GB compact flash (even with the BIOS set to IDE mode) and so I wasn't able to experience how this sounded but I know of a few things I can try and so I haven't given up.  What I will say is that when connected to a standard SATA cable, my 2TB Samsung EVO SSD sounded the worst of the lot while the Intel X25-E sounded best with the Toshiba hard drive somewhere in the middle but closer to the Intel X25-E than the Samsung SSD.  With the Pachanko SATA Reference cable, the differences diminished considerably although the X25-E was still noticeably more to my preference (just a calmer presentation with a smoother edge).  During this testing, these drives were powered by an LPS-1.  

 

I also tested how the X25-E sounded when connected to the SYBA PCIe SATA adapter vs the motherboard's native SATA3 and SATA2 ports.  The difference wasn't as large as I imagined it would be and once again, the Pachanko SATA cable and LPS-1 may have something to do with it.  Also, since this is a SATA II SSD, regardless of whether it is placed on fast or slow bus, its throughput will always be slower than a SATA III drive.  Despite the smaller differences, I did find a preference for when the X25-E was attached to the SYBA PCIe SATA adapter. The good news is I heard none of the issues that @lmitche heard with his adapter from the standpoint of a thinner sound and female sibilance.  Tonality was quite rich and full bodied and there was a greater immediacy to the sound when compared to the motherboard's native SATA ports.  Whether this will lead to fatigue with long-term listening, I'm not sure but I'm hoping that things will only improve once the clock on this PCIe card has been upgraded and its switching regulators have been replaced.

 

One other area I got to test was 2GB vs 4GB of RAM.  While the difference wasn't huge, 2GB for sure sounds better and I will see if I can get 2GB to work out.

 

 

 

On 2/21/2017 at 3:34 PM, romaz said:

Here are some excerpts I've collected from Paul's e-mails to me over the past months. Hopefully, you'll find them useful as you make your own decisions. I've never personally met Paul (he is based in Scotland and I am in California) but based on numerous e-mail correspondences over the past year, I feel like I've gotten to know Paul pretty well and he has been a wonderful source of education for me. As you'll see from his correspondences, he is very articulate. He is also one of the nicest human beings I have met on my audio journey. I have spoken much of his SR7 on this thread and his power supplies really deserve a dedicated thread of their own but since this thread is about the things that "massively" improve either the microRendu or sMS-200 and since I consider his SR7 as the single most important component that I own that literally massively improves either of these NAAs, I feel it appropriate to to respond to your question comprehensively. Full disclosure -- I have no business relationship with Paul and I am a paying customer like anyone else.

 

Here is his response to me more than a year ago as I was comparing my HD Plex and its LT1083 regulators to Paul's own PR3 regulator design:

 

"The PR3 regulator topology used in the SR7 power supplies is my

proprietary discrete component design optimised for high quality audio

use. All the important audio related performance parameters of the PR3

regulator modules are considerably better than the LT1083

specifications.

 

The PR3 series voltage regulator noise measurements using an HP3561a

spectrum analyser specialized for low frequency measurement down to the

sub 1Hz region

 

The noise floor of 3561A is typically 15nV/sqrt Hz and the PR3 series

regulator was around 44nV/sqrt. When the 3561A noise floor is factored

into the measurement the PR3 series regulator would display lower noise

than 44nV/sqrt.

 

The ALW version of the Walt Jung ultra low noise regulator, which is the

lowest noise of the generally available DIY regulator circuits, was also

tested.

 

Here are the NF at 10Hz to 100 KHz:

PR3 series reg – 44nV/sqrt. Hz

ALW series reg – 70nV/sqrt. Hz

(40 nV/rtHz = 5 uVrms).

Measured from DC to 100 KHz the output impedance remains below 3

milliohms

 

Regulation operating bandwidth is from DC to >300 MHz allowing for device

tolerances.

 

Transient response for a 5A load current change is typically

<100nanoseconds and the settling time is also <100nanoseconds.

 

Transient current delivery for PR3HD modules >30A.

Supply line rejection >80dB DC to 100Khz (>150dB DC with double regulated PSU)

I use Panasonic FC, FM and FR low ESR high ripple current, high

temperature energy storage capacitors for excellent transient current

delivery and long working life. Alternatives can be fitted but it is

very important to ensure adequate power rating of power supply

capacitors in high current linear power supplies like the SR7. The

Panasonic capacitors are very well built and give consistently high

sound quality in capacitor tests on various audio forums."

Many have privately approached me with questions about how best to improve SQ and then stop me once I start talking about power supplies. Their response typically goes like this: "You don't have to convince me of the importance of good power, I already know this because I own an HDPlex." These folks have no idea. No offense to HDPlex but the model I own is nowhere in the same vicinity as the performance of my SR7. I also have a Teradak, Kenneth Lau, Paul Pang, iFi, and various DIY battery supplies that are now collecting dust. The only thing that comes close (that I own) is the LPS-1.

 

As I started to talk to Paul about the importance of the output impedance of a PSU, here was his informative response:

 

"Power supply output impedance is an important parameter, as are transient response, settling time, operating bandwidth and noise. For exceptional power supply design it is important to consider all of these parameters and optimise them to the best of your ability. This is what I do. The ideal power supply would have zero impedance at all frequencies of operation as you cannot develop any voltage into zero impedance no matter how much current passes through it. In the real world all circuits have some level of impedance and any current passing through this impedance will generate a voltage fluctuation, which in reality becomes an additional noise source in the system, which degrades signal integrity. It is therefore important to minimise the impedance to reduce this disturbance to the lowest level. This impedance reduction is usually achieved by an error amplifier using high levels of negative feedback, which introduces all manner of problems with operating bandwidth, transient response and settling time. This is a big subject to consider so I will not go into detail here. Suffice to say I do not use typical circuit topologies in my voltage regulator designs to achieve low impedance over a very wide operating bandwidth."

 

As you've seen above, Paul lists his measurements for these parameters. No one else seems to do this either because they don't have the measuring equipment to do so or they don't believe these parameters are even important. Without measurements, it's hard to even know how to shop for a good PSU and so the end user is forced to do his/her own listening comparisons or else make a purchase based on speculative comments.

Here is his response regarding the importance of using high quality, expensive discrete components. Paul's supplies are not inexpensive but it's not because it's all going into his pocket. Most importantly, IMO, the performance for the dollar is there:

 

"Some industry participants have been minimising the importance of “expensive” discrete component high performance voltage regulators in their marketing and concentrating their marketing stance on other parameters. I have to disagree here as I have conducted extensive testing in these areas over many years. Everything you use to build a power supply will have a sonic signature that will imprint on the achievable sound quality. Using low cost industrial voltage regulators with limited performance will just bottleneck the overall performance that is achievable and no amount of attention to detail in the other areas will override this."

Paul makes 3 lines of PSUs, from the SR3 to the SR7 and here is how he describes them:

 

"The SR3 was originally custom designed for a customer to improve on the performance offered by the Optima Red Top car battery used for the Altman Attraction DAC. As I foresaw additional sales for the SR3 for a wider variety of equipment I wanted it to be small enough and light enough to ship world wide via the Royal Mail small packet rate (under 2Kg) and of reasonable price. The mains transformer was the best of the “off the shelf” transformers available in the UK and the “off the shelf” chassis limited the transformer size to 60VA. The current PHD SR3 is rated at 30W continuous delivery and 240W transient delivery.

 

The SR5 was designed in response to requests for Mac Mini rated power supplies. The power rating had to be 80W continuous and it can also provide 350W transient delivery. This is achieved by increasing the mains transformer rating to 160VA, additional energy storage capacitance for robust power delivery with large dynamic load transients and the design of a custom chassis to house this and the electronics and heat sink required for this power upgrade. I also increased the power ratings of the Schottky Barrier rectification and the output stage device to provide the power rating with comfortable margins for safety and long life. More space within the chassis allowed me to design a mains transformer that would address the shortcomings of “off the shelf” mains transformers. In particular, core saturation is a big issue as the transformer operation stalls when this happens. Other big issues are electrical and mechanical noise. The SR5 and SR7 mains transformers are carefully wound on manually operated winding machines using over sized grain orientated silicon steel cores and high quality wire to minimise these issues. They are designed for low impedance operation and can deliver very large transient currents to the load.

 

The SR7 is essentially a higher power version of the SR5 with increased ratings for all power devices and a larger 250VA mains transformer giving a 125W continuous power rating and a 480W transient power rating for the more power hungry applications.

 

Many customers asked for a “Multirail” power supply to reduce space requirement for multiple power supplies in systems and also to reduce overall cost so I then designed the SR7 Multirail with up to four galvanically isolated supply rails to avoid interaction between the various items of equipment to be powered. The SR5 can also be configured as a Multirail but space limits the number of rails to two.

 

The best way to describe the performance between the three power supply levels is to use the car engine analogy. A small runabout with a 1.2 litre engine, a hot hatchback with a 1.6 litre fuel injected engine, a larger family saloon with a 3 litre fuel injected engine and a sports car with full engine management. Braking systems for these cars will be suitably rated to cope with the engine power. All four cars will get you from A to B but the ride will be progressively more responsive, dynamic and stable as you go up the range of cars. The runabout is the equivalent of a low cost power supply upgrade, the SR3 is the hot hatchback with acceptable all round performance and a lively drive, the SR5 is the 3 litre family saloon with a more stable but responsive drive and the SR7 is the sports car with even better performance.

 

As you go up the range in power supplies this translates into the traits you were hearing at the listening trials. The musical performance in all areas just improves with a more stable three dimensional presentation that is more robust when the going gets busy with the better power supplies."

More recently, he stated it to me this way:

 

"The SR range of high performance power supplies were designed for powering both analogue and digital audio and video equipment. The same proprietary ultra low noise high performance discrete component voltage regulator circuit topology is used in the SR3, SR5 and SR7 power supplies. The SR3 uses a standard 50VA mains transformer, the SR5 uses a custom manufactured 160VA mains transformer and the SR7 uses a custom manufactured 250VA mains transformer. The SR7MR uses a custom manufactured mains transformer with up to 500VA rating depending on the overall rail requirements. The custom manufactured mains transformers use oversized grain orientated silicon steel cores and are wound to avoid core saturation in use and to operate quietly both electrically and mechanically. Schottky Barrier rectifiers are used throughout the range, as they do not generate reverse recovery transients and their associated harmonic distortions.

 

The error amplifier used in the voltage regulator modules has the following specification :-

 

Noise voltage < 0.5 nanovolts root Hz

Operating Bandwidth > 300 Mhz

Supply line rejection > 80 dB DC to 100 KHz

Output impedance < 3 milliohms DC to 100 KHz

Transient response and settling time < 100 nanoseconds

 

As the SR3, SR5 and SR7 all use the same high performance regulator circuit topology they all have a similar sonic signature musically. Moving up the range allows better quality lower impedance mains transformers and up-rated rectification and regulator output stage providing a reduction in power supply output impedance, which in turn reduces interaction with the load. The increased energy storage capacitor bank also reduces rectifier ripple noise and RFI break through from the mains supply. Another benefit of the increased energy storage capacitance is with transient response and settling time. The net effect musically is to provide progressively larger, more stable and more robust soundstage particularly where large dynamic load current swings occur, as well as, a lower noise floor and improved timbre and temporal accuracy.

 

These power supplies are available with fixed voltage output from 1.6v to 30v or variable voltage output with a 10 volt span on voltage setting, within this range, using the precision adjustment potentiometer internally situated on the regulator module.

 

Continuous power output ratings – For fixed output voltage versions the SR3 provides 25W, the SR5 provides 80W and the SR7 provides 125W and the SR7MR chassis can support up to 250W spread across the rails. The SR5 can support one 6A module and the SR7 can support one 10A module. The output voltage and output current can be specified within this power rating using the formulae :-

 

V = W/I

I = W/V

 

Where V is output voltage, W is the available power in Watts and I is the output current in Amps.

 

If you require help with power supply specification I will be pleased to help you.

 

For variable output voltage versions of the power supplies, set at the maximum output voltage of the range, the SR3 provides 25W, the SR5 provides 80W and the SR7 provides 125W. Lower voltage settings than maximum will increase the voltage across the regulator output device, which will increase the heat generated in this device. To maintain safe operating temperatures and long term reliability the current rating should be de-rated by 8% per volt when reducing the output voltage level on variable output voltage versions.

 

The SR5 and SR7 power supplies are available in Multirail versions SR5MR and SR7MR with galvanic isolation between the supply rails to avoid ground return current intermodulation (ground loops) where more than one item of equipment is powered from the same power supply. The SR5MR can support one 6A regulator module and one 3A regulator module or two 3A modules. The SR7MR can support one 10A module and up to three additional 3A modules or two 6A modules and up to two additional 3A modules.

 

XL ultra low impedance (< 1 milliohm) connectors and fine silver internal wiring between capacitor banks, regulator modules and the output connectors, can be fitted to the SR5 and SR7 power supplies.

 

The SR3, SR5 and SR7 single rail supplies are also available in DR versions where two of the high performance voltage regulators are cascaded to a give supply line and rectification interference rejection exceeding 150 dB from DC to 100 KHz. This provides lower overall noise levels than the standard power supplies.

 

Current Prices 200117

 

SR3 £300

SR3DR £500

 

SR5 £600

SR5XL £678

SR5DR £800

SR5DRXL £885

SR5MR2 £800

SR5MR2XL £970

 

SR7 £750

SR7XL £870

SR7DR £950

SR7DRXL £1080

SR7MR2 £950

SR7MR2XL £1190

SR7MR3 £1150

Here is his pricing for his DC leads:

 

Current Prices 200107

 

All prices are for1 metre lengths with a Switchcraft DC plug. Alternative lengths and connectors can be quoted for if required. Cable impedance reduces with higher current rating. XL ultra low impedance (< 1 milliohm) connectors can be fitted to the SR5 and SR7 DC Leads.

 

3A current rating :-

DC3C annealed copper with Teflon insulation £050

DC3FS annealed fine silver with Teflon insulation £085

 

6A current rating :-

DC6C annealed copper with Teflon insulation £075

DC6FS annealed fine silver with Teflon insulation £145

DC6FSXL annealed fine silver with Teflon insulation £186

 

10A current rating :-

DC10C annealed copper with Teflon insulation £100

DC10FS annealed fine silver with Teflon insulation £229

DC10FSXL annealed fine silver with Teflon insulation £270

There is also the option for Paul to make you a double regulated PSU (with cascading regulators) and this can be discussed with Paul for those who want his very best. I cannot yet tell you what this sounds like because I haven't received mine yet. Considering the performance I am getting from his SR7, it is my opinion that this is some of the best money I have ever spent on audio. Without question, quality is the foundation of anything that is good in audio. Paul does not have a presence here on CA and for those who wish to correspond with him, you can do so via e-mail at [email protected].

 

On 3/18/2017 at 3:23 AM, romaz said:

So it would seem that moving up to this larger microATX ASRock motherboard will bring with it both good things and bad but hopefully, the positives will outweigh the negatives.

 

The Good:

Two PCIE 3.0 slots that allow the most direct access to the CPU that is possible. These slots have the same low latency access to the CPU as RAM. All the other buses must go through the Platform Controller Hub to access the CPU. One slot will be used for my Intel dual LAN card and the other slot will be used for my SATA to PCIE card where my OS drive and one of my data drives will connect.

 

The Bad:

More noise.

 

I can power the CPU directly with a 12V lead from my SR7. This should be a significant advantage over my Mac Mini and hopefully will help negate some potentially serious disadvantages.

 

My mATX motherboard will require a high current 5V, 3.3V and 5V standby rail in addition to possibly another legacy rail and so Paul Hynes has told me that I will not be able to use my SR7 to directly power these rails without a DC-ATX converter. This means I have placed an order for the HDPlex 300W-HiFi-DC-ATX converter based on the recommendation of many. From my search, I have not found another DC-ATX converter that is convincingly better. With this HDPlex 300W-HiFi-DC-ATX converter in place, I can then use my SR7 to feed it. I have had another SR7 on order since January and Paul has told me it should be finally ready to ship in the next week or two. Upon hearing my dilemma, Paul was kind enough to agree to convert my 12V rail to a 19V rail so that I can feed the HDPlex DC-ATX converter the voltage it requires. He has also agreed to fabricate for me a high performance 24-pin to 24-pin ATX cable so that I can connect the HDPlex DC-ATX converter to my motherboard. I really can't say enough great things about Paul. For those with custom LPSU needs, I can't recommend him more highly.

 

Anyway, now that I'm on this path, I did some digging around and here are some comparative values. They don't guarantee how good something will sound but they do provide perspective.

 

Since only noise floor data is provided by most manufacturers, I've gone ahead and listed what I have found. Noise is listed in mV and the lower the value, the better:

 

Generic switching PSU ~ 60mV

EVGA SuperNova 1.6kW Platinum ATX PSU (100% load) ~ 12.4 - 22.7mV

EVGA SuperNova 1.6kW Platinum ATX PSU (20% load) ~ 4.0 - 13.4mV

Corsair AX1200i 1.2kW Platinum ATX PSU (100% load) ~ 7.3 - 13.7mV

Corsair AX1200i 1.2kW Platinum ATX PSU (20% load) ~ 5.6 - 11.4mV

HDPlex 400W ATX LPSU (100% load) ~ 3-5mV

HDPlex 100W LPSU with LT1083 regulators (100% load) ~ 2-3mV

Paul Hynes SR7 (100% load from 10Hz to 100KHz) <4uV = <0.004mV

 

Compared against either HDPlex LPSUs, the SR7 is in the order of being nearly 1000x quieter and nearly 10,000x quieter than a generic switching PSU.

 

Unfortunately, because I will have to use the HDPlex 300W-HiFi-DC-ATX converter which uses regulators with measured ripple values of 10mV (more than 2000x noisier than the SR7), the ultra low noise floor of the SR7 will be buried by the much higher noise floor of the HDPlex DC-ATX converter. From a noise standpoint, almost any decent switching PSU may possibly sound just as good because this converter will be the limiting factor.

 

This then raises the question of just how noisy the numerous switching regulators buried within the motherboard are. Quite possibly, they may be even noisier than 10mV and so from a noise standpoint, especially with a motherboard's 3.3V rail (which is the noisiest rail on an ATX motherboard), all of this may be moot. Quite depressing.

 

CONCLUSIONS?

1. We really need someone to design an audiophile-class motherboard without noisy switching regulators.

2. In the absence of such a motherboard, because noise is potentially additive, it probably still makes sense to use a low noise PSU but its benefits will be at least partially negated by the noise created by the motherboard (or the DC-ATX converter).

3. It would seem that ATX power supplies designed for high power output (specifically high current output) will generally also have a higher noise floor but in this regard, the HDPlex ATX LPSU stands out as being exceptional.

4. Noise aside, low impedance still matters. To my ears, this may be a more important quality because with the aid of grounding boxes, noise filters and galvanic isolating devices, there are ways to mitigate noise but no way to undo the suffocating impact of a high impedance PSU. It's quite possible the EVGA excels here but there's no way to know except through comparative listening since no measurements are available.

 

On 8/13/2017 at 5:12 PM, romaz said:

 

Very nice post.

 

As has been the case in the past few months, it seems it is only at airports when I can find the time to catch up with forum activity and that is the case once again as I wait for my flight to Fiji to take off.

 

I, too, have an sPS-500 along with SOtM's copper single DC lead (not a Y-cable).  To my ears, this is a very good PSU, easily better than my HDPlex and so I would agree with several posts that a linear PSU is not always better than a switching PSU.  Having spoken with Lee about this PSU back in Munich, this switching PSU has a measured ripple noise of about 100mV, which is horrendous!  How does a PSU with such high ripple noise sound so good?  I have already stated this in previous posts but if you haven't figured it out for yourselves by now, Lee's (SOtM's) products are all about filtration.  I liken Lee to a PhotoShop artist.  Those of you who are photographers know that it's generally always better to get it done right in the field than to have to fix it in post but post-production tools like PhotoShop have gotten so good that even when the shooting environment is less than ideal, PhotoShop can lead to excellent results.

 

Case in point, I have found a reputable motherboard manufacturer who is willing to develop and build an Intel X86-based audiophile-grade motherboard for me (for a development fee that would probably start at around $1,500).  I would envision this motherboard would contain no noisy internal switching regulators and having discussed this idea with Lee, while he had some interest in the idea, he also felt it wasn't that important because he believed he had already largely figured out how to filter out the noise that comes from many of today's noisy motherboards.  His filtering methods are on display with his current Ultra endpoints and especially with his Ultra PCIe USB cards and so I will leave it to those who own such devices to decide how successful Lee's filtering methods are.  In my own situation, I have stopped pursuing the development of my own motherboard.  

 

This filtration approach is also the basis for his sPS-500 PSU and his upcoming mT-1000 AC power strip.  While SOtM makes their own linear and battery-based PSUs, reading between the lines with the comments they have made to me, they believe their switching sPS-500 is their best sounding PSU.  Are his filtering algorithms impervious to the very worst AC conditions?  Not surprisingly, his answer is "no."  Just like with PhotoShop, the better the starting material, the better the final product.  According to Lee, the sPS-500 benefits from good starting power and especially a good AC power chord.  

 

What do I think of the sPS-500?  I have to be careful with what I say here because my sPS-500 doesn't have a lot of hours on it yet (I've been gone a lot lately) and so I suspect it will continue to improve.  I have read Steven Plaskin's review and I respect his findings but I don't agree with them.  To my ears, against my HDPlex, it is A LOT better and so we are both in agreement there.  Against my LPS-1, I find the LPS-1 easily superior with respect to dynamic contrasts, both macro but especially microdynamics.  Against my SR7, the gap is even larger in all the ways that you have found.  As good as its filtration schemes are, I think its limiting factor is its output impedance, which according to Lee, measures about 50 milliohms.  This is more than 16x the output impedance of the SR7.

 

With regards to the impact of DC cables, I am of the opinion that they all cause harm and just like with clocks, the very best they can do is to cause no harm.  With the very best low impedance PSUs, the best DC leads can make a noticeable impact from the standpoint of keeping line resistance (and, therefore, impedance) as low as possible.  Unlike with AC leads, the dielectric used (which can be the basis for why some AC cables cost so much) doesn't really matter since DC cables don't have to contend with issues such as HF smearing or roll-off.  Metallurgy matters but only from the standpoint of line resistance.   Silver is a better conductor than copper (and I generally use high purity silver every chance I get) but silver is only a 5% better conductor than copper.  Large gauge copper can be more effective and a better value.  I have 12 inch 18 AWG UP-OCC silver DC lead that is noticeably improved upon by Sonore's 12 inch DC4 cable that uses 15 AWG copper.   If you're using 14 AWG silver, your cable for sure is making a difference.

 

With regards to star quad (vs twisted pair), with short cables, to my ears, I can't hear a difference, at least not with DC cable.  I would put my money on the largest gauge and shortest cable you can get away with.

 

I think the bigger problem (and perhaps the limiting factor) is not the DC cable but the barrel connectors used in most of these small devices.  I use Oyaide barrel connectors exclusively as I don't know of a better brand but barrel connectors are not low impedance connectors and serve as a bottleneck to the very best PSUs.  Here is what Paul Hynes has to say:

 

"I usually fit Switchcraft barrel connectors, as they are a good solid 
connector that fits well and the metal surfaces are well plated unlike 
the lower cost connectors often used. However, I can fit Oyaide Barrel 
connectors on request... The Oyaide connector 
offers a small but noticeable improvement over the Switchcraft.

To be frank, I am not a big fan of barrel connectors as they are not low 
impedance connectors. Usually the mating PCB connectors fitted to the 
equipment PCBs are poor quality connectors and this would be the 
limiting factor even with Oyaide barrel connectors. In my product range 
I use precision gold plated DIN for low cost applications (approx 5 
milliohm), silver plated XLR for my standard power supplies (approx 3 
milliohm) and military spec silver plated XL (not XLR) connectors on all 
high performance applications (leas than 1 milliohm). These offer 
progressively better performance than barrel connectors with the XL 
connectors being exceptional and the best I have tried. Unfortunately, 
with many of today’s audio products there is just not the room to fit 
these connectors so I usually end up migrating circuit boards to larger 
enclosures to allow their use with high end applications."

 

Each of my SR7s have Paul's XL connectors on the PSU end but Oyaide barrel connectors on the component end.  With the tX-USBultra, which will be the endpoint that connects directly to my DAC,  I am contemplating hard-wiring the DC lead directly to the PCB.

 

29 minutes ago, sandyk said:

 

There are ATX PSUs available these days that have very low SMPS ripple. Many also have speed controlled fans  .

Well implemented Coax SPDIF from a decent soundcard into a DAC with a well implemented SPDIF input (e.g. Input transformer for isolation)  has the potential to crap all over typical USB Audio, where you often need to spend more on other ancillaries to get the best from it, than the computer itself cost !

Coax SPDIF is capable of far wider bandwidth than currently used, as is Glass Optical Fibre.

 In my case, I also use the attached DIY very low noise PCB for vastly improved power derived from the internal +12V SMPS, which is regulated down to 2 separate isolated +5V supplies for my internal SSDs .The electrical noise output from each half is around 4uV. Click on the image for a larger image.

 

 

@sandyk Instead of rehashing everything that's been discussed in this thread to date about PSUs I've linked a few key posts which will provide insight into why we're so bullish about the Hynes SR7 and a low power approach.  Since then the sMS-500 has been released with a large 100mv ripple, but superior filtering.  There's been limited comparison between the two, but you can guess the SR7 came out ahead.

 

These posts also go into the logic behind it all.  Considering there was no response to my question about using a faster processor to push RF/EMI into a higher spectrum I'll assume this was theory on your part but I'd like to learn more about this if you have some factual background.

 

Here's a noise floor comparison for you.  The SR7 LPSU has the same noise output as the DIY you use for your SSDs.

 

Generic switching PSU ~ 60mV

EVGA SuperNova 1.6kW Platinum ATX PSU (100% load) ~ 12.4 - 22.7mV

EVGA SuperNova 1.6kW Platinum ATX PSU (20% load) ~ 4.0 - 13.4mV

Corsair AX1200i 1.2kW Platinum ATX PSU (100% load) ~ 7.3 - 13.7mV

Corsair AX1200i 1.2kW Platinum ATX PSU (20% load) ~ 5.6 - 11.4mV

HDPlex 400W ATX LPSU (100% load) ~ 3-5mV

HDPlex 100W LPSU with LT1083 regulators (100% load) ~ 2-3mV

Paul Hynes SR7 (100% load from 10Hz to 100KHz) <4uV = <0.004mV

 

[Johnseye]

29 minutes ago, lmitche said:

LOL, Johnseye, you remind me of the dude in this ad.

 

https://www.ispot.tv/ad/wk55/homeadvisor-mailboxes

 

 

Not sure how to take your last two posts towards me Larry but at the moment I'm insulted. Why don't you take it to a PM and while you're at it come by and repair my roof. 

[Johnseye]

6 minutes ago, lmitche said:

Great to know you have a sense of humour Johnseye!

 

Glad you caught that. I wasnt sure you would. 

 

Look Larry, there's nothing wrong with your ATX PSU. I'm sure it performs very well and does what you want it to.  I'm not knocking it and it obviously suits your need. I just went down a different path. Not a better path, but a different one. I see value in your approach, hopefully you see why I'm trying this one. 

[Johnseye]

14 minutes ago, austinpop said:

 

You and @Johnseye seem to have gotten pretty evangelical about the single server approach.

 

I would caution people not to think this is the One True Path. There are valid reasons for going with either approach. 

 

Until either of you have done a direct comparison with a well designed trifecta, I would caution against making these blanket assertions. At this point, it’s just your opinion.

 

Roy’s findings are in the context of his system which has extreme optimizations most people can’t afford.

 

My point is - I want this thread to be focused on observations, not speculation.

 

24 minutes ago, Johnseye said:

Look Larry, there's nothing wrong with your ATX PSU. I'm sure it performs very well and does what you want it to.  I'm not knocking it and it obviously suits your need. I just went down a different path. Not a better path, but a different one. I see value in your approach, hopefully you see why I'm trying this one. 

 

Agreed Rajiv.   That's why I said what I did in my last post.  While it may appear I'm pushing the approach, I'm really defending it.  The last thing I want is to come off the wrong way, so apologies if that's how it sounds.

[Johnseye]

7 minutes ago, austinpop said:

 

No worries then!

 

Change the topic to politics or religion? 

[Johnseye]

2 hours ago, TubeMan said:

" It may also use a high quality soundcard with Coax SPDIF output "

any suggestion of that?
I have looking for one at long time but not find any good PCIe card

 

+1

 

I started down that path long ago and not only found the options limited, but not as good.

[Johnseye]

4 hours ago, ElviaCaprice said:

I couldn't disagree more.  There is nothing extreme about Roy's optimizations in his sCLK-EX server approach that others could not afford.  And until someone else can disqualify his findings/comparison, I would say it stands. 

 

I'm definitely not going to waste time and money making the comparison myself, as Roy didn't think it worth, or far enough apart for those with the trifecta to make a change. 

 

But, if your new to the sCLK-EX, I would think it wise to weigh the options from a price/software/format wise comparison.  Which in my eyes, favors the sCLK-EX server option.

