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

[austinpop]

5 hours ago, AmusedToD said:

I just didn't get it, can the switch work without the master clock in the Tx board or not?

 

Again, needs to be a FAQ, the number of times this gets asked.

 

No it cannot. The switch is modified to get its clocking from the sCLK-EX board, in whichever Ultra chassis you're using in the chain - the sMS-200ultra, the tX-USBultra, or the dX-USB HD Ultra.

 

The switch no longer works standalone.

[austinpop]

4 hours ago, AmusedToD said:

 

This is what May wrote to me:

 

- Enabling sCLK-EX clock output : USD50/an output

- Installation cost by SMB connector: US$70/ a tap

- Cost for switch : Depends on which model you would buy.

- Switch modification :

1. adding or replacing caps on the switch normally requires at least US$20

2. switching power regulators normally requires US$50/ a point

 

Yup exactly. In my Zyxel switch, they modded:

• enabled sCLK-EX input $50
• replaced 2 regulators for linear $100
• replaced 1 capacitor $20

[austinpop]

7 minutes ago, lmitche said:

The endpoint based approach.  That's why I wanted to know what your backend looks like.  If you don't want to share the details that's OK.

 

Ah OK. Thanks for clarifying. I'm happy to share, and will take any and all suggestions under advisement!

 

So, to answer your questions. 

• sMS-200 runs as Roon endpoint.
• No upsampling in Roon. I've tried Roon and HQPlayer upsampling extensively - posted about this in other threads - and for my DAC, native sample rates sound best to these ears. So for me, it's native all the way.
• Music is on NAS, mounted on server as SMB share. I have experimented with music on a local drive, both on the system drive SSD, and on an internal SATA 1TB HDD. Made no difference, and I really prefer to keep my content on NAS, from which I have - some would say overly baroque - triple redundancy backup redundancy schemes!
• W10 server: 12GB RAM, 256GB Samsung EVO SSD system disk, 1TB internal SATA HDD. By MB, assuming you mean motherboard? Nothing special - just a stock Dell XPS 8700. Processor is i7-4770 I believe. Since I do no upsampling, it barely breaks a sweat.
• Using Audiophile Optimizer (AO) 2.2beta5, as well as Process Lasso, in Bitsum Perfect mode.
• Bridging network adapters are internal PCIe connected to router, USB 3.0 adapter connected to sMS. Both employ the Realtek chipset.

[austinpop]

33 minutes ago, AmusedToD said:

@austinpop, would you share the details regarding your DAC, amp and speakers? You don't have to, of course. But it would be good to know how resolving your system is. Just to try and make a reference as to whether lower end systems can benefit from all those reclocking points upstream. I believe @romaz has a Chord Dave which he uses to drive his highly sensitive speakers directly, therefore I guess any change upstream can be clearly heard and differentiated through his system (even though Mr. Rob Watts thinks his DAC is immune to USB sources)  

 

All in my sig!

 

see below. Not visible on mobile browsers.

[austinpop]

1 hour ago, lmitche said:

Hi Rajiv,

 

Got it, 3 computers make the chain

 

I guess so, depending how you're counting! If you mean NAS, Dell desktop, and the sMS-200, then yes.

 

1 hour ago, lmitche said:

Is the Dell machine a single purpose, music playing machine?  If it has an NVME slot, an Optane stick may be able to replace the SSD and you could pull the HDD entirely.  You also only need two 4 GBs sticks of ram in the two slots closest to the CPU.  Also consider a second USB 3 NIC for the bridge and turn off the internal ethernet port as well as SATA controller. There will also be a bunch of PCIE bus lanes you can turn off in Windows devmgmt. You could also upgrade the power supply to one of the new titanium ultra low ripple ATX SMPSes.

 

OK, I'll look into these.

 

Quote

I would think a second SOTM switch between the server and NAS would also add benefits.  

 

You do realize there is no such thing as an SOtM switch, right? The reclocking switch I'm using was modded to use a free sCLK-EX tap from my tX-USBultra. Now that I've used all 4 taps. I would need to buy another Ultra component and then use a free tap on a modded switch.

 

This is where I think I'm just going to wait for the dust to settle with Roy's experiments. I am happy to piggyback on his results!

 

1 hour ago, lmitche said:

On my QNAP NAS there was a noisy DOM usb chip used for the QNAP OS.  In the end, I pulled the DOM and since it was an atom machine configured it to boot Linux and run SMB that way.  The NAS only used HDDs. Lastly I had the two NAS nics bridged here as well.  In my case I had three EMOsystems HD EITs one between each machine and the router along with BJC Cat 6A. And the NAS was powered by a very high quality Sigma 11 based LPS.

