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

rickca · February 23, 2018

15 minutes ago, limniscate said:

I did like the tX on HD-PLEX when the dX was also on the tX--just not when the dX was on the SR4 and the tX was on the HD-PLEX.

OK I am clear about that now thanks.

austinpop · February 23, 2018

2 minutes ago, lmitche said:

The FS switches are 100 mbps, so that means you liked the 1 gb connection best. Over here, the 1 gb nic sounds best set at 100mbps via it's driver configuration.

I actually misspoke earlier. Here is the corrected list. I've also annotated it to indicate the speed the SE server sees.

Configurations:

Wall > SE <---------- sounded the best 1Gbps
Bridge > reclocked switch (LPS-1, but clock supplied by tX-USBultra on HDPlex) > SE 100Mbps
Wall > Netgear FS105 (HDPlex 12V - does not work with LPS-1 at 7V) > SE 100Mbps
Wall > Netgear FS105 (HDPlex, with ground shunted via iFi Groundhog) > SE 100Mbps

@limniscate - that sound right to you? I'm pretty sure this is it, as I was the one crawling around making cable changes while you wallowed in the sweet spot as the blind listener.

welldone · February 23, 2018

3 minutes ago, rickca said:

OK I am clear about that now thanks.

I'm intending to buy a new HDplex 200 to feed 2 Lps1 (1 for SMS200, 1 for TX-USB Ultra). Is that a good idea? Please give your recommendation. Thanks.

limniscate · February 23, 2018

18 minutes ago, lmitche said:

Which hdplex?

Second generation, so not the current one but the 100 watt with the USB-C.

pas · February 23, 2018

6 minutes ago, welldone said:

I'm intending to buy a new HDplex 200 to feed 2 Lps1 (1 for SMS200, 1 for TX-USB Ultra). Is that a good idea? Please give your recommendation. Thanks.

After the first 250 LPS-1.2 orders are shipped, we will begin offering our 7.5V/4.8A/36W adapter separately for $15 to those who want one. Until then, there is a CUI brand 12V/2.5A/30W unit available in the US from DigiKey and other distributors (https://www.digikey.com/products/en?keywords=102-4104-ND. This is one that I personally confirmed connects AC ground to DC output "ground."

Hope this helps.

--Alex C.

I recently purchased and received the CUI supply that Alex mentioned above and the results have been fantastic. I am powering two LPS-1's, which in turn are powering my sMS-200 and HDTA Serenade Pro Dac respectively. My system has never sounded better and they were only $17 each. UpTone will be offering a similar item in the future. FYI!

lmitche · February 23, 2018

22 minutes ago, limniscate said:

Second generation, so not the current one but the 100 watt with the USB-C.

Ok, that makes sense. The current model is said to be a great improvement. I hear there are inherent stability issues with the 100 watt unit.

My 3 year old 250 watt DC ATX was just replaced by the newer 400 watt DC ATX and the latest generation product sounds better, with improved bass, (I had to dial the sub back), and top end extension. Clarity and presence is off the charts with the dual PSU solution.

KingRex · February 23, 2018

What is controlling the SE. Did you bridge the IP address, if that is even possible.

Also, is the SE sensitive to the upstream device controlling it. I thought I read somewhere the Ultrarendu are. Maybe I'm remembering wrong. Just wondering about the Influence that has on an SE.

BigAlMc · February 23, 2018

Hi Eric / @limniscate - great post and it sounds like you and Rajiv are having some real testing fun.

I also have the Cerious Technologies graphene speaker cables after stumbling upon rave reviews on another forum. I'm a fan now and have ordered balanced XLRs too. I can't say I tested against many rivals but I like them and think the punch above their weight for a pretty reasonable price point. Bob Grost is also a pleasure to deal with.

In terms of future tests can I ask that you consider whether LPS-1 vs LPS-2.1 is possible to test?

Cheers,

Alan

Summit · February 23, 2018

9 hours ago, austinpop said:

I think we can also credit the isolation built into the Zenith SE ethernet port. As Nuno wrote on this other forum:

The ethernet ports on the Zenith are no ordinary Ethernet ports you find on normal "consumer-grade" boards. The ethernet ports have 4 separate isolation transformers each to further isolate the connection (4 small grey boxes right before each ethernet port). We also do extensive anti-vibration treatment of the chassis, the LPSU and even the SSD, converting vibration into very small amounts of heat. This makes a very audible difference in sound quality.

The Ethernet board is important, no doubt. It made a big difference going from mobo Ethernet to JCAT LAN. Probably of even more significance if bridging them like I do.

rickca · February 23, 2018

9 hours ago, welldone said:

I'm intending to buy a new HDplex 200 to feed 2 Lps1 (1 for SMS200, 1 for TX-USB Ultra). Is that a good idea? Please give your recommendation. Thanks.

Sorry I've never used an HDPlex 200 so I can't answer your question. I'm sure @Superdad could give you a definitive answer.

seeteeyou · February 23, 2018

16 hours ago, auricgoldfinger said:

I've had the dCBL-UF in my system (between the tX-USBultra and Auralic Vega) for the past 4-5 weeks. I have the silver version with the 5v power. I was skeptical that this cable could possibly be worth anywhere near the price I paid, but the sound is simply stunning. I am extremely happy with it as a replacement for the Lush cable. When I added a single ferrite, the sound improved further.

They're also working on dCBL-BNC as linked below, it's gotta be pretty dope as well

http://www.sotm-audio.com/sotmwp/english/portfolio-item/dcbl-bnc/

Most likely that's meant to match the performance of sCLK-OCX10 so let's see how it goes.

rickca · February 23, 2018

9 hours ago, lmitche said:

Clarity and presence is off the charts with the dual PSU solution.

Can you please document your dual PSU solution? I know part of it is using the HDPLEX 400W HiFi DC-ATX.

auricgoldfinger · February 23, 2018

34 minutes ago, seeteeyou said:

They're also working on dCBL-BNC as linked below, it's gotta be pretty dope as well

http://www.sotm-audio.com/sotmwp/english/portfolio-item/dcbl-bnc/

Most likely that's meant to match the performance of sCLK-OCX10 so let's see how it goes.

I actually bought two dCBL-BNC when I ordered the sCLK-OCX10 since they were being offered at a discounted introductory price when bundled together.

auricgoldfinger · February 23, 2018

10 hours ago, pas said:

After the first 250 LPS-1.2 orders are shipped, we will begin offering our 7.5V/4.8A/36W adapter separately for $15 to those who want one. Until then, there is a CUI brand 12V/2.5A/30W unit available in the US from DigiKey and other distributors (https://www.digikey.com/products/en?keywords=102-4104-ND. This is one that I personally confirmed connects AC ground to DC output "ground."

Hope this helps.

--Alex C.

I recently purchased and received the CUI supply that Alex mentioned above and the results have been fantastic. I am powering two LPS-1's, which in turn are powering my sMS-200 and HDTA Serenade Pro Dac respectively. My system has never sounded better and they were only $17 each. UpTone will be offering a similar item in the future. FYI!

Did you have to buy a separate DC cable? The photo only shows the adapter.

lmitche · February 23, 2018

1 hour ago, rickca said:

Can you please document your dual PSU solution? I know part of it is using the HDPLEX 400W HiFi DC-ATX.

The other part is using my evga 1600 watt for the 12 volt CPU power only.

kilroy · February 23, 2018

12 hours ago, lmitche said:

Ok, that makes sense. The current model is said to be a great improvement. I hear there are inherent stability issues with the 100 watt unit.

My 3 year old 250 watt DC ATX was just replaced by the newer 400 watt DC ATX and the latest generation product sounds better, with improved bass, (I had to dial the sub back), and top end extension. Clarity and presence is off the charts with the dual PSU solution.

Not disagreeing, but this is interesting in light of what Larry told me just a few weeks ago. I have the previous DC-ATX unit and asked him about sound quality improvements with the 400 watt unit. He said there are no differences in SQ, the 400W was designed only to provide more power.

My impressions from dealing with Larry on several products in the past few years is he does not design specifically for sound quality at all. He has also said that about the different versions of his LPSs.

There are some users (forget if here or elsewhere) that say HDP products do not compare favorably with some other high end ones known to the audiophile community. But I am intrigued to hear that someone says that maybe this is not the case.

Cornan · February 23, 2018

40 minutes ago, kilroy said:

Not disagreeing, but this is interesting in light of what Larry told me just a few weeks ago. I have the previous DC-ATX unit and asked him about sound quality improvements with the 400 watt unit. He said there are no differences in SQ, the 400W was designed only to provide more power.

My impressions from dealing with Larry on several products in the past few years is he does not design specifically for sound quality at all. He has also said that about the different versions of his LPSs.

There are some users (forget if here or elsewhere) that say HDP products do not compare favorably with some other high end ones known to the audiophile community. But I am intrigued to hear that someone says that maybe this is not the case.

