[Wijnand]

I just built the Ferriet core solution for that last meter ethernet. I hear a difference 🙂

[Wijnand]

I will look into it for sure thanks for the link. However I don’t use Wifi at all. About the batteries I once did a test with a Wavelet Dac of a friend. I had a few 18650 cells from Sanyo Li-Ion. And also from panasonic. For the comparison I created a Mosfet 12Vdc powersupply. I must say it was a close call, but the powersupply came on top in a blind test. After this test I have used the batteries in my network for a while and disconnected the chargers when listening. It was pretty good and cheaper than a good powersupply. But then there was Farad.

Are you familiar with the noise floor of a battery? Farad power supplies are lab tested and have a significant lower noise floor than a battery. A few companies that used batteries for testing sensitive equipment switched over to use Farad Super 3’s for even better results. Not to mention the Super 10’s

The wrong reply button happened to me as well in another thread 🙂

[Wijnand]

Here’s also a picture from a router on a switching powersupply for the comparison with the above

[Wijnand]

In addition to the above, the more I cleaned up the stream, the more trouble seemed to appear. Normally I would say if noise is the problem then it must be audible immediately. It should not take a few days to get effect. With crystals it’s another thing they can slowly synchronize to other crystals. And this seems to be the problem when a stream became to clean. I think that is the problem I’m facing.

[Wijnand]

Hi Tobias, Thanks! and yes of course I understand what you’re telling me. The reason I connected the PC was because my network was in trouble. The only thing in my network was a router fed by an ultra clean powersupply. I also made scope pictures back in 2020 which made clear the data stream was very clean. 

[Wijnand]

About point 5 ethernet cables. I must say that I was convinced for more than a decade. That the quality of ethernet cables are very important for high end audio systems. I think that I wrote over 500 articles on the PS audio forums back in 2o12 and also more than a few back in 2015 on the Alpha Audio website. I tested and bought really high grade cables in the past and they did make a clear difference, but in the past few years I also found out that when an ethernet system is tuned to the max. That the difference in cabling will be still there, but I can’t say I like high grade cables over normal cables any longer. Actually normal cables sound more neutral than

[Wijnand]

Thanks for this article!

In the first paragraph is written; common mode current on the wires. I have to think about this, because an ethernet connection to a switch is always galvanic isolated with the shield as an exception.

In the second paragraph is written; That data is not being corrupted in any way. I know that’s true, but what about the shape of a bit? The bits are just an analog signal that is interpreted as digital data. And on the DAC it’s the other way around. I think the shape of the bit is an important factor.

In the third paragraph is written; In other words that data is just passed trough, but what about noise? I think that noise can be past true as well. Think about this: an electrical noise spike is converted to light and on the other side converted back from light to an electrical noise spike.

Personally I found out that at least in my system fiber optics can’t compete with copper devices. I really tried hard. I still have a few hundred meters of fiber optic cables and several converters. I think the clock signal is passed trough. The data tap is built by a clock and sent by a clock and catched by a clock.

I will look into the link you have sent and maybe see you back in a few months. Good luck!

[Wijnand]

Yes, I had that as well the first few times. Now I copy my text into a text file

[Wijnand]

Hi Paul,

Well let’s find out about point 3, 5 and 7

About point 6, my experience is that it always start good, but when the crystals seem to synchronize the sound ‘collapses’

But I like your statement line better 🙂

 

Point 3, is the one I’ve been puzzling on for 3 years now. I had a Pink Faun X2.16 streamer with 3 Ultra OXCO clocks inside. At that point I was still using my old NAD M12 modified with 2 Pink Faun clocks. One Audioclock and one Clock on the BluOS module. I also replaced the 2 power supplies by two Farads Super 3’s. One 24Vdc and one 5Vdc.

The problem was that new devices, new cables always sounded great, but after 3 day’s the sound always became dull. From that point the search began. I replaced my cabling, my power blocks, modified the streamer over and over, replaced power supplies and drilled a new ground electrode. Replaced my Dac 2 times. And after that I found out it was the ethernet which caused all the trouble. Not at once, but in day’s.

