Layout of Ethernet Isolators

I wouldn't know about the USB output but I suspect these won't be clean as well.

Although I have a weak spot for the Touch is has long been surpassed by many better devices. The attractive features of the Touch today are the touch display (many newer devices don't have a display at all) and general appearance, not its audio qualities. They once were though. Operating them is a joy IMO and I miss mine.

Please be warned that the PCB material of the Touch is very weak. One has to be a real "Bringer" to have the guts to work on it. I used to change the SPDIF RCA to good industrial quality BNC 75 Ohm connectors and this took many hours and it was a high risk job. I thought changing the XO's and all surrounding stuff to be a too high risk job and never did that. Decoupling the power supply lines also was rewarding. As you found out the switcher as delivered with the Touch is of no use in audio. It is a good transmitter though and a convincing example to ban SMPS from audio completely.
 
Last edited:
I wouldn't know about the USB output but I suspect these won't be clean as well.

Although I have a weak spot for the Touch is has long been surpassed by many better devices. The attractive features of the Touch today are the touch display (many newer devices don't have a display at all) and general appearance, not its audio qualities. They once were though. Operating them is a joy IMO and I miss mine.

Please be warned that the PCB material of the Touch is very weak. One has to be a real "Bringer" to have the guts to work on it. I used to change the SPDIF RCA to good industrial quality BNC 75 Ohm connectors and this took many hours and it was a high risk job. I thought changing the XO's and all surrounding stuff to be a too high risk job and never did that. Decoupling the power supply lines also was rewarding. As you found out the switcher as delivered with the Touch is of no use in audio. It is a good transmitter though and a convincing example to ban SMPS from audio completely.

I am also in love with this little box. This is the reason i have not retired the touch yet.
As i only use it as a Ethernet to USB Bridge for my RME ADI 2 DAC, i dont care about its sound quality of its DAC.
 
I really want to stay with the LMS solution, because I can use it to adjust the stream from Qobuz to my LS50 Wireless via EQ to the room via SOX.

I'm also converting the stream from Qobuz to PCM on the LMS to relieve the touch.

As an alternative, a PI 4 would still be possible instead of the Touch, after having solved the performance problem with USB.
But I'm not sure if this would really be an upgrade.

Allo USBBridge is also an option - but I suspect that this was only built because the PI 3 USB had problems and is now playing at the same level as the PI 4.
 
No one told you not to but please note so called "system choices" are not the most optimized choices when one wants performance to be top notch. Before you know it you have replanted so much trees in the forest that you can't find the way out. Well experience tells that then "beliefs" start to roar their ugly head just as "testing by ears only".

Many here will be pretty much convinced that using standard (i.e. not hampered with) UTP cables to just 1 switch will also improve matters. Sometimes one must go back to the basis.

Did I read "wireless"?
 
Last edited:
The frequencies are of course not in the audio range, but can have a negative influence on the conversion from D to A.

There are also common mode chokes (CMCs) in these xformers.

Assuming the transformer (incl. CMC) has positive effects, it may be that this is mainly due to the CMCs.
Would it then make sense to use only CMCs instead of an isolator (transformer + CMCs)?
For the frequencies >10Mhz the xformer does not isolate due to its interwinding capacitance. In that case the cmc choke is the only option that makes sense. To check this, you can wind you cable through several ferrite clamps. This is an effectice CM-choke for these frequencies.
Btw I am not convinced that sound is affected at all by some RF-noise. Do you have any measurements showing this noise, distortion or whatever this is?
 
The frequencies are of course not in the audio range, but can have a negative influence on the conversion from D to A.

There are also common mode chokes (CMCs) in these xformers.

Assuming the transformer (incl. CMC) has positive effects, it may be that this is mainly due to the CMCs.
Would it then make sense to use only CMCs instead of an isolator (transformer + CMCs)?
For the frequencies >10Mhz the xformer does not isolate due to its interwinding capacitance. In that case the cmc choke is the only option that makes sense. To check this, you can wind you cable through several ferrite clamps. This is an effectice CM-choke for these frequencies.
Btw I am not convinced that sound is affected at all by some RF-noise. Do you have any measurements showing this noise, distortion or whatever this is?
 
Btw I am not convinced that sound is affected at all by some RF-noise. Do you have any measurements showing this noise, distortion or whatever this is?

