Spdif coax input circuitry

Hello, considering jitter, noise, distortion, sonic signature etc.. which is the "best sounding" spdif coax input diy typology of circuit:

1) simple 75 ohm resistor (NON isolated) :eek:
2) 1:1 pulse transformer plus resistor (isolated):rolleyes:
3) pulse transformer plus resistor plus 74HCU04 with resistor feedback (isolated):eek:
4) pulse transformer plus resistor plus fast comparator (AD8611 or AD8561) (isolated) :hbeat:
5) resistor plus 74HCU04 with resistor feedback (NON isolated)
6) resistor plus fast comparator (AD8611 or AD8561)(NON isolated)

all obviously including 10/100 nF capacitors

The coax cable is professional but 2 meters long
The source could be any common commercial digital CD or network players
Input will be CS8412/14 for sure.

Thanks
 
Except for possible hum due to ground loops, I doubt very much whether there are any audible differences, but from a technical point of view:

A transformer can eliminate hum due to ground loops. It also limits the bandwidth to some extent, so it will probably make jitter due to the noise of the receiver slightly worse. As the source can be anything, you can't rely on the source having an output transformer.

It seems unlikely that a plain old CMOS inverter has better jitter performance than a circuit designed as an AES3 / SPDIF receiver. An AD8611 might be better, as it is apparently made in a bipolar process and therefore probably has less 1/f noise.

All in all, probably either 2 or 4 will work best.
 
Regarding SPDIF jitter effects, it depends on the dac.

ESS Sabre dacs use a state machine to decode the SPDIF waveforms without regard to jitter variations in edge timing.

Other dacs may use a PLL in an effort to synchronize the dac with the incoming SPDIF clock. If edge timing is jittery, the PLL can attenuate it, but not ignore it.

So, if using a Sabre dac I would probably like transformer isolation best, but maybe not for another dac.
 
Disabled Account
Joined 2002
Hi, although I like galvanic insulation the most the waveform is generally best without transformer so option 1. In my current setup the source has a non-PE PSU (adapter) and the digital amplifier (FDA) also lacks a solid PE connection which makes at least one of them "floating". In this case an SPDIF transformer did not make things better. In many other cases results were better with transformer. I recall Squeezebox Touch being such a case. Nowadays I like to use insulated and non-insulated BNC SPDIF inputs with GND also being switched. Why not have both? One source likes galvanic insulation while another source doesn't. Having a few inputs makes comparing also easier ;)

Tips: Murata DA101C is a very nice performing SPDIF transformer. Avoid 74HCU04. Use a resistor at the primary side and at the seconday side of the transformer for better results.

Input will be CS8412/14 for sure.

You must know that these receivers make the discussion a bit blurred as they are long superseded by receivers with less jitter (or having jitter reducing techniques). With SPDIF situations "best sounding" is indeed "best measuring". If you want to have the most out of it you will have to use a better receiver chip combined with one of the better input circuits.
 
Last edited:
To get the best sounding input you'll need to pay attention to more than just the kind of circuitry at the input. You need to consider where the common-mode current from the S/PDIF input is going to flow. If you have any common ground impedance for that noise current with any audio grounds then that'll degrade your SQ. Isolation increases the CM impedance so I'd definitely use a pulse transformer but watch your layout.
 
Thanks for the replies.
After massive internet and DIYAUDIO navigation I think I will adopt Quanghao’s configuration, that uses a pulse trafo and a comparator. (also reported in this forum somewere)

de-r-inputs.jpg


I have simulated importing AD8611 model from ADI and other LTC comparators in LTSPICE with success. Attached. It works great!

1) I would decrease the input resistance in order to lower input noise (and increase capacitance accordingly, to 220nF). Is this right or should I have to stay with 10K+10K voltage split?

2) I would put a zobel on input…mmm I don’t know if it is correct and beneficial

3) BIG QUESTION: at the inputs of CS8412/4 should I put just the positive comparator's output to RXP and ground to RXN, or it is better to input to RXP and RXN the differential signal between positive and negative outputs of the comparator?

NB: I'm ready for LIQUID MUSIC and this is my first DAC construction
 

Attachments

  • SPDIF COMPARATOR INPUT(1711).asc
    2.7 KB · Views: 119
  • SPDIF input with AD8611 scheme.jpg
    SPDIF input with AD8611 scheme.jpg
    27.2 KB · Views: 1,040
  • SPDIF input with AD8611 signal.jpg
    SPDIF input with AD8611 signal.jpg
    33.5 KB · Views: 976
Last edited:
Having C2 rather defeats the purpose of the transformer. Sure you've still got isolation at DC but it's going to let through a lot more noise than the interwinding capacitance of a typical pulse transformer.

The zobel as shown on your schematic isn't such a good idea - at higher frequencies its just another termination resistor but you already have one in R2 so this is going to mis-match your cable.
 
Thanks, but ... this does't reply to my original question....;)

Another time LTspice helps: it looks like that Differential, Unbalanced and Capacitive inputs work fine, but not floating.

So I think I will test by ear at least Differential and Unbalanced, as I read here (
Digital transformer ) from Elso Kwak that removing caps sounds better! And Elso used direct coupling of both comparator outputs. So I think "differential" could be the best sounding interface.
 

Attachments

  • SPDIF input FLOATING.jpg
    SPDIF input FLOATING.jpg
    234.8 KB · Views: 581
  • SPDIF input MIXED.jpg
    SPDIF input MIXED.jpg
    234.7 KB · Views: 564
  • SPDIF input CAPACITIVE.jpg
    SPDIF input CAPACITIVE.jpg
    226.1 KB · Views: 566
  • SPDIF input DIFFERENTIAL.jpg
    SPDIF input DIFFERENTIAL.jpg
    223.6 KB · Views: 759
  • SPDIF input UNBALANCED.jpg
    SPDIF input UNBALANCED.jpg
    225.1 KB · Views: 721
  • SPDIF COMPARATOR INPUT Differential (1711).asc
    4 KB · Views: 84
Last edited:
Disabled Account
Joined 2002
Hi, you are right. I experimented quite a lot but never (AFAIK as it is long ago that I used CS8412) tried the differential approach so I can not add anything. Tried the comparator though but soon enough reverted to simplicity and changed to better chips when they were introduced. As said, using a better SPDIF receiver makes way more difference and it is more worthwhile even if one has to design a new PCB. WM8804 has comparator input by default, beware for some errors in the datasheet. Never tried software control but heard many a curse from guys that did. Depending on implementation of course but jitter is a multiple lower than CS8412/14 and thus can easily be heard. It may very well be that CS8412 even with the most optimal input circuitry can not beat other more recent receivers with just a transformer at the input :) To be precise: it can't. So improving matters with a classic receiver really is a waste of time compared to designing a new front end.

Let the oscillosope do the judging when experimenting with input circuits.
 
Last edited:
And Elso used direct coupling of both comparator outputs. So I think "differential" could be the best sounding interface.

Did you even test a simple buffer(74HC125) instead of the comparator?

WM8804 has comparator input by default, beware for some errors in the datasheet.

That's what I thought. Why not just a buffer after the transformer?

DIR9001 also have comparator inputs?
 
Last edited:
Did you even test a simple buffer(74HC125) instead of the comparator?

Hi, no I just build (attached) the differential version with AD8611, and it works great!
Being the current fastest single supply comparator in SOIC (at only 4 ns), I don't have any reason to downgrade to 74HC (9-17ns).

I'm working now with 74AS logics for PLL, reclock etc.. (with 1kohm pull-up at the output to match CMOS levels), which are dip and have few ns of delay.

Attached actual schematic (to be fair: not mine but a mix of two circuits found in google images and forum).

(FYI I built also the reset circuitry for CS8412).
 

Attachments

  • IMG_2618.jpg
    IMG_2618.jpg
    819.5 KB · Views: 418
  • Input SPDIF wave.jpg
    Input SPDIF wave.jpg
    173.1 KB · Views: 409
  • Input SPDIF.jpg
    Input SPDIF.jpg
    139.8 KB · Views: 415
  • LTSPICE.zip
    4.1 KB · Views: 60
Last edited:
Hi, no I just build (attached) the differential version with AD8611, and it works great!
Being the current fastest single supply comparator in SOIC (at only 4 ns), I don't have any reason to downgrade to 74HC (9-17ns).

I'm working now with 74AS logics for PLL, reclock etc.. (with 1kohm pull-up at the output to match CMOS levels), which are dip and have few ns of delay.

Attached actual schematic (to be fair: not mine but a mix of two circuits found in google images and forum).

(FYI I built also the reset circuitry for CS8412).
Hello did you finish this project in the end?
I am building a spdif input with cs8416 . It is software controlled via I2C and Arduino.
I am very curious...