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

[incus]

43 minutes ago, Advieira said:

Look my system based on Sotm Diagram.

 

 

 

I get your stab at humor but the diagram is no more complicated than any serious listener's rig -- power supplies, 2 clocks, 2 switches, server, USB card, dac, amp and speakers, along with cables to connect them. Obviously not many people are going to have two clocks and two switches but the rest is pretty standard.

[Balázs]

2 hours ago, guiltyboxswapper said:

@Balázs in your PSU setup above, is it me or are the LT3045s being fed from an already regulated PSU, thereby somewhat limiting final performance?  i.e. the slower regulators in the chain can end up degrading performance.

 

Yes the LPSU is based on the standard output of the HDPLEX 200.

 

First I tried to get direct AC input from the transformer. Although it worked (I had to use diodes on the Stammheim board) I didn't notice any sonic benefit against the DC driven approach. I am no expert at all but I think the slower regulators don't really matter if they are before the faster ones in the regulation chain.

 

On the other hand my goal was to improve on the HDPLEX 200W I already had not to build a new one. I didn't really want to bypass its protection circuits (Overvoltage , Overload protection,Overcurrent protection, short circuit protection)

[Balázs]

2 hours ago, sunny_time_99 said:

Did you compare other regulators, before you build this monster psu? imho you can paralell as much 3045 as you want, it will never sound really good. 3045 is not specially designed for audio use, like others. It got nice specs but it allways sounds flat. Ripple is only one aspect in psu design. There are many ways to reduce ripple...

 

You are absolutely right and I didn't invent to design a new PSU. I just wanted to have a single source of good quality DC instead of many smaller PSU based on my existing HDPLEX 200. I really like what the 3045 did to the HDPLEX. It can be that it's not the 3045 that makes the difference but the HDPLEX represents a rather low quality power.

[sandyk]

1 hour ago, Balázs said:

I am no expert at all but I think the slower regulators don't really matter if they are before the faster ones in the regulation chain.

 

 In fact they will make life easier for the faster regulators that will then see a much more stable voltage input, and only need to deal with HF rubbish that can sail almost unattenuated through a generic slower voltage regulator. 

 However it may be worthwhile to increase the capacitance at the output of the slower regulator to help maintain a flatter output impedance to well past 100kHz. Remember that the LT3045 etc.  are designed mainly for extremely low noise performance for instrumentation power supplies , and the manufacturers make no claims in their Data sheet as to their suitability for Hi Fi Audio.

 

See also post #13999 by sunny_time_99

[drjimwillie]

3 hours ago, Balázs said:

 

You are absolutely right and I didn't invent to design a new PSU. I just wanted to have a single source of good quality DC instead of many smaller PSU based on my existing HDPLEX 200. I really like what the 3045 did to the HDPLEX. It can be that it's not the 3045 that makes the difference but the HDPLEX represents a rather low quality power.

 I have the 300 W HDPLEX ATX LPS.  I am not using my 19 W output. I’m having a regulator built for me to step down and 19V to 12 V for my NAS. 

 Currently the HDPlex ATX powers my pipeline server/HQPE player. The 12V powers an LPS 1.2 for my iso-regen.  

I would like to use regulators to clean up the power from my ATX but if I put LT3045s in line won’t it drop the voltage and reduce the compatibility with the  computers requirements?

 

 Also I would be using the adjustable output to power the LPS-1 for my SOTM USB  card but the DC cord from the LPS-1 does not fit into the SOTM USB card. Has anyone else had an issue like this?  I ordered a new DC cord from Ghent hopefully the tolerances will be better?   SOTM  confirms that the external power input is 5.5, 1.2 

[seeteeyou]

i9-9900K
https://ark.intel.com/content/www/us/en/ark/products/186605/intel-core-i9-9900k-processor-16m-cache-up-to-5-00-ghz.html

 

i9-9900KF
https://ark.intel.com/content/www/us/en/ark/products/190887/intel-core-i9-9900kf-processor-16m-cache-up-to-5-00-ghz.html

 

i9-9900T
https://ark.intel.com/content/www/us/en/ark/products/191044/intel-core-i9-9900t-processor-16m-cache-up-to-4-40-ghz.html

https://www.kikatek.com/Product/1177160-CM8068403874122-INTEL-CLIENT-CPU-TRAY-CORE-I99900T-210GHZ-LAUNCH-15TH-MAY-IN

https://www.morecomputers.com/a?pid=CM8068403874122>>Intel

 

Finally we're getting 8 relatively affordable cores at 35W TDP, though here's another choice with 30W TDP

 

https://www.gamepc.com/shop/products?sku=M11SDV-8CT-LN4

https://www.wiredzone.com/supermicro-components-motherboards-single-processor-m11sdv-8ct-ln4f-10028801

[guiltyboxswapper]

The problem with the multi-core stuff is when used heavily, will they induce their own noise signature (= thereby impact audio performance) ?  I suspect some trial/error (e.g. underclocking the CPU in general) to find this out will be required.

 

I'm personally keeping an eye out on the AMD r1000 embedded series, for this very reason.

[seeteeyou]

Both AMD R1606G and R1505G are dual core processors, therefore still counted as multi-core stuff. Their TDP should be ranging anywhere from 12W to 25W so we could compare them to quad core processors like i7-8650U or i7-8565U

 

https://ark.intel.com/content/www/us/en/ark/products/124968/intel-core-i7-8650u-processor-8m-cache-up-to-4-20-ghz.html

https://ark.intel.com/content/www/us/en/ark/products/149091/intel-core-i7-8565u-processor-8m-cache-up-to-4-60-ghz.html

 

Those mobile processors should be meant for a Roon Endpoint since much lower TDP should be preferable.

 

OTOH, we're definitely looking for something more powerful for building a Roon Server while its TDP should be low enough to satisfy the requirements of a fanless chassis. Hence AMD EPYC Embedded 3000 looked so interesting with 8 cores at only 30W TDP. An i9-9900T should be another favorite with 35W TDP that's coming later this month.

 

BTW, I'm looking forward to yet another powerful NUC in the future with either i9-9980HK or i9-9880H as they're mobile processors

 

https://ark.intel.com/content/www/us/en/ark/products/192990/intel-core-i9-9980hk-processor-16m-cache-up-to-5-00-ghz.html

https://ark.intel.com/content/www/us/en/ark/products/192987/intel-core-i9-9880h-processor-16m-cache-up-to-4-80-ghz.html

 

If Akasa Turing (105 euros + VAT) were any good for Bean Canyon with 28W TDP without having to worry about throttling issues, then I suspect that something similar could also be done for mobile i9 with 45W TDP

 

http://www.akasa.com.tw/search.php?seed=A-NUC45-M1B

http://www.fanlesstech.com/2019/04/the-akasa-turing-is-official.html

[Balázs]

12 hours ago, drjimwillie said:

I would like to use regulators to clean up the power from my ATX but if I put LT3045s in line won’t it drop the voltage and reduce the compatibility with the  computers requirements?

 

This is someting I'd also like to do but , there are some challenges that should be addressed first.

 

ATX specification

The specification of the ATX24 interface needs 3.3/5/12/-12VDC. This mean you should "produce" or convert a lot from your HDPLEX to get to these voltages.

 

Power

Another issue the power need. According to Molex  a fully populated 24 pin Wire-To-Board connector could deliver 8A per pin. The HDPLEX 400 specification of the ATX output: 
 

Output Voltage

Max DC Current

Full Load DC Current

Output Voltage Tolerance

Ripple (mV)

---

+12V	20A	18A	2%	3mV
-12V	1A	0.5A	2%	5mV
+5V	16A	13A	2%	3mV
+3.3V	16A	13A	1%	3mV
5V SB	3A	12A	1%	5mV

The "strongest" LT3045 Stammheim board has 5A max... 

 

Voltage drop

The Schottky diodes drop 0.4V  and the LT3045 has also some loss but very low. The Stammheim board drops 0.8V of DC input. I'd really use a higher DC output of the HDPLEX to get down to the neeed exact voltage value.

 

I would suggest to get a single DC powered motherboard like the ones @romaz and @seeteeyou mentioned but you'd like to optimize your Roon/HQP  server that takes much power. This will be a rather difficult task

[seeteeyou]

Supermicro M11SDV-8CT-LN4F turned out to support a single 12V rail for its power input

 

https://www.supermicro.com/manuals/motherboard/EPYC3000/MNL-2172.pdf#page=21

https://www.supermicro.com/manuals/motherboard/EPYC3000/MNL-2172.pdf#page=38

 

Maximum power consumption seemed to be 50.4W

 

https://www.servethehome.com/supermicro-m11sdv-8ct-ln4f-review-with-amd-epyc-3201-in-mitx/4/

 

Since the output current of LSIB-HPULN PS is up to 5A, that means 12V output should be good for 60W and it's more than enough

 

https://www.mpaudio.net/product-page/lsib-hpuln-ps

[guiltyboxswapper]

Here's what i'm tracking 

https://www.mouser.co.uk/ProductDetail/UDOO/SC40-1000-0000-C0?qs=sGAEpiMZZMspCjQQiuQ1fFTDrDpp2YD1F9tsYJDSIn3bTNxS3kflpQ==

[rickca]

11 hours ago, seeteeyou said:

Akasa Turing (105 euros + VAT)

In the USA, I now see it at Newegg for $150.

[lmitche]

2 hours ago, rickca said:

In the USA, I now see it at Newegg for $150.

In a vertical orientation this case puts the cables a long way from the surface of the table or shelf underneath. In the horizontal orientation it's better, but roughly double the height of the X7 or X8  pizza box design.

 

Outside of aethestics, a matter of taste, what are the benefits of this new case? Better cooling maybe? 

[BigAlMc]

"Perfect for audiophile enthusiasts"

 

First time I've noticed that on the Akasa site. I guess we've been buying a lot of them 😁

 

Would be more perfect if it would fit on my rack! But interesting all the same.

 

Cheers,

Alan

[seeteeyou]

13 hours ago, rickca said:

In the USA, I now see it at Newegg for $150.

 

98.66 quids and 879 RMB respectively, $130 or so

 

https://www.lambda-tek.com/Akasa-A-NUC45-M1B~sh/B43300798

https://item.taobao.com/item.htm?id=592557668564

 

Basically there's a previous generation called Baby Canyon while the processors were quite from each other

 

https://ark.intel.com/content/www/us/en/ark/products/codename/95586/baby-canyon.html

 

7.5W to 15W TDP

https://ark.intel.com/content/www/us/en/ark/products/95442/intel-core-i3-7100u-processor-3m-cache-2-40-ghz.html

 

9.5W to 15W TDP

https://ark.intel.com/content/www/us/en/ark/products/97539/intel-core-i5-7260u-processor-4m-cache-up-to-3-40-ghz.html

 

23W to 28W TDP

https://ark.intel.com/content/www/us/en/ark/products/97541/intel-core-i7-7567u-processor-4m-cache-up-to-4-00-ghz.html

 

Then Akasa started selling their Plato X7 despite such significant gap in TDP, in the end they just added this kinda "friendly reminder" specifically for i7 model(s) even though that's pretty much like BS

 

http://www.akasa.com.tw/download/compatibility/Akasa 7&8th Gen NUC Chassis Compaibility V13.pdf

Quote

** TDP/ BIOS may need to be limited/configured down

 

Could Domino's / Papa John's actually take the heat? Not THAT much room inside the box and therefore we could tell what the deal is. IMHO you're a sucker if you're expecting their Plato X8 to be any good for Bean Canyon since all of them are sharing the same 20W to 28W TDP (i.e. throttling should be no surprise?)

 

https://ark.intel.com/content/www/us/en/ark/products/codename/126382/bean-canyon.html

 

BTW, they also "experimented" with Skull Canyon while we're talking about 35W to 45W TDP for an i7-6770HQ

 

https://ark.intel.com/content/www/us/en/ark/products/89187/intel-nuc-kit-nuc6i7kyk.html

https://ark.intel.com/content/www/us/en/ark/products/93341/intel-core-i7-6770hq-processor-6m-cache-up-to-3-50-ghz.html

 

Honeymoon didn't seem to last very long and that NUC ended up with premature death

 

https://www.reddit.com/r/intelnuc/comments/6zq4yo/my_akasa_galactico_case_review_for_intel_skull/

 

What about Turing then? My uneducated guess would be turning the entire case into a gigantic heat sink and see how that might / might not be able to handle something with 20W to 28W TDP.

 

IMHO that could be potentially a logical explanation since their Galactico was more like a joke, though we still couldn't tell if Turing were good enough for Bean Canyon yet

 

https://www.notebookcheck.net/Intel-NUC-Kit-NUC8i7BEH-i7-8559U-Mini-PC-Review.360356.0.html

Quote

When under higher loads, consumption will jump to 47 W or higher.

 

The highest consumption we can record is 78 W from the small (~12.3 x 4.5 x 3.0 cm) 90 W AC adapter when running both Prime95 and FurMark simultaneously.

 

Obviously that's something to be taken into consideration if we're running something like Extreme 2 with AL, otherwise the role of Roon Endpoint etc. shouldn't consume that much power to begin with.

[lmitche]

34 minutes ago, seeteeyou said:

 

98.66 quids and 879 RMB respectively, $130 or so

 

https://www.lambda-tek.com/Akasa-A-NUC45-M1B~sh/B43300798

https://item.taobao.com/item.htm?id=592557668564

 

Basically there's a previous generation called Baby Canyon while the processors were quite from each other

 

https://ark.intel.com/content/www/us/en/ark/products/codename/95586/baby-canyon.html

 

7.5W to 15W TDP

https://ark.intel.com/content/www/us/en/ark/products/95442/intel-core-i3-7100u-processor-3m-cache-2-40-ghz.html

 

9.5W to 15W TDP

https://ark.intel.com/content/www/us/en/ark/products/97539/intel-core-i5-7260u-processor-4m-cache-up-to-3-40-ghz.html

 

23W to 28W TDP

https://ark.intel.com/content/www/us/en/ark/products/97541/intel-core-i7-7567u-processor-4m-cache-up-to-4-00-ghz.html

 

Then Akasa started selling their Plato X7 despite such significant gap in TDP, in the end they just added this kinda "friendly reminder" specifically for i7 model(s) even though that's pretty much like BS

 

http://www.akasa.com.tw/download/compatibility/Akasa 7&8th Gen NUC Chassis Compaibility V13.pdf

 

Could Domino's / Papa John's actually take the heat? Not THAT much room inside the box and therefore we could tell what the deal is. IMHO you're a sucker if you're expecting their Plato X8 to be any good for Bean Canyon since all of them are sharing the same 20W to 28W TDP (i.e. throttling should be no surprise?)

 

https://ark.intel.com/content/www/us/en/ark/products/codename/126382/bean-canyon.html

 

BTW, they also "experimented" with Skull Canyon while we're talking about 35W to 45W TDP for an i7-6770HQ

 

https://ark.intel.com/content/www/us/en/ark/products/89187/intel-nuc-kit-nuc6i7kyk.html

https://ark.intel.com/content/www/us/en/ark/products/93341/intel-core-i7-6770hq-processor-6m-cache-up-to-3-50-ghz.html

 

Honeymoon didn't seem to last very long and that NUC ended up with premature death

 

https://www.reddit.com/r/intelnuc/comments/6zq4yo/my_akasa_galactico_case_review_for_intel_skull/

 

What about Turing then? My uneducated guess would be turning the entire case into a gigantic heat sink and see how that might / might not be able to handle something with 20W to 28W TDP.

 

IMHO that could be potentially a logical explanation since their Galactico was more like a joke, though we still couldn't tell if Turing were good enough for Bean Canyon yet

 

https://www.notebookcheck.net/Intel-NUC-Kit-NUC8i7BEH-i7-8559U-Mini-PC-Review.360356.0.html

 

Obviously that's something to be taken into consideration if we're running something like Extreme 2 with AL, otherwise the role of Roon Endpoint etc. shouldn't consume that much power to begin with.

Thankfully the 15 watt Comet Lake U-series is on the way, so we can skip the 28 watt Bean Canyon release and can continue to avoid high current machines. In the meantime the 15 watt Kaby Lake R NUC machines sound terrific and run cool in the Akasa X series cases.

[Balázs]

Has anyone tried an Intel NUC Roon endpoint with the JCAT femto cards yet? Using a PCI riser dual adapter for the external USB and NET cards could be a SOtM reclocking alternative, couldn't it? 

[davide256]

6 hours ago, lmitche said:

Thankfully the 15 watt Comet Lake U-series is on the way, so we can skip the 28 watt Bean Canyon release and can continue to avoid high current machines. In the meantime the 15 watt Kaby Lake R NUC machines sound terrific and run cool in the Akasa X series cases.

In the FWIW column,  just discovered with the 8i3BNH if you select "custom power" mode in BIOS, allows you to select a low power mode... which still won't work

with an LPS 1.2

[Dutch]

18 hours ago, Balázs said:

Has anyone tried an Intel NUC Roon endpoint with the JCAT femto cards yet? Using a PCI riser dual adapter for the external USB and NET cards could be a SOtM reclocking alternative, couldn't it? 

 

SOtM say they will release a mPCI(M.2) and PCIe compatible network card themselves in Munich. Not sure what it’ll look like and if it’s small enough to fit inside a NUC or one of the Akasa NUC enclosures. 

 

https://www.sotm-audio.com/sotmwp/english/highend2019/

 

Check this link for a ‘how to’ JCAt NET card “inside” a NUC. I don’t believe anyone active in this thread has done this. Several did install the NET card inside the server.

https://jcat.eu/install-jcat-cards/

[Balázs]

53 minutes ago, Dutch said:

 

SOtM say they will release a mPCI(M.2) and PCIe compatible network card themselves in Munich. Not sure what it’ll look like and if it’s small enough to fit inside a NUC or one of the Akasa NUC enclosures. 

 

https://www.sotm-audio.com/sotmwp/english/highend2019/

 

Check this link for a ‘how to’ JCAt NET card “inside” a NUC. I don’t believe anyone active in this thread has done this. Several did install the NET card inside the server.

https://jcat.eu/install-jcat-cards/

 

Thanks @Dutch. A couple of days ago I ordered an M.2>PCIe adapter and a dual slot riser card. As soon as they arrive I will be able to try them with my JCAT cards and NUC7i7DNKE.

 

https://de.aliexpress.com/item/M-2-NGFF-NVMe-M-Key2280-To-PCIe-3-0-4x-Riser-Card-Cable-PCI-Express/32833359557.html?spm=a2g0x.10010108.1000016.1.7b204ca20TfFxB&isOrigTitle=true

 

https://de.aliexpress.com/item/Dual-slot-pice-PCI-E-X16-extension-card-2U-PCI-E-video-card-graphics-video-card/33003556007.html?spm=a2g0x.10010108.1000016.1.5d6c48f6Vea8Z8&isOrigTitle=true

 

One more gadget I'd like to play with is the picoPSU. I know that these units have already been discussed here but I'm not sure about how they perform when connected with the 19V output versus the ATX output of the HDPLEX 400W. The single 19V option also provides an easy way to improve by regulators (LT3045 etc.) Last but not least I found a pretty strong 200W picoPSU "clone" called Dynamo Mini by KMPKT. Interestingly, it's HDPLEX who builds this PSU for KMPKT.

 

https://www.sfflab.com/products/kmpkt_dynamo_psu

 

P. S. Thanks a lot for the information regarding the coming SOtM netword card. Now I understand why May refused to clock the JCAT cards saying they "don’t modify the products for other brand products"  

 

 

 

 

[Marcin_gps]

Hello Guys, 

 

Just a quick update: both JCAT NET Card FEMTO & JCAT USB Card FEMTO are back in stock and all pre-orders will ship on Monday. 

Thanks to @romaz & @austinpop reviews there was a huge demand for the cards. 

 

BTW - if anyone is in Munich next week for the High End show and has any questions or just want to say hi, please visit Hall 2, Booth E01. I will be there 

 

Regards, 

Marcin

[RickyV]

10 hours ago, Marcin_gps said:

Thanks to @romaz & @austinpop reviews there was a huge demand for the cards. 

 

👍 good for you, Marcin👍👍

[str-1]

Just received a new tX-USBultra with all the upgrade options, including silver dc wiring, Evox caps, eABS paper and 75ohm master clock input.  I’ve been told it will take about a week to fully run-in but being a little impatient I’m keen to learn from others what to expect from the Evox caps in particular as they run-in.

[auricgoldfinger]

4 hours ago, str-1 said:

Just received a new tX-USBultra with all the upgrade options, including silver dc wiring, Evox caps, eABS paper and 75ohm master clock input.  I’ve been told it will take about a week to fully run-in but being a little impatient I’m keen to learn from others what to expect from the Evox caps in particular as they run-in.

 

You should expect a more dynamic presentation.

[tedwoods]

On 5/3/2019 at 12:49 PM, Dutch said:

Not sure what it’ll look like and if it’s small enough to fit inside a NUC or one of the Akasa NUC enclosures. 

It looks like that: