Improving digital coax cable by galvanic isolation

[jls001]

This is not analog interconnect but SPDIF coax is still an interconnect so I thought I'll add it here.

Here's a very interesting recipe to improve your digital coax cable by providing galvanic isolation between the source and destination by inserting an isolation transformer in the cable itself.

Improve Your Soundcard's S/PDIF Output - All

You'll need an old Ethernet LAN card (10/100 will do). Each LAN card is equipped with a pulse transformer module which is usually installed very close to the actual RJ-45 port. The pulse transformer contains a number of tiny isolation transformers (typically four) that completely isolates (galvanic) the Ethernet transceiver circuitry from the UTP cable. The transformer can handle upto about 20 MHz frequency, which is sufficient to handle 24/192 data.

Intrigued by the idea, I risked cutting a perfectly working Belden 1855A SPDIF cable, about 1.5 inch from one RCA connector end and soldered the pulse transformer.

It has drastically raised the performance of the cable. The sound is just cleaner. In my chain, what used to be a soft bass when using this cable has tightened up substantially.

The only difficult things in this project are desoldering the pulse transformer from the LAN card, and making an acceptable housing for this "network".

To desolder, you need a suction gun and lots of patience. First suck away as much of the solder as possible from each pin, then heat up one row of pins while patiently prying it out. Gripping the pulse transformer in a vice grip helps immensely. Do be gentle and do not force it. I broke two pins due to my impatience.

For covering it, I used 12 mm diameter heatshrink to cover the assembly. After adjusting the heat shrink, use hot air gun to shrink it, then tie the shrunken portion to the cables using small nylon cable ties. Use one more layer of heatshrink to hide the cable ties, etc.

Experiment with direction - keeping the transformer nearer to source, or nearer to destination.

 

[captrajesh]

Though the link contains some pics, we would greatly appreciate if you could post some pics of your mods.

 

[elangoas]

Am planning to use one of the PC's Motherboard SPDIF out (Co-axial) header to connect to AVR, also have a unused LAN card with 10/100 capability.

Would this DIY help to improve the output from PC ?

 

[rikhav]

Not much as the clock on the mother board are really bad ones

I have read about people even upgraidng the motherbosrd clock. Check out on computer audiophile forum

 

[keith_correa]

Not much as the clock on the mother board are really bad ones

I have read about people even upgraidng the motherbosrd clock. Check out on computer audiophile forum

Apples and Oranges! Galvanic isolation and the clock are different animals. One has little to do with the other.

 

[rikhav]

Apples and Oranges! Galvanic isolation and the clock are different animals. One has little to do with the other.

Hi Keith
My point was that for mother board putting this kind of diy effort is not worth maybe

 

[jls001]

Actually, if you read the Instructables article, the mod is meant to isolate the "gunk" from the computer going into the DAC. So it should work for elangoas as his specific setup is similar.

But it should work in any SPDIF application, I think. Works very well for me to isolate DVD player (coax SPDIF output) which I'm using to spin CDs and DAC (actually PC sound card SPDIF input; I'm using the sound card as DAC).

 

[manniraj]

@jls, I have a spare Pulse transformer with me lying around. Does this work as per the instructions provided for the digital cable?

 

[jls001]

Please check for what application this pulse transformer is intended to know whether it can handle the max sampling frequency of SPDIF interface (192 kHz).

For example, the pulse transformer in the 10/100 Mbps LAN card is more than sufficient to carry 24/192 bit rate without degrading the signal as this pulse transformer is designed for much, much higher bit rate of 100 Mbps.

Also, there doesn't seem to be a standard pin out. In your link, the transformer has four pins so it is a straightforward case of finding which are the two pins for source end and destination end (this is not just notional and MAY NOT BE safe to interchange terminals as some pulse transformers have built in low pass filter. The LPF end is intended to face the computer side.

Also, in the interest of full disclosure, please understand that the characteristic impedance of Ethernet lines are 100 ohms and not same as the 75 ohms of SPDIF lines. However, I'm ignoring this inconvenient bit since the cable is very short (1 to 1.5m at the most) so reflection losses due to mismatched transmission lines is low. The proof is that I still use the same volume level as before.

 

[jls001]

Some pictures to show construction steps.

Below is a test setup to determine the correct leads where the two ends of the coax have to be connected. Solder short leads of about 1.5" to the leads after identifying the best candidates from the pulse transformer data sheet. On the printed surface of the pulse transformer you'll see a small dot. Just like in integrated chips this dot signifies pin number 1. The correct pins in my case are pin 10-11 and pin 15-16.

Solder the coax cable as shown below. Just two tiny soldered points per coax cable, with no scope for tying the cables to the transformer body.

Cover the assembly in appropriate sized heat shrink (12 mm heatshrink worked for me), then use cable ties to secure it at both ends.

Additional layer of heat shrink tubing for some measure of cosmetics. Still it ain't gonna win no beauty contest

 

[Fantastic]

Maybe you should check if your DAC does have a transformer at it's input already. Many above average systems do. Often they have a dc blocking capacitor as well . So checking with a multi meter may not be possible. You could open it up and check it.

 

[jls001]

Yes, Fantastic that's true. Some CD players (and DVD players) are equipped with the pulse transformer at the SPDIF output. I guess even some DACs would be, too.

Made another cable today for a friend. Curiously, the pulse transformer, taken from another LAN card, is of the same brand but had only 12 pins unlike the earlier one which had 14. Had to start afresh in figuring out the correct pins but in the end it worked.

PS: wondering aloud if this would do any good in an analog interconnect? My last pulse transformer can be sacrificed for this important scientific endeavour

 

[rikhav]

Thank you Joshua for taking the pains of modding my coax cable

I use a burson dac and as source use a 2 PC setup (amd machine as music server with daphile or jriver and raspberry pi as music renderer with moode audio. All with dlna streaming

I was using my cdp lesser and lesser as comparing to my computer source it was sounding way too dead. Dead as in no life in music, no energy when music had.
Mind you there is nothing wrong with the cdp. Its just that I was getting mode used to resolving sound of the PC as transport

I tried the modded coax cable between my rpi and dac and after listening to a few tracks I could easily make out the sound was mucb more cleaner then before and this resulting in details coming out better
Bass was more tighter then before but as it is before too was quiet good. Tonally it was not changing much and I was sort of happy with it as I was happy with the tonality I used to get.
Whatever it had achieved was also quiet astonishing as I never expected my setup to be more resolving then what it was

Today I used the modded coax cable between cdp and dac. After listening to about 4 albums I can say I have new found respect for my CD player again

Okay, suddenly its not become as resolving as my PC setup. But it's just way more musical then before. I don't find the sound anymore dull or lifeless
The mids are more resolving so listening to vocals is treat and way the macro and micro dynanmics sound jow , it never used to sound before
I find it hard to explain but things like pulling of strungs in a guitar, some scratching sound is way more audible then before

If you go to see the overall sound of the setup , the mod in the cable has made it more musical and resolving. Sort of bringing out the strengths of the cdp and PC in a better way

 

[elangoas]

@jls001

My friend helped me make the SPDIF Out (Co-Axial out) from Motherbaord's SPDIF Out header.. We had used cabinet's front panel connector wires to do this..http://www.hifivision.com/home-thea...pdif-out-back-panel-bracket-2.html#post687436

Weird thought - Plz help to clarify..

Is it a good idea to bring the galvanic isolation chip in-between the motherboard's SPDIF out header and the Co-axial out of the recent made bracket?

 

[jls001]

Yes, you can put the isolator between motherboard and the bracket. In fact that's a better place than on the SPDIF cable.

 

[Fantastic]

PS: wondering aloud if this would do any good in an analog interconnect?

These are ferrite core types. Only suitable for high frequency use. Low frequency audio would saturate the core and distort the signals.
Incidentally the use of an isolating (!) transformer will not affect the 'frequency response' of the signal. If you hear differences it would have some other explanation.

 

[omishra]

PS: wondering aloud if this would do any good in an analog interconnect? My last pulse transformer can be sacrificed for this important scientific endeavour

For analog - you already get ground isolator transformers (used in Car Audio as ground loop hum and whine removal).
example - http://www.amazon.com/PAC-SNI-1-3-5-3-5-mm-Isolator/dp/B001EAQTRI
http://www.amazon.com/Raptor-GL15-G...56205741&sr=1-5&keywords=ground+loop+isolator
They are in-fact 1:1 line level transformers.

 

[jls001]

Thanks for the pointers.

I still feel that the transformers could at least behave as protective elements, without degrading/improving the sound

Of course to find out I need to sacrifice a pair of perfectly working analog ICs. Tough choice