I think there is to many Larry's here to keep my mind straight, or is @lmitche the Larry you are talking about? ?

Cornan · February 23, 2018

14 hours ago, austinpop said:

I actually misspoke earlier. Here is the corrected list. I've also annotated it to indicate the speed the SE server sees.

Configurations:

Wall > SE <---------- sounded the best 1Gbps

Bridge > reclocked switch (LPS-1, but clock supplied by tX-USBultra on HDPlex) > SE 100Mbps

Wall > Netgear FS105 (HDPlex 12V - does not work with LPS-1 at 7V) > SE 100Mbps

Wall > Netgear FS105 (HDPlex, with ground shunted via iFi Groundhog) > SE 100Mbps

@limniscate - that sound right to you? I'm pretty sure this is it, as I was the one crawling around making cable changes while you wallowed in the sweet spot as the blind listener.

1Gbps always sounded better than 100Mbps in my setup (except during a short period of time with my 100Base-TX experiments with silver plated starquads and FMCs). Direct connection from main router to switch sounds better than through a wireless bridge when the AC mains is "fully" isolated. With a not so isolated AC mains the wireless bridge is better. Aqvox switch-8 always improves SQ. BTW, I only use cloud based music (as you know...but maybe not everyone else).

When it comes to fuses. Try the AHP fuses. They are cheaper and they are better than everything else, but Furutech fuses IME. I have tried most of them. Be sure to treat the fuses with DeoxIT. It makes a big difference!

pas · February 23, 2018

6 hours ago, auricgoldfinger said:

Did you have to buy a separate DC cable? The photo only shows the adapter.

No, it is attached.

austinpop · February 24, 2018

22 hours ago, limniscate said:

@austinpop and I did a lot of testing today on my system and had some interesting findings.

We did sighted comparisons at first and then blind testing on me. I definitely learned that testing on my own is useless because I'm always biased against improvement. Before today, I thought the tX-USB Ultra and Mutec Ref10 basically did nothing to my system and couldn't tell any difference between the Pasternack and Digikey clock cable either.

Before I list the tests and results, here is a reminder of my system:

Zenith SE -> tX-USB Ultra via Phasure Lush -> dX-USB Ultra via Curious -> Schitt Yggy Rev. A via Transparent Reference AES -> ARC Ref6 via Shunyata Python XLR -> Sanders Magtech via Groneberg TS Premium single-ended -> Magnepan 3.7i via Cerious Technologies Graphene Extreme speaker cables. Mutec Ref 10 clocking the dX with a Pasternack 50 ohm BNC to SMB cable and an Amphenol 75 ohm BNC to BNC to the tX.

We did the swap the Curious for another Lush at one point. The dX and tX were powered by a 2nd gen HD-Plex at 12v at first. We switched in a Paul Hynes SR-4 later on and also tried two Uptone LPS-1 in series (which we've tried before).

First we turned off the Mutec Ref 10 with the dX-USB Ultra and tX-USB Ultra in my system. Like the Cybershaft, the Mutec added a sense of three dimensions and realness. The image seemed larger as well.

Next, we took out the dX and just used the tX directly from the Zenith to the Yggy. This is when we swapped in the other Lush, which remained for the rest of the testing. As expected, the Yggy sounded best on AES. It'll be interesting to see how much the gap closes if I get the Gen5 USB upgrade on the USB.

Then we took out the tX-USB Ultra and put the dX back in. As reported before, the tX-USB definitely added weight.

Then we swapped the Paul Hynes SR-4 for the HD-Plex on the dX-USB and put the tX-USB back into the chain. Wow, I had forgotten how sensitive SOtM gear is to power supplies. This was the most obvious change of the day.

At this point, we started blind testing on me. I believe @austinpop started with the Mutec Ref 10 in and out of the chain as the first test. Like the sighted test, I was able to correctly identify whether it was in the chain or not.

Then he tested me on the tX. This is where it got interesting: so the system sounded worse with the tX on the HD-Plex but the dX on the SR4. Remember above that my system sounded better with the tX in it, but that was with the dX and tX both on the HD-Plex! This was shocking to both of us and underscores just how important a good power supply is.