Last year I modified a switch with an Ultra OCXO clock with incredible results, but after 48 hours the sound was more dead than alive. The interesting part is that it takes 48 hours as well for the clock to stabilize. In other words the more it stabilized the worse the sound. Of course the switch itself was not the problem. The problem is that other ethernet devices seemed to synchronize more and more to this Ultra clock.</p>

[Wijnand]

Well maybe some pictures can be added later. With my current network I can outperform my internal SSD with an internet stream. To do this I split my network in 2. The first part is from the internet connection directly to my PC’s mainbord ethernet connection. And the second part is from the PC’s Jcat Net Card XE directly to my streamer DAC. The internet fiber converter (WAN) is fed by a Farad Super 3 powersupply with a QSA fuse. The router is a semi professional one (without Wifi) from Draytek. In which I drilled a hole for an external DC connection. The Draytek is Fed by a 12V Farad Super 10. Then the second part, the Jcat is Fed by a Farad Super 10 as well. The Super 10’s are switched in floating mode which means the DC side is floating from ground, but the external housing is connected with ground and also connected with the Router and the Jcat.

And now comes the fun part 🙂

The network is as simple as can be and fed by high grade power supplies. The sound of the audio system became dark in only 1 hour. So it seemed the signal is too clean and the crystals synchronized almost immediately. What I did then was, I added a spare switch to the network from which the only connection was to the router. I used the stock powersupply for the switch. The idea was to inject noise into the router, in other words to destabilize the routers crystal. Immediately the sound cleared up. After this I gave it a few hours and it seemed it was not enough yet. So I connected a computer to the switch and now the sound cleared up completely.

I think the result can be explained best with a metaphor. It was as if a high grade 8K television had almost no contrast at all, details were kept hidden. There was almost no backlight and the picture was dark. After I connected the switch to inject noise into the router the television reached Full HD immediately. After I connected the computer with the switch the television reached ultimate pictures and showed it is a true 8K television with mind blowing detail.

The best switch for the Job seemed to be A D’link DGS108. This switch measured the strongest crystal signal. Higher voltage then other switches. Higher voltage logically means higher noise.

[Wijnand]

I once looked into the relation between input and output impedance. The higher the input impedance the easier it will be to drive. I’ve tested devices up to 200K Ohms. For example a Classé CP800 has a 100K input impedance and a CP700 had 200K input impedance. And a 100 Ohm output impedance. This means the current that is needed to drive the CP700 is only 50% for what is needed for the CP800. Poweramps and subwoofers are most of the time somewhere in the range of 15K to 50K Ohms.

Now you can ask yourself what the difference is. Well this depends on the unit that’s driving. Just make sure that the unit is able to drive lower impedance than the unit that must be driven. Besides this on higher input impedances I would use even better cables, because the signal with less current is most logical more easy to be influenced by electromagnetic fields.

How they will sound together is depended on far more parameters

Hope this helps!

[Wijnand]

Tidal is een streaming dienst. Dat is geen installatiesoftware. Je zult een programma moeten installeren waarmee je Tidal kunt gebruiken. Bijvoorbeeld Roon, Eupony Stylus, Jriver, Audirvana etc.

[Wijnand]

Is there some way to share a few scope images here? Or should I go to a forum group for this?

[Wijnand]

I think these statements will cover the complete path to understanding. How to create the ultimate ethernet system.

In the past 3 years I addressed a problem with my system. I had dark sound and my system was not satisfying for the tiniest bit. I worked on it as a second job… almost 8 hours a day and it caused me many nights. Frustration levels reached the top of a mountain. In total I have tested literally more than 200 situations. Which made me conclude that the manufacturers were not able to set us on the right path yet. This actually is pretty logical, because they do not build equipment to create the ultimate network. They create equipment to deal with a noisy network.

The average audiophile probably will be aware that ethernet is an important aspect for the quality of streaming audio. When adding just a high grade switch, cable or a powersupply. Then most of the time this will help to get a better sound. When the sound is still better after 48 hours then I would say that your ethernet system is not on the edge of what is possible.

[Wijnand]

7. Realy stable clocks should be avoided (when the network is aready high grade)

[Author]

Posts