 

 

To be fair, the current discussion in this thread specifically began with afrancois and was a defense of sCLK-EX modified server and tx-USBexp.  No one ever said flat out that this approach was clearly better than the Trifecta.  To my knowledge no one but Roy has compared the two in an A-B comparison, let alone DBT or anything with a strong foundation.  Elvia stated "Roy didn't think it worth the effort for someone with a well designed trifecta to switch over to a sCLK-EX server.  It's just that the software advantages for his needs overcame the trifecta limitations."  I also stated I don't have an sMS-200 ultra so could not draw that comparison.

 

The discussion had a side component of a large cased PC with ATX PSU and a sound card with sandyk's posts, not to mention lmitche's input.  There is a lot to debate there if one wants to.  I still don't see the point of going down that path but to each their own.  By all means, throw a lot of power and speed into your system and add some fans.  To me that equals more noise/distortion which equals a higher noise floor.  Nothing is wrong in this hobby, there are just different strokes.  Some like analog, some like digital, some like DSD, some like PCM.  I will never say someone's preference is better than another.

 

I also think it's important to point out that the difference between the Trifecta and my approach if I use the tx-USBultra, is that my server will be modified instead of the switch and I'll be using a tx-USBexp instead of an sMS-200 ultra.  It's still a trifecta, but one that doesn't pass over ethernet between server and endpoint.  This is a different approach, not a better one.  It's one where I can easily pull out the tx-USBultra to see how much of an impact it has if any.  I'd prefer not to have the tx-USBultra but since I use the ISO Regen I really want to know how it compares.

 

If anyone lives in the Chicago area, you're welcome over for a listen.  If you have a Trifecta and want to bring or ship it to me I'd be happy to compare the myriad of options available.

[Johnseye]

13 hours ago, austinpop said:

BTW @Johnseye thanks for the info about the GS105.

 

I just got my GS105v5 delivered today. Swapped it out for my grounded FS105v3, and SQ is just as good, 

 

Thanks for lending your unit to JS!

 

Because my NAS is connected to this switch my nightly backups and any audio data I copy goes over it.  100MB is painfully slow by comparison, even though it doesn't matter when streaming audio.  The server can still be set at 10 or 100MB if desired.

 

2 hours ago, austinpop said:

 

I know this is still months away, but if I do make it out to Axpona next April, I can try to bring my trifecta to Chicago, if i can figure out how to pack it all securely in my luggage!

 

I hope to see you then.  It will be a fun weekend, especially if you're able to bring the trifecta.

[Johnseye]

1 minute ago, elan120 said:

It would be great to see approach to higher power sCLK-EX based server that is capable running HQPlayer that can up-sample to DSD512, and this is what I am researching currently to see what server components I need to put one together and perhaps send to SOtM for sCLK-EX upgrade.

 

In order for HQPlayer to upsample to DSD512 you need a high speed processor.  You do not need a high amount of power.  Your PSU only needs to be sufficient in providing enough power to the components of your server/pc.  You don't need an nVidia Cuda GPU to upsample DSD512.  The latest Intel and AMD CPUs have 16-18 cores but they run up to 165w.  This isn't necessary, even if you're into processors which I have to admit is sexy stuff.

 

Find out what processor you need to effectively upsample to DSD512.  Design everything around that.  I'm guessing you're looking at a 95w proc.  What we've been kicking around is that there are very few motherboards capable of external DC in that support this proc.  That leaves you with needing either an ATX PSU adapter like that from HDPlex or an actual ATX PSU.  This is what's led to the discussion in approach to power this kind of system.

[Johnseye]

3 minutes ago, elan120 said:

I currently lean more towards 65w processor instead of 95w without Cuda Offload, which will need a power supply no more than 200w, perhaps around 160w even.  Finding a capable low noise power supply is one of the key search now in addition to finding a suitable Mobo that will have txUSBexp taking one of the 4 output points sCLK-EX.  I still have a lot of research to do, that is the main reason I hope to see if there are more inputs to help put this server together.

 

My SR7 will handle that.  It will provide 190w at 19v.  The ASUS board I chose can handle the 95w series of i7's although I first went with the i7 35w proc and was planning on seeing how that performed first.  That board doesn't have a PCI slot so when I decided to go the tx-USBexp route I needed to find a new board.  I then tabled my DSD upsampling plans, but will one day revisit.  Let me know what you end up with.