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

[Middy]

Theres 2 free tips you can do then on your gear...a little bit of silicone around network ports for anti vibration and desolder noisey but useful LEDs.. courtesy of innuos..

I havent seem anyone try that silicone trick on Caps 3 dots at the base or O-rings.. if it doesnt hurt then it doesnt hurt to try... i tried o rings from a trick they use on valves..

Would the desoldered leds need to be bridged to complete the circuit??

Nice to see if it does help SQ...

Or just buy the innous..😁

[Fourlegs]

1 hour ago, str-1 said:

Expensive new entrant to the audio-grade switch market https://innuos.com/phoenixnet/

I guess I will have to get one on home demo! They must be pretty sure it does improve things with that price tag.

[Jonder]

I had a Melco S100 on loan and I couldn’t tell the difference with the led’s on or off!

[barrows]

10 minutes ago, Jonder said:

I had a Melco S100 on loan and I couldn’t tell the difference with the led’s on or off!

LEDs are extremely low noise devices.  One "might" be able to hear a difference in the context of a state of the art world class phono stage (super high gain analog circuit) but no one will ever hear a difference in the context of a noisy, high speed, digital device such as a Network switch.  I would not trust anyone's claim of hearing a difference in that context-furthermore, I would have little faith in any manufacturer who claims LED noise contribution is audible in the context of a Network switch, as this would be a clear indication that they are making things up, rather than actually measuring noise levels.

Let us all consider that LEDs are often used as ultra low noise references for both current and voltage regulation in very high end analog circuitry, which is much more nose sensitive than a Network switch.

[Blackmorec]

Quoting from the Melco S100 Network Switch Manual 

 

LED Off Switch

Press this switch to turn off the link/act LED

Leaving the link/act LED on may affect the quality of sound. We recommend turning off the link/act LED for typical music playback. 

 

I imagine Melco and I know Innuos measures extensively during development and both recommend skipping the LEDs for critical listening.  Similarly SoTM offers a simple means of disabling all their LEDs via a 3 position power switch. 

 

 

 

 

[barrows]

8 hours ago, Blackmorec said:

Quoting from the Melco S100 Network Switch Manual 

 

LED Off Switch

Press this switch to turn off the link/act LED

Leaving the link/act LED on may affect the quality of sound. We recommend turning off the link/act LED for typical music playback. 

 

I imagine Melco and I know Innuos measures extensively during development and both recommend skipping the LEDs for critical listening.  Similarly SoTM offers a simple means of disabling all their LEDs via a 3 position power switch. 

 

 

 

 

Here are the key wordings from the above:

 

"may affect the quality of sound"  When they say "may" they are clearly admitting that they do not know the answer, if they did know, they would say so.

 

and:

 

"I imagine Melco..."

 

Yes, anyone can imagine things, but imagining things has nothing to do with what has actually occurred.

 

I stand by my analysis that the relatively tiny amount of noise which LEDs make is inconsequential to the overall noise footprint of a component like a Network switch, which includes very noisy processing chips in its circuitry. 

 

[MarkusBarkus]

[Blackmorec]

13 hours ago, barrows said:

Here are the key wordings from the above:

 

"may affect the quality of sound"  When they say "may" they are clearly admitting that they do not know the answer, if they did know, they would say so.

 

and:

 

"I imagine Melco..."

 

Yes, anyone can imagine things, but imagining things has nothing to do with what has actually occurred.

 

I stand by my analysis that the relatively tiny amount of noise which LEDs make is inconsequential to the overall noise footprint of a component like a Network switch, which includes very noisy processing chips in its circuitry. 

 

Semantics aside,  the Melco includes 11 LEDs and the SoTM sNH-10G has 19, all powered from the same power supply that’s driving the data stream. 

 

I have found power supplies on network switches to have a disproportionally large affect on SQ.  If you were designing a streaming device to place in front of a server with the express purpose of improving sound quality, and your plans include an ultra-low noise, multi rail LPS specifically designed to place the final regulators right next to the active components in order to avoid picking up any interference, why would you then include a bunch of flickering LEDs fed by the same power supply?  

Innuos’s design engineering addresses several key factors:

Power supply quality

Power supply regulator intimacy with the powered components in order to avoid any cabling on the DC supply to chips and clocks

Clock intimacy with the devices they time and control.

Vibration avoidance and suppression

Internal EMI avoidance

External EMI shielding 

Clock accuracy and minimisation of noise interference 

 

I have found that applying these same criteria to the entire network, from start to finish results in increasingly large SQ gains and the inversion of the law of diminishing returns. The more noise you take out of your network supply, the better it keeps getting and the greater the gains become. What would be an insignificant and perhaps infinitesimal improvement on a noisy network becomes a massive gain when most noise sources have either been removed or ameliorated. This is the reason Melco use the word “may” rather than “will” as the actual result achieved depends entirely on the status of the rest of the network.  What they are clearly acknowledging is that by definition, they can’t know what the final outcome will be, when they don't know the network status. 

 

My system uses a wi-fi (input) to ethernet (output) bridge, powered by a highly optimised LPS. The bridge includes the capability to switch off all LEDs for light pollution reasons; however doing so also has quite easy to hear benefits on sound quality. 

[Mahler and Bach on Computer]

8 hours ago, Blackmorec said:

My system uses a wi-fi (input) to ethernet (output) bridge,

Hi, wonder if you can share how to achieve this from the network settings for a computer.  I have been searching this for a while and had posted a question here as well.  Many thanks!

[R1200CL]

Some of us is dependent on good USB to SPDIF converters. I as many others, bought a modified Singxer SU-1 based on recommendation in this tread. 
 

Have anyone upgraded to SU-2 ?

And of cause is SQ equal or better ?

 

My reason to consider this upgrade, is the external clock input.

I assume I still can use external 5V input? (As with SU-1).

 

Kitsune https://www.kitsunehifi.com/product/ktesu2/ is selling an “upgraded” version. Any idea idea if it’s worth the double price tag ?

 

[Blackmorec]

11 hours ago, Mahler and Bach on Computer said:

Hi, wonder if you can share how to achieve this from the network settings for a computer.  I have been searching this for a while and had posted a question here as well.  Many thanks!

This is based on hardware. When i first set up my system i needed a way to get the data stream from the ISP router up to my hi-fi room, a direct distance of ca 15m. 

I evaluated several different means including some cheap Cat5 ethernet cable, CAT6 Synergistic Research ‘Actve SE’ ethernet,  wi-fi direct, mesh wi-fi (2 different brands). 

By far the best turned out to be a TPLink RE650 extender using its 5GHz band, with the 2.4Ghz band, polling and all LEDs switched off.  Adding a TPLink Archer AC5400 tri-band router meant that i could dedicate one 5GHz band solely to audio. I then modified the RE650, removing its built-in 230v SMPS and wall plug and replacing it with a custom built anti-vibration wall mount and Sean Jacobs 5V LPS. 

 

[Summit]

10 hours ago, R1200CL said:

Some of us is dependent on good USB to SPDIF converters. I as many others, bought a modified Singxer SU-1 based on recommendation in this tread. 
 

Have anyone upgraded to SU-2 ?

And of cause is SQ equal or better ?

 

My reason to consider this upgrade, is the external clock input.

I assume I still can use external 5V input? (As with SU-1).

 

Kitsune https://www.kitsunehifi.com/product/ktesu2/ is selling an “upgraded” version. Any idea idea if it’s worth the double price tag ?

 

 

I would advice you to also take a look at Digital To Digital Converters from Audio GD and Denafrips.

 

[Mahler and Bach on Computer]

3 hours ago, Blackmorec said:

This is based on hardware. When i first set up my system i needed a way to get the data stream from the ISP router up to my hi-fi room, a direct distance of ca 15m. 

I evaluated several different means including some cheap Cat5 ethernet cable, CAT6 Synergistic Research ‘Actve SE’ ethernet,  wi-fi direct, mesh wi-fi (2 different brands). 

By far the best turned out to be a TPLink RE650 extender using its 5GHz band, with the 2.4Ghz band, polling and all LEDs switched off.  Adding a TPLink Archer AC5400 tri-band router meant that i could dedicate one 5GHz band solely to audio. I then modified the RE650, removing its built-in 230v SMPS and wall plug and replacing it with a custom built anti-vibration wall mount and Sean Jacobs 5V LPS. 

 

@BlackmorecThank you very much for spending time to write your implementation up.  I really appreciate it.  The power supply replacement is a good thought.  

[TheAttorney]

4 hours ago, Blackmorec said:

By far the best turned out to be a TPLink RE650 extender using its 5GHz band, with the 2.4Ghz band, polling and all LEDs switched off...  I then modified the RE650, removing its built-in 230v SMPS and wall plug and replacing it with a custom built anti-vibration wall mount and Sean Jacobs 5V LPS.

How big a SQ jump was gained by replacing the p/s? Compared to, say, the improvement of the stock RE650 over one of its rivals?

 

And how easy was it to do the swap? E.g. could the case be opened up by undoing a few screws, or was it molded shut?

[di-fi]

Hi, Blackmorec generously shared detailed information about the tp-link WiFi-router modifications (with pictures). See the link above.

[Blackmorec]

48 minutes ago, TheAttorney said:

How big a SQ jump was gained by replacing the p/s? Compared to, say, the improvement of the stock RE650 over one of its rivals?

 

And how easy was it to do the swap? E.g. could the case be opened up by undoing a few screws, or was it molded shut?

The jump in sound quality was pretty large....for starters it removed a substantial emphasis in the treble region that resulted in everything sounding a lot more natural, pure and enjoyable. Not to mention making the sound more detailed and spacious. 

 

The actual mod was quite easy. Undo 4 corner screws and release a couple of side clips and the RE650 splits down the middle, leaving the SMPS and wall plug in one half and the entire networking infrastructure + heat sink in the other side.  Very convenient. There is a square six pin socket on the network side that mates with a fixed 6 pin plug on the power supply side  3 x 5V and 3 x 0V so its simply a matter of fabricating a 6 pin socket using a small piece of robust circuit board and mating that to an end-mounted 2.1mm barrel connector socket. I removed the 230V wall plug and replaced it with single point ‘button connector’. I mounted the female side of the button connector to a piece of Perspex, which is itself mounted to the wall via 4 o-rings. This mod achieved the following:

Avoids vibration and noise caused by 230V - 5V rectification

Maintains antennae orientation

Removes the cheap-as-chips SMPS

Removes vibration coming from the SMPS’s transformer

Isolates the assembly from wall-borne vibration 

 

Both the AC5400 router and RE650 bridge share a software package called Tether, which allows you to set up security, define/restrict clients, define whether the RE650 is used as an extender or an access point, enable/disable bands, polling etc., switch LEDs on or off etc. 

 

 

   

 

 

[Jonder]

I’ve replaced the smps on my Vodafone router with an £80 Chinese lps from Aliexpess. That made the same sq improvement that @Blackmorec has referred to. Age old thing, get rid of cheap smps!

[nvitorino]

On 2/26/2021 at 11:18 PM, barrows said:

Here are the key wordings from the above:

 

"may affect the quality of sound"  When they say "may" they are clearly admitting that they do not know the answer, if they did know, they would say so.

 

and:

 

"I imagine Melco..."

 

Yes, anyone can imagine things, but imagining things has nothing to do with what has actually occurred.

 

I stand by my analysis that the relatively tiny amount of noise which LEDs make is inconsequential to the overall noise footprint of a component like a Network switch, which includes very noisy processing chips in its circuitry. 

 

Hi, 

 

Just a quick one regarding the LEDs on the ports and the reasoning behind it. This is stated clearly in the network switch design guide (which is a good thing to read if you are designing a network switch from scratch):

 

"If the designed product works inside an electrically noisy outside environment, it is not recommended to use RJ45 with integrated LED. This is because the outside interference signal or voltage is coupled to the LED circuit through the line side of RJ45 due to the LED circuit directly connected to chip and system power/ground."

 

The issue is not just the noise the LEDs create, but the interaction with outside interference. So the difference may be bigger or smaller depending on your environment.

 

Nuno

[robi20064]

Hello Nuno,

 

Which "network switch design guide" is that? Could you please provide a source to that or some ISBN if it was part of a book?

[seeteeyou]

http://ww1.microchip.com/downloads/en/DeviceDoc/KSZ8895RQX-Hardware-Design-Checklist-00003335A.pdf#page=5

http://ww1.microchip.com/downloads/en/DeviceDoc/KSZ8895FQX-Hardware-Design-Checklist-00003367A.pdf#page=6

Quote

• The user can utilize the RJ45 connector with integrated LED components if the product working environment is
not very noisy.
• If the designed product works inside an electrically noisy outside environment, it is not recommended to use RJ45
with integrated LED. This is because the outside interference signal or voltage is coupled to the LED circuit
through the line side of RJ45 due to the LED circuit directly connected to chip and system power/ground. It is better to use independent LED components.
• If the user needs to utilize the RJ45 with an integrated LED circuit in a noisy environment, consider adding TVS
diodes to protect the chip.

 

[BOBO]

have you guys tried to overclock the dac`s usb port to 1000hz using hidusbf ?

[Soul Analogue]

Indeed on the buffalo, turning off the leds does sound substantially better than keeping them on...

much less harshness and having more clear and holographic image

so i purposely put a toggle switch to turn off the leds...

[lwr]

1 hour ago, Soul Analogue said:

Indeed on the buffalo, turning off the leds does sound substantially better than keeping them on...

much less harshness and having more clear and holographic image

so i purposely put a toggle switch to turn off the leds...

Very interesting!  Please provide a photo of the point in the circuitry where such an interrupter switch is connected.  Thanks!

[sandyk]

On 2/26/2021 at 4:48 AM, Middy said:

Theres 2 free tips you can do then on your gear...a little bit of silicone around network ports for anti vibration and desolder noisey but useful LEDs.. courtesy of innuos..

 

Modern  LEDs are not noisy, unless they are multiplexed as in frequency displays etc. It is possible though that if the power supply itself is noisy, that  there may be a small amount of radiation from the leads powering the LEDs .

LEDs are often used as  low noise voltage references in Amplifiers etc.  with voltages typically varying from around 1.5V to 3.3V depending on their colour and materials used in their construction.

[Fourlegs]

1 hour ago, sandyk said:

 

Modern  LEDs are not noisy, unless they are multiplexed as in frequency displays etc. It is possible though that if the power supply itself is noisy, that  there may be a small amount of radiation from the leads powering the LEDs .

LEDs are often used as  low noise voltage references in Amplifiers etc.  with voltages typically varying from around 1.5V to 3.3V depending on their colour and materials used in their construction.

 

See Nuno's post in link below which clarifies why he thinks the inclusion of LEDs can be noisy.