IC1		= LM6172  (tried fet input types OPA2604, OPA2134, but prefer the LM one)
VR1, VR2, VR3	= LM431, TL431 shunt regulator
C3, C4, C6	= 15u/25V Sanyo OSCON
C5, C7		= 1st order filtering (I didn't install these),
		  something like 1n (@56kHz) or higher
		  MKP, polysterene, film/foil
C9		= 100u/10V Sanyo OSCON

I used carbon resistors throughout, don't like metal film ones.

After experimenting I settled on adding 100u/10V Sanyo OSCON across R14, R3, R13 as well.
This will give deeper and tighter bass response.  It filters the internal Vref(=2.5Vdc) of the shunt regulators.

I use the following shunt currents:
50mA for the +14.1Vdc	->	R2 = (Vin - 14.1Vdc)/50mA	->	Vin=18V, R2=78 ohm @ 195mW , use 2 parallel 150 ohm
50mA for the -14.1Vdc	->	R6 = R2
80mA for the +5Vdc	->	R1 = (Vin - 5Vdc)/80mA		->	Vin=8V, R1=37 ohm @ 240mW, use 2 parallel 68 ohm.
Have a look in the datasheet for the LM431 to understand how they work.

I created a prototype pcb and now a final pcb measuring ca. 5cm * 7cm


HOW DOES IT SOUND ???
One of my (old and 2nd hand) cd players is converted to a non oversampling player with 3 parallel TDA1543's resembling the TNT Convertus DIY DAC.  This sounds quite nice and is quite open and clear due to having no active output stage, just passive I/V conversion.  I use 3 in stead of 4 with different values for the bias and I/V resistor for even better results (but just 1Vrms in stead of 2Vrms output).  This player has an old style LCaudio-Clock low jitter clock as well, plus some other improvements (diodes, snubbers, psu, improved local decoupling etc.)

Using this new output stage in a similar but unmodified player, there is just no contest.
The sound is so clear, basses are deep and tight.  The circuit behaves like passive I/V (fet functions as a variable resistor here), but with the benefits of an active output (output drive power and incredible deep and tight bass.)

Coming up:
Same output circuit for TDA1545.  This will resemble this schematic a lot.
Can be used in the Philips CD713/CD723 for example, to make these killer machines.  I know, I owned the whole series and are going to modify one these weeks with this output stage.
Let me tell you, with a low jiter clock and this output stage, it will put cd players costing $1000 and up to shame...

Coming up later:
A Philips CD713/CD723 converted to non oversampling, reclocking with low jitter clock and again this output circuit again...

Happy DIY.

Regards,
Rudolf Broertjes