[Wijnand]

Hi Tobias, I’m aware that the clock at the switch has no importance from the timing perspective. Still, normal crystals are pretty noisy. The Ultra clock is one of the best in the world. It has phase noise levels above 130 dbv at 10hz. When you swap the crystal by this clock it’s clear in seconds. You feel almost like a life attendant at a concert. At first when the clock is still cold, the phase noise levels are not that good, but after a few hours it’s already much better and only after 48 hours it’s fully stabilized.

In the first few hours it only gets better, until it decreases again after 6 hours. Slowly the DAC clock is pushed out of phase. This happens everytime when the stream becomes too clean.

I think when the signal is weakened by the filters and edges become less steep due to smoothing then everything is okay. I’m really positive now that it’s not only about noise. A stream can get too stable as well and cause trouble to the DAC clock.

[Wijnand]

There actually are two things of importance for audio networks.

1. Noise

2. Clock synchronisation

Unfortunately those two are biting each other. Noise can be prevented by using high grade power supplies and or noise can be canceled by ethernet filters.

When there’s almost no noise at the ethernet signal the blockwave will stabilize and push other devices with its stable signal. Other devices will slowly follow the rhythm of this stabilized stream. This means the clock of the audio device will synchronize (in a certain amount) as well to this stable signal, but the audioclock should not run at 25Mhz, it should run at 24.756Mhz. When the audioclock is out of phase the sound will darken.

The winding patrons of cablechokes can cancel noise and it can cause or prevent clock synchronisation. The cleaner the signal the more important it is not to create induction at the signal. Induction will cause more explosive pulsations which will cause clock synchronisation.

So the real deal is to create a noise free blockwave, but also preventing the network clocksignal to cause trouble.

What I have done Yesterday with the Ultra clock switch was really incredible. But I must switch off this clock each time after listening. This will prevent the clock to stabilize. Really the first few minutes are on a level that you can’t even imaging even when you have heard it 🙂

[Wijnand]

Hi Paul,

I can’t tell why Eric choose that winding patron. Maybe he can answer that question 😉

Yes, I made changes over and over. I tested so much.

SFTP cat5e sounds just better in my system. I have tried at least 5 UTP versions. Also a CAT8 version, but the CAT8 clearly sounded worse (more woolly). I think CAT8 can transport noise better as well

You can see in the picture that I don’t have a shield connection at the switch 🙂

[Wijnand]

The ultimate network. This clock brings extreme quality to music. In my network are 3 core boxes and one ultra switch. The coreboxes brought a major impact on the sound, but still the clock dubbels the impact. I have to wait at least 24 hours before I can say if the clock can stay in the network, because in the past it always dubbeled the network trouble as well. But when life music is played with this clock switch even the public applauding will bring goosebumps

[Wijnand]

This balanced winding patron prevents energy fields in the core. An energy field will create noise on the stream. On low quality cores its less important.

[Wijnand]

This winding patron is just wrong. With strong cores the problem is clear audible. I will show a better patron tomorrow 🙂

[Wijnand]

Well maybe I can build a filter for that 😉

[Wijnand]

Thanks Paul, Yes I’m still very happy with the sound. I’m listening to all kinds of setups. So there’s not one particular answer to give on your question.

For my setup there’s one thing that I really dislike and that is ferrite. When I add ferrite to the circuit then the sound is not that good. Ferrite blocks on the output cable have a certain impact ‘impedance’ like for example 200 Ohm at 100Mhz, but also for example 90 Ohms at 25 Mhz. Which means when there are several cores than also the normal data range is affected. I guess in systems with switchmode powersupplies and other devices that causes noise. Then ferrite could be better.

[Wijnand]

Just edit and remove the lines that are weird 🙂

 

 

[Wijnand]

That would be nice. Than can work 4 days a week 😉

[Wijnand]

[Wijnand]

[Wijnand]

I’m working on a first non commercial batch of 5 core boxes. These are built after I have tried several prototypes.

[Wijnand]

This seems to work to upload a picture in the private section and then it can be used on the forum

 

In this picture you see a balanced wound core. This core is even bigger than the ones I used before. From the JCat to the core the cable is shielded. In the core and behind the core the cable is unshielded. There’s a slieving around the cable. The core is filled with hot glue

 

[Wijnand]

Today I have tried to take the core box another step up, but that seemed not that easy. I have tried some extra windings on the main cores. I have tried a second (smaller) cores per pair. I have tried copper tape to create internal shielding. I have tried aluminium foil in layers. I have tried EMI absorbers inside. I have added some cores on the output cable. It all has an effect.

The internal copper tape shielding is a yes!

The EMI absorbers are a Maybe

The pair of smaller cores is a big NO! (A strong field inside the box was created)

The aluminium foil is a no

The extra windings on the main core is a no (because a stronger field is created)

The external cores on the output cable is a yes!

I will soon build a serie of 6 for some friends 🙂

 

[Author]

Posts