 

Actually, my NAS is a Synology 916+ 4-bay. As you may recall, I beat my head on bridging the 2 NICs before giving up.  

 

1 hour ago, lmitche said:

I hope this post gives you some ideas for improvement.

 

Thanks for the suggestions!

[austinpop]

1 hour ago, AmusedToD said:

Great system indeed! I am not much of a headphone guy, although I tried.

 

I would love to have a standalone listening room with speakers, but I live with a house full of women (wife and daughters) who get bothered when I play music at "realistic levels." I do have a home theater with speakers, but find I'm always turning the volume way down, except when we all watch movies together.

 

I switched to headphones when my girls were little, and have not looked back since. The best modern headphones (and I count the Sennheiser HD800S in that group, even though there are many pricier 'phones) still cannot give you the sense of physical volume and space that a speaker based setup can, but in terms of resolution and coherence, they rival and IMO, exceed most speakers except perhaps the very rarefied high end. This is just my opinion, so please let's not go down the headphone/speakers rat hole. There are some fine threads here on CA where you're welcome to join the discussion. These invariably end up on my ignore list once the discussion devolves.

 

1 hour ago, AmusedToD said:

I was thinking of getting the Ayre Codex myself, but went for an R2R DAC instead.

 

So many good DACs out there. My library is about 25% DSD, so that played a factor in my choice. I cannot praise the Codex highly enough.

 

1 hour ago, AmusedToD said:

Anyways, regarding the main topic, I decided to put the upgrades on hold for now and wait for things to clear up a bit. I am especially curious as to whether modding the clock at the very source (Intel NUC for example) would bring even more improvements than the switch (if the NUC is connected directly to the sms200Ultra, bypassing the switch).

 

Yes, this is where Roy's results with give us much to chew on.

[austinpop]

3 minutes ago, Superdad said:

 

Uh oh.  Larry wants to look at Rajiv's back end.  What sort of site is this becoming?

(But I'll show you mine if you show me yours...)  

 

Thanks for your ass-inine post, Alex. 

 

It's these kind of lofty standards that distinguish us from the other sites.

[austinpop]

28 minutes ago, Superdad said:

 

OT: Rajiv, I see you own the Focal Elear headphones.  I have not heard them yet (they are likely not on display within 300 miles of me), but I keep reading nice things about them.  Am beginning to hunt for some "reference" cans for some upcoming product development work.  Would you recommend them--confidently enough for me to purchase "blind" (or would that be deaf?)?

 

Hi Alex, yes this is OT, and we should probably take it to PM if needed. It really depends on:

• your price range
• your musical preferences
• open vs. closed 

Given the Elear's can be had for around $800, it's a real standout in value. I wrote a review on them in head-fi: 

https://www.head-fi.org/f/threads/focal-elear-in-stock-and-a-loaner-program.821542/page-4#post_13056744

 

Take a look.

 

28 minutes ago, Superdad said:

Others to consider? 

The only pair of cans that made me go WOW--when I was at the 2015 RMAF--were the Zenith Audio modified PM2 (neither of the stock Oppo cans did anything for me).

 

In the sub-$1k range, not much. Some people swear by the Audioquest Night Hawk and Night Owl, but I haven't heard them. Once you get in the sub-$2k range, the Mr Speakers EtherFlow and EtherFlow C are very good. A lot of people love the Audeze line - I have uniformly found they do not suit me. They're too heavy on my head, and tonally too dark for my taste.

 

Sadly, I don't recommend buying any headphones blind/deaf, unless you have a clear return option - with no restocking fee.

 

You could try to see if Todd (the Vinyl Junkie) is still running his loaner program - the thread my review is in above. Other than that, your best bet is to go to a CanJam or head over to one of the stores in the "big city." I know The Source AV in LA (Torrance) carries a lot of 'phones on display.

 

Oh and final idea - The Cable Co's headphone lending library: https://www.thecableco.com/Product/NEW-Headphone-Lending-Library

[austinpop]

@Superdad check your PM.

[austinpop]

8 hours ago, tapatrick said:

The point I'm making and I'm sure Rajiv will concur the MBP as source is not a good comparison with his trifecta. 

 

1 hour ago, mozes said:

Yes the iphone in airplane mode was clearly better than the MBP.

 

OK guys - got it. MBP bad. ✓

 

Now let's wait for someone who has a legitimately optimized direct USB setup compare against an Ultra trifecta chain.

 

Clearly won't be me! 

[austinpop]

21 minutes ago, Always.Learning said:

Yesterday I received the final part of my "Ultra trifecta" package from SOtM. I had hoped to be able to run both the SOtM-modded switch (a D-Link supplied by Crux Audio) and the sMS-200 Ultra from a single LPS-1 using a Y-cable supplied by Ghent Audio. I had SOtM modify the switch so that it accepts a 7 volt power supply so that I could (theoretically) run the LPS-1 at 7 volts for both switch and sMS-200 Ultra. I can now confirm that this combination is just too great a load for the LPS-1 -- the power supply would alternately power on and off when I hooked up both components with the Y-cable. Just powering the sMS-200 Ultra, the LPS-1 gets quite toasty. 

 

I know Rajiv has successfully used an LPS-1 and Y cable to power his modded sMS-200 and modded Zyxel switch, but he was very careful to point out that the sMS-200 Ultra might require more power. He was right. 

 

Ugh, bummer! But worth a shot.

 

21 minutes ago, Always.Learning said:

I now have another question for Rajiv and anyone else who wants to chime in. I will be purchasing another LPS-1 so that I can power my modified Ultra switch at 7 volts. I'm also thinking of purchasing the ISO REGEN and using my Y-cable to power both the Ultra switch and the ISO REGEN at 7 volts. IIRC, Rajiv has successfully run both his modified Zyxel switch and ISO REGEN using a Y-cable and LPS-1. Rajiv, could you confirm that this is true? Does anyone else have relevant opinions or feedback?

 

I've not actually tried that - running the switch and the ISO-R off of the same LPS-1. I'm fairly confident that current capacity-wise, it should work. The only caveat is the following. Consider the chain:

• modded switch > sMS-200ultra > ISO-Regen > tX-USBultra > DAC
• __________________________________| |_________________________
• __________________________Isolation Moat_____________________

Now remember - the ISO-Regen is bus powered upstream of the moat. It's own PSU - the LPS-1, say - powers the circuitry downstream of the isolation moat. If you use a single LPS-1 to power the switch and the ISO-R, that LPS-1 is spanning across the moat, and theoretically, defeating the isolation.

 

How much does this matter, practically and sonically? Well, only your ears will tell. If you want to be absolutely pristine, obviously then you need separate LPS-1s. But that's an expensive option. The better approach may be to start with a single LPS-1 and experiment. Compare:

• LPS-1 powering modded switch AND ISO-Regen via Y-cable to:
• LPS-1 only powering ISO-R, and use an alternate supply - LPS or battery - to power switch.

 

21 minutes ago, Always.Learning said:

Although I have only had one short listening session with the tX-USB Ultra and sMS-200 Ultra in my system and the components are (obviously) still burning in, I easily heard very significant improvements from these two components. Some of those improvements may be due to the new SOtM dCBL-Cat7 ethernet cable I am running between source and sMS-200 Ultra. FWIW, my source is a Mac Mini with the MMK mod by Uptone, powered by Uptone's JS-2, running Roon. I'm looking forward to inserting the Ultra-fied switch and possibly ISO REGEN in the future and will report back. 

 

Awesome! Can't wait to hear your impressions.

[austinpop]

2 minutes ago, Johnseye said:

 

If you want to send me your ultra trifecta ...

 

Sorry - no can do. Due to the hardwired clock cables, I'm paranoid about how much physical stress it can tolerate in shipping.

 

In fact, I am getting close to biting the bullet and just sending it back to SOtM to be re-done using the SMB terminations, so I can use detachable Digikey clock cables. At that time I'll probably just spring for them to add a master clock connector for future proofing.

[austinpop]

8 minutes ago, Always.Learning said:

 

Rajiv -- thank you for the information and advice. A question for you:  Your sig says you are running 3 LPS-1s and yet, unless I'm mistaken, you are powering four components with these three power supplies:  the Ultra trifecta plus the ISO REGEN, correct? So doesn't that mean that two of those components are sharing an LPS-1, or are you using a different supply for one of the components (e.g., the switch)? How do you avoid the "moat" issue you describe above?

 

Ah - the key difference is I don't have an sMS-200ultra. I have the regular sMS-200, modded. I believe it is the sCLK-EX board (in the Ultra boxes) that requires significant current. I only have one Ultra box, the tX-USBultra. You have 2 - the sMS-200ultra and the tX-USBultra.

 

I run the switch and the modded sMS-200 off of one LPS-1 no problem, and since both are upstream of the moat, I don't span the boundary. And as I've noted with my tests with the GI switch on the ISO-R, I do hear a (small) benefit from the isolation.