Next, he tested me on something that we had tested in the past but is baffling to both of us. He tested me with the SOtM modded switch in and out of the system. The switch was powered by an Uptone LPS-1 at 5v and was connected to the tX via a Digikey 50 ohm SMB to SMB clock cable. Upstream of the switch were two SOtM dCBL-CAT7 with a SOtM iSO-CAT 6 to a USB NIC that's connected to a USB hub, which is connected to a standard PC (bridge mode). Another dCBL-CAT7 connected the modded switch directly to the Zenith SE.

With the switch out of the system, the Zenith SE was connected directly to an ethernet port on the wall which connects to a switch and router in another room. Therefore, this mode has two more SMPS upstream of the Zenith SE compared to with the modded switch in the chain. Amazingly, I preferred the chain without the modded switch! The image sounded bigger and more dense on the low end. This is the second time I blind tested this and chose the chain without the modded switch. How can this be?

Our first hypothesis was that maybe the iSO-CAT6 was causing the difference. Remember, we've tested the iSO-CAT6 before and couldn't distinguish any effects. So Rajiv blind-tested me without the modded switch but with and without the iSO-CAT6. Neither of us could distinguish any effect. Our second theory is that it might have something to do with the HD-Plex powering the tX, which is connected to the switch via the Digikey clock cable. More testing will have to be done when Rajiv gets his LPS 1.2 and I get my SR7.

The next blind test that we did was with grounding a Netgear switch aka John Swenson trick. It was very subtle, but I think I did prefer the grounded Netgear switch to the ungrounded Netgear switch.

After that, we tried the grounded Netgear switch versus no switch. Amazingly, we again preferred no switch in the chain!

The last test that we performed was a Pasternack 50 ohm BNC to SMB clock cable from the Mutec Ref10 to the dX versus a Digikey 50 ohm BNC to SMB clock cable from the Mutec Ref10 to the dX. Surprisingly, I preferred the Digikey. We used Daft Punk's Game of Love HDTracks version, and the cymbals were definitely crisper with the Digikey than the Pasternack.

TL;DR

1. Power supplies make a big difference on SOtM gear.

2. Mutec Ref 10 has a nice effect on realism (more 3D-sounding) and image size.

3. tX does seem to add a heft/weight on the bottom end.

4. For some unknown reason still, the modded switch sounds worse in my system than direct to the wall.

5. Digikey BNC to SMB cable sounds better than Pasternack.

Upcoming tests:

Mutec Ref 10 versus Cybershaft OP-14

LPS 1.2 versus SR4

Clarity USB cable versus Lush

As an aside, I haven't posted about my Zenith SE versus SOtM gear because I want to wait for my SR7. I suspect the gap will close considerably when the SOtM gear is on the SR7.

19 hours ago, austinpop said:

I actually misspoke earlier. Here is the corrected list. I've also annotated it to indicate the speed the SE server sees.

Configurations:

Wall > SE <---------- sounded the best 1Gbps

Bridge > reclocked switch (LPS-1, but clock supplied by tX-USBultra on HDPlex) > SE 100Mbps

Wall > Netgear FS105 (HDPlex 12V - does not work with LPS-1 at 7V) > SE 100Mbps

Wall > Netgear FS105 (HDPlex, with ground shunted via iFi Groundhog) > SE 100Mbps

@limniscate - that sound right to you? I'm pretty sure this is it, as I was the one crawling around making cable changes while you wallowed in the sweet spot as the blind listener.

These findings at Eric's house had me very intrigued - not to mention puzzled - so I had to try it chez moi. My results are described below. A couple of clarifying points:

Eric doesn't stream music from Tidal, or even from a local NAS. His music was flowing from a local SSD straight out through USB. Therefore the network's role is only to allow communication between a Roon control app and the Roon Core. The fact that the network connection has an audible effect was puzzling. Also - several of you mentioned pulling the plug on the ethernet cable to see what happened. On Roon at least, it appears that causes playback to stop. Maybe there's a trick to make this work, perhaps with the internal Squeezelite player instead.
In my house, "Wall" refers to an upstream router on the far end of my house, behind about 100 ft of Cat 5e in-wall cable. While the router is on an LPS, as is my broadband modem, and an attached switch, I do have a few SMPS-powered devices up there, most notably a NAS.

My findings

In my setup too, my Roon Core on the Zenith SE serves up music stored on the local SSD. However, I do stream from Tidal as well. My network connection options were:

Wall > dCBL-Cat7 > SE
Wall > BJC Cat 6a > Netgear GS-105 (Breeze LPS, JSGT) > dCBL-Cat7 > SE
Wall > BJC Cat 6a > Netgear GS-105 (Breeze LPS, JSGT) > Supra Cat 8 > sCLK-EX modded Zyxel (LPS-1, JSGT) > dCBL-Cat7 > SE

A. Playback from local SSD

I first tried this mode, mimicking Eric's experiment. In my house, and with my network, I can sum up the result succinctly: no difference between configurations 1, 2, and 3.

B. Playback from Tidal

Here, the results are different. As expected, configuration 3 sounds quite dramatically better, while 2 is just slightly better than 1. In mathematic terms, 3 >> 2 ≥ 1.

In summary, my findings are the same as I've reported before, but it always helps to revisit findings from time to time. It also goes to show that every situation and every system is different!

rickca · February 24, 2018

On 2/19/2018 at 6:10 PM, rickca said:

Bottom line seems to be that powering stuff from the motherboard (USB, SATA disk, ethernet) is just bad for sound quality.

So now I have an SR4 powering my internal X25-M SSD and an LPS-1.2 powering my Startech PCIe USB 3.1 Gen2 card. My micro iDSD with an ISO REGEN/LPS-1 is plugged into the Startech card. I plan to get a Samsung T5 external SSD for music files and plug it into the Startech card as well. The T5 is bus powered, but now clean bus power is supplied by the LPS-1.2 powering the Startech card.

I may decide to switch the 3 power supplies around ... for example, I may prefer to use the LPS-1.2 for the ISO REGEN, and I may need to use the 2A SR4 to provide enough juice for the Startech USB card peripherals.

I also have a Startech USB3 ethernet adapter plugged into the Startech card. This means my ethernet adapter is similarly getting clean power from the LPS-1.2. It is part of a bridged network adapter and my ultraRendu is attached to the Startech USB3 ethernet adapter via Blue Jeans cat 6A cable.

To clarify, I don't use the ISO REGEN and ultraRendu at the same time. I either attach the micro iDSD to the ISO REGEN or to the ultraRendu for a listening session, depending on whether I want a direct USB or a network streamer configuration. I'm hoping that A+ for Windows comes out soon.

So this is a pretty versatile setup. I'm just using a Z170 motherboard and an i7-6700K CPU. This rig was originally built as a gaming/productivity machine rather than an audio PC.

The funny thing is that these 3 LPS probably cost as much as the rest of the rig.

PS: My main audio system has an Zenith SE controlled by iPeng9 on an iPad Pro. I need a reliable turnkey solution somewhere! This other stuff is for experimentation to see how scarily close I can make it sound.

austinpop · February 24, 2018

Hello? Is this thing on?

A little tip for those of you using a reference clock with the tX-USBultra. Nominally, there is no visual indicator on the exterior to show whether a reference clock attached to the BNC input on the back is working correctly.

Well, there is. All credit to @limniscate for pointing this out to me. I just never used my eyes to notice this before.

Peeking through the top of the tX into the box, you'll notice a blinking green LED on the sCLK-EX board. What's less obvious, is that adjacent to it is another LED that is unlit by default. When you connect a clock on the input, and it correctly syncs with the sCLK-EX board, this 2nd LED lights up with a steady green.

Take a look and tell me you're seeing this too.

kilroy · February 24, 2018

7 hours ago, Cornan said:

I think there is to many Larry's here to keep my mind straight, or is @lmitche the Larry you are talking about? ?

Neither. I'm sure @lmitcheknows who I'm talking about.

d_elm · February 24, 2018

26 minutes ago, austinpop said:

Hello? Is this thing on?

A little tip for those of you using a reference clock with the tX-USBultra. Nominally, there is no visual indicator on the exterior to show whether a reference clock attached to the BNC input on the back is working correctly.

Well, there is. All credit to @limniscate for pointing this out to me. I just never used my eyes to notice this before.

Peeking through the top of the tX into the box, you'll notice a blinking green LED on the sCLK-EX board. What's less obvious, is that adjacent to it is another LED that is unlit by default. When you connect a clock on the input, and it correctly syncs with the sCLK-EX board, this 2nd LED lights up with a steady green.

Take a look and tell me you're seeing this too.

Those with the sMS-200ultra are not so fortunate as the sCLK-EX is mounted upside down and the LEDs can only be seen when

the boards are removed from the box. T8 six point torex for the box screws.

rickca · February 24, 2018

7 hours ago, Cornan said:

I think there is to many Larry's here to keep my mind straight, or is @lmitche the Larry you are talking about?

There's also a Larry at HDPLEX.

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

Recommended Posts

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Create an account or sign in to comment

Create an account

Sign in