Building the ultimate NOS DAC using TDA1541A

Hi luxury54,

if it's possible for you to attach a basic schematic of this concept for testing purposes I'm sure many of us will appreciate it, ... ONLY if this is possible

No problem, I attached concept schematics.

Schematics shows a basic rectifier, C1 provides some mains RF attenuation.

The actual common-mode capacitance multiplier consists of two ripple reduction circuits / capacitance multipliers, one in the plus and one in the minus rail. It provides over 500,000 times ripple and interference attenuation. So with typical 100mVpp input ripple, output ripple would be approx. 200nVpp.

Separate transformers or windings are required for multiple voltages as there is no longer a common GND path starting at the smoothing cap.

Multiple voltages are achieved by connecting +V or -V terminals of multiple regulators. Each power supply could be viewed as a "floating" battery power supply.

I currently use discrete voltage regulators based on bipolar transistors, or the more refined FET regulator shown in the schematic.

The FET regulator consists of a cascoded CCS based on 2 JFETs. Series regulation is done using a MOSFET series transistor. L1 serves as voltage reference, the current through L1 depends on JFET CCS properties. JFET T6 compares reference voltage with attenuated output voltage and controls MOSFET T3 gate.

I use 4 of these regulators in the SD-player, one for the SD-transport and 3 more for DAC power supplies.
 

Attachments

  • comcap1.jpg
    comcap1.jpg
    94 KB · Views: 1,874
Hi fffo,

does this concept schematics work without exploding/ problem? If so, I will give a try.

Yes, well except if you reverse polarity of C2.

Tip, when experimenting with circuits, connect them through an energy monitor that displays circuit power consumption. Add a mains (safety) switch between monitor and circuit. If power consumption exceeds expected levels, mains can be switched-off instantly before problems occur or circuits get damaged.

It usually takes some time for a reverse connected electrolytic cap to heat up and finally explode or vent, or a power transistor to heat up and short out. During this time the energy monitor would display unusual high power consumption.

Increased transformer hum (core saturation) can also be a warning sign that something is wrong.
 
Hi ryanj,

is the role of T4 and T5, along with the resistors to attenuate the spike in positive flow after T3 starts conducting?

T4, T5, R3, and R4 form a cascoded constant current source. Cascoded constant current sources offer much more stable output current and higher impedance.

This constant current runs through T6 and voltage reference L1 (1.6V). T6 (JFET) can be viewed as a variable (voltage controlled) resistor. When resistance varies, and current is kept constant, the voltage across T6 will vary. This varying voltage is then used to control MOSFET T3.

The refinement over a bipolar transistor based regulator is that gate currents are virtually zero, so they don't change while the circuit is keeping the output voltage constant. This means that the full cascoded current source output current has to flow through L1, all the time, regardless of correction voltage on the gate of series regulator T3. Also, voltage divider R5 / R6 can't dump current into L1 because gate current of T6 is virtually zero.

T4, T5, and T6 are low noise FETs, this helps to achieve low noise DC output voltage.

With the earlier version based on bipolar transistors, the LED reference current