Problem with this is noise is ... noise. A while ago I tried to make the most audible noise that would barely show up on a FFT. A spike at random intervals did the trick, very audible, yet it is not periodic, so it just shows up as a raised noise floor on audio frequency FFT, which looks innocuous. Yet I noticed a... correlation... between the HF noise floor and the sound quality (and also jitter).

Which of these two ES9023 DACs sounds better than the other?
 

Attachments

  • 40k -66dBfs fs96_freq4.png
    40k -66dBfs fs96_freq4.png
    99.5 KB · Views: 130
...I tried to make the most audible noise that would barely show up on a FFT. A spike at random intervals did the trick, very audible, yet it is not periodic, so it just shows up as a raised noise floor on audio frequency FFT, which looks innocuous. Yet I noticed a... correlation... between the HF noise floor and the sound quality (and also jitter).

IMHO people should be doing more of those types of experiments and measurements. Its something that has been overlooked in audio for too long.
 
  • Like
Reactions: 312elements
Unfortunately, I don't have any suitable measuring instruments myself.

All I can say is that I have just connected a self-made isolator (without termination of the center taps) with separation of TX and RX to two separate transformers to reduce cross talk to my LS50 Wireless and I am overjoyed with the sound. ....
 
IMHO people should be doing more of those types of experiments and measurements. Its something that has been overlooked in audio for too long.

In order of increasing difficulty:

- talking about it
- measuring what is measurable
- measuring what is relevant

Attached:

40k -66dBfs fs96_time.png

Time domain output of same previous ES9023 DACs, one sounds good, the other does not. I don't know if it is as simple as saying "HF noise influences sound quality". That's probably too simple. It could be something a bit more subtle like: a lot of HF garbage in the output indicates the product was not properly designed, which will probably correlate with bad sound quality.

Back to topic, this is like saying "this DAC is so transparent that it allows to hear a difference between digital interconnects!!!!!!!!!!!!!"

Translation: the DAC has no jitter rejection, so it needs a special snowflake SPDIF cable.

TomJones explained the ethernet side of things in this post but not the audio side of things.

If TomJones is feeding SPDIF from squeezebox into DAC through transformer, and squeezebox is connected to ethernet through transformer, and adding another isolation transformer changes the sound, then... something somewhere is being a special snowflake about noise, probably the DAC.

What is the DAC and how is it connected to the squeezebox?
 
Hi Tom, there are some wrongs in your post 🙂 How on earth can you design and build a DIY isolator and not measure it?
The transformer is not made by me, i just used transformers like this: H1102FNL Transformer: Ethernet SMD 0-70degC Trans: RX 1: 1,TX 1: 1-1.1dB PULSE: Amazon.de: Beleuchtung
This was the first i built:

In order of increasing difficulty:

- talking about it
- measuring what is measurable
- measuring what is relevant

Attached:

View attachment 945507

Time domain output of same previous ES9023 DACs, one sounds good, the other does not. I don't know if it is as simple as saying "HF noise influences sound quality". That's probably too simple. It could be something a bit more subtle like: a lot of HF garbage in the output indicates the product was not properly designed, which will probably correlate with bad sound quality.

Back to topic, this is like saying "this DAC is so transparent that it allows to hear a difference between digital interconnects!!!!!!!!!!!!!"

Translation: the DAC has no jitter rejection, so it needs a special snowflake SPDIF cable.

TomJones explained the ethernet side of things in this post but not the audio side of things.

If TomJones is feeding SPDIF from squeezebox into DAC through transformer, and squeezebox is connected to ethernet through transformer, and adding another isolation transformer changes the sound, then... something somewhere is being a special snowflake about noise, probably the DAC.

What is the DAC and how is it connected to the squeezebox?

Squeezebox is connected via USB with Intona Isolator and then with RME ADI 2 DAC
 
Last edited:
You have something going on with isolators....

I just love the German way of overengineering/overreacting. I was in a training for heavy industrial equipment with German engineers, there were also some African engineers as guests. They blew up a machine by accident (it could have happened to anyone, don't get me wrong) and the German engineers all were at once very busy meticulously examining the defects from all angles. The African guys sat together. I went to them and asked what was up. They replied that if stuff blows and matters gets complicated the first thing they are used to is sit together and have a laugh about what went wrong, all glad that no one was hurt 😉 The solution would still follow...
 
Last edited: