My friend 'brother Y' was impressed with the sound of his S1 chips sitting in my hardwired DAC and rightfully asked a balanced topology, I couldn't agree more. Balanced topology requires pair match of everything, from DAC chip down to Jfet and opamp output stage down to single resistor & capacitor in the signal line. After long study at DIYAudio.com (as always, thank you guys..) I've found the way how to do balance DAC using I2S splitter, yup, there is one good man behind this, Pedja Rogic, kudo's to you sir!.
The digital data from CD transport is first entering the receiver CS8414, which gives I2S data out, this data then go into splitter which feed 2 TDA1541A, one chip received Left data, the other Right data. Now the splitter does something else, it split the audio data into + data and inverts the other half of data into -data. So one dac chip will output L+ and L- analog out, the other R+ and R-, clever isn't?
From there I will have 2 pairs of (now) my favorite discrete 2SK170BL's single ended current to voltage converter I/V as in my hardwired DAC. From the JFET's the the L+/L- and R+/R- signal are fed to balance - single ended converter op amp :(...but this is the easiest & speediest way. After op amp you can do tube out, just what I did, later on that... :)
Meanwhile, I did got busy making that double sided PCB which since the start of layout I've incorporated possibility for paralel and balance configuration. Not a world class digital PCB I admit, but it works wonderful. I got also many valuable advices from the NOS TDA1541 man himself : Thorsten Loesch, thank you T.., and this means the layout is evolving :).
Well, quite an introduction to this challenging DIY project, now (finally) some pictures, I've made my PCB's using my (DIY) mini CNC machine that I've made back in 2004, told you that I'm a hardcore DIY guy :)
CNC ing is very nice if you do only limited quantity, and when you experimenting (read: playing :)) but take bit of time of tracing & converting ECAD data to usable GCode for machining, but what we get is nice clean process without chemicals and of course nice copper area for power and digital ground plane...
The long awaited parts were finally complete, specially the hard to find old Philip chip for I2S splitter, the famous Toshiba 2SK170-V for the 4 Salas regs, and the most important, a pair of dual matched 2SK170-BL, quad match were simply not available anymore. Got 1 quad match in my MC phono pre though he he..hope to share this as well soon.
Here is how the DAC and splitter board looks like
Top of pcb, the splitter with 3 red cerafines and the left channel DAC, bottom pcb are for the CS8414 and the right channel TDA1541A, notice detachable risers for ground return and power supplies.
And does these things worked? You bet!..perfectly by the way :)... here are early measurement using TDA1541 non A and John B's single ended Jfet, tested here the L+ and L- signals, the usual 500Hz/0dB
Green is the L+ signal, red the L-, they are exactly, well..within the limitation of our beloved VisualAnalyzer of course :), have you got one installed yet?
You see, the whole balanced thing comes to to this, you add 2 exact but opposite phase signal, summing will get rid of all even harmonics (what????...said a nice tube guy over there...:;D..), what left are the odd harmonics...and in total lower distortion figure.....why oh why???...said another tube guy....:(...
That yellow curve is full of even harmonic and it IS being removed. This cleaned signal then enters the op amp to be made single ended again, want to add flavour? you can, I'll add tube out just to add a pinch of 2nd harmonic. My brother Y will have options of DAC outputs to choose, SE JFET, Balanced JFET, SE opamp out, or .....SE Direct Heated Triode tube out
stay tuned & have fun