 

Here's the detailed LPS-1 usage:

1. LPS-1 #1: modded switch and modded sMS-200, using Ghent Y-cable
2. LPS-1 #2: ISO-Regen
3. LPS-1 #3: tX-USBultra

In order of toastiness, my LPS-1 #3 is the toastiest, followed by LPS-1 #1, followed by LPS-1 #2.

 

Hope this helps.

[austinpop]

6 minutes ago, Johnseye said:

You think we go to extremes.  Look what this guy did to get good power.

 

 

 

 

 

Wait - that 2nd guy who has to move his couch to open the fridge has a girlfriend?! Somehow I doubt it. 

[austinpop]

1 minute ago, Always.Learning said:

Does anyone have a recommendation for a cheap linear power supply or battery supply that could supply 7 volts to power my SOtM-modified switch -- I'd like to conduct the experiment that Rajiv has suggested a few posts above. My first thought was a Jameco LPS but they don't seem to make a 7 volt version. Any and all suggestions appreciated!

 

I'd strongly suggest you try the LPS-1 sharing across the moat anyway. It's quite possible the practical loss of SQ is small. After all, there are many people using the mR + ISO-R with one LPS-1, and this crosses the moat too.

 

Also, note that even if your modded switch accepts 7V, it may well work at 5V too. Mine works fine at both 5V and 7V. Jameco makes a decent 5V PSU.

 

Let me rummage through my drawers, I may be able to loan you something. You US-based? I'll PM you if I have something.

[austinpop]

9 minutes ago, austinpop said:

Let me rummage through my drawers, I may be able to loan you something. You US-based? I'll PM you if I have something.

 

I have a tiny Teradak U9VA I could send you. It has a 5v output via USB A, but I have a cable for that too.

 

6 minutes ago, kennyb123 said:

 

You are welcome to borrow my "el cheapo".

 

This may be even better. El cheapo has an internal screw that can dial down the voltage to 7v if you want to go strictly by the book.

[austinpop]

45 minutes ago, Always.Learning said:

Thanks for the generous offers on power supplies, Rajiv and Ken! Now that I think about it, I've actually got a 5v Jameco that I could try. And I will of course try spanning the moat with the LPS-1 -- assuming a negligible or zero sonic hit, that would be my preference. My DAC is a Chord DAVE, so I'm hoping that its own galvanic isolation will help (though I recall someone saying that they preferred to turn on the GI with their ISO REGEN even when using the DAVE). 

 

Awesome. You're all set. Keep us posted.

[austinpop]

11 minutes ago, Johnseye said:

 

... qualified server ...

 

How would you define this? What "qualified server" is a fair comparison with an endpoint-based trifecta?

 

Something like @mozes's VR Mini-powered Nimitra server? Or @lmitche's optimizations? It's a moving target.

 

I've said it before - the only one on this thread who is truly spanning both worlds is @romaz, and I for one, am content to enjoy my setup until he returns with a report of his explorations.

 

 

[austinpop]

1 hour ago, lmitche said:

...many of us tried the 2 and 3 box solution 2 and 3 years ago. We then simplified to a one box solution.

 

...Later generations of CAers try or retry the multi-box solution.  That's great,  but some of us have been there and done that.

 

My sense is it's more of a pendulum than a linear evolution!

 

But whatever floats your boat.  

[austinpop]

A few days ago, @romaz wrote this about his "reviewer friend":

 

On 7/2/2017 at 6:36 PM, romaz said:

My reviewer friend now has both the IR and the tX-USBultra in his possession.  Not wanting to steal too much of his thunder as he has yet to publish his review on his site, what I will say is that he loves them both and prefers them both together just as you do although he prefers his IR after the tX-USBultra and so once again, personal preference.

 

And today, this came out. https://audiobacon.net/2017/07/11/uptone-audio-iso-regen-review/

 

Looks like this may be the friend! His conclusions seem to be pretty much in line with Eric's and my review, with the exception of the position of the ISO-Regen in the chain.

[austinpop]

Another 50 hours on the ISO-R (up to about 170 hours), and I feel it continues to improve even at this stage of burn-in. I've spent another long session this evening with another album I keep coming back to of late:

 

 

I've really run out of superlatives. I find myself sitting down for a few minutes and spending hours!

 

I did want to mention a couple of tweaks that I think have finally banished the last remnants of the slightly thin SOtM house sound that the ISO-Regen didn't quite mitigate.

• Suppress the noise backwash from the Meanwell energizing supplies - I have 3 of these, remember!
  • Put all of these on a PS Audio Dectet strip and attach them to a B&K 1604A isolation transformer
  • Replace the Meanwell power cords with modestly upgraded cords - Pangea AC-14SE MkII
  • Both of the above contributed to a blacker background, and sounds more relaxed and smooth
• Swap the balanced interconnects from the DAC to the amp. I've been enjoying the Ayre Signature's for quite a while now, but I happened to pick up a pair of really old (10+ years at least) Cardas Golden Cross cables from a friend.
  • Man, these just vanquished the last trace of thinness!
  • This is an amazing cable, as it should be for something that retailed for more than $1k in the day. 

In the next day or two, we should be doing another listening session at Eric's, so will report on that soon. I suspect many folks will want to know how this combo sounds:

• modded switch > sMS-200ultra > ISO-Regen > SU-1

More soon.

[austinpop]

6 hours ago, BigAlMc said:

Hi Rajiv, 

 

Thanks for the post. Though I'm not sure I follow the above. 

 

Are the 3 meanwell's the power supplies for your LPS-1 collection? 

 

And the Dectet is a power conditioner. So are you finding that even though the meanwell's are switching PSUs themselves they still benefit from getting a cleaner AC supply? 

 

I'm curious as my serious kit is all on an IsoTek conditioner. Or on LPS-1"s. But the supplies for everything else are on a £50 basic socket with a filter. I forget the name. 

 

So is your hypothesis that "dirty" AC noise still makes it through the LPS-1? 

 

Many Thanks, 

Alan 

 

Yes the Meanwells are for the LPS-1s.

 

My system is already pretty resilient to AC mains noise. My P5 Regenerator, which completely regenerates the input AC, has sonically demonstrated its positive impact.

 

I don't put too much stock in the Dectet other than it's a good power strip. It was previously plugged into an outlet on the P5, and the Meanwells plugged into the strip.

 

I think my experiments are confirming that while the LPS-1s do an effective job blocking AC mains noise from getting to the DC output, the Meanwells are putting noise back into the AC, as SMPSes do. 

 

The B&K IT between the P5 and the Dectet strip blocks that "backwash."

 

The effect of the power cable swap is fairly typical, I'd say.

[austinpop]

4 hours ago, lmitche said:

Rajiv, 3 Meanwells wow! Looks like you've done a great job of isolating them, however you may be surprised by the sound quality impact of a good, not great, LPS replacement for the Meanwells.

 

With a 12 volt Sigma 11 based lps to power 2 lps-1s  the difference in SQ is significant here. The B&K 1604a IT powers the Sigma 11 lps with a Supra AC cable.

 

 

 

Yeah Larry,

 

I need to try LPSes to energize. Is the Sigma 11 a DIY, or something you can buy?

 

I have one el cheapo on the shelf, so I'll try that on one and see if I can hear anything further. I need to confirm with Alex though - wasn't there some issue wth it not handling instantaneous current peaks well?

 

But yes, more experiments necessary. Always!

[austinpop]

2 hours ago, AmusedToD said:

 

@austinpop, eager to read the impressions, my setup for now is sms200Ultra > Curious link USB > SU-1 (stock) > i2S HDMI > Holo Spring level 3

 

I wonder if you could start with this setup (just the sms200Ultra > SU-1), and then add component after component (the modded switch, then the ISO Regen) and try to describe the level of perceived sonic improvements. It would mean a lot to me in terms of making a buying decision in the future.

 

Many thanks in advance!

 

I'm sure we can do that. But the effect of the switch is easy. We've done that before. It always makes a huge positive impact. Regardless of whatever else is or is not in the chain.

[austinpop]

1 hour ago, Johnseye said:

 

Good stuff Rajiv.  Lots of changes going on.  When I hear you say "finally banished the last remnants of the slightly thin SOtM house sound" and "these just vanquished the last trace of thinness" I think to myself, what are we really trying to achieve with these devices?  If we think in painting terms, are we just painting different shades or colors of the music to achieve a sound that is more pleasing to our ears?

 

Some would call it coloring - and I wouldn't argue. But to me, every system needs to be dialed in to be ultimately pleasing to the listener for the "long term." Loaded term, for those of us constantly tweaking! 

 

Quote

And I wonder how you came to the conclusion that SOtM products sound thin because I've come to that same conclusion myself.  What is your point of reference in which a device sounds fatter that you are measuring against?  I found that the microRendu sounded fatter.  However the sms-200 was slightly clearer while also sounding thinner at the same time.  I think it may be hard to get one without the other, but I chose thin and clear vs. fat and clouded.

 

Hah - great minds (ears) think (hear) alike! I felt this dichotomy too when I did my first comparisons of stock sMS-200, mR, and Aries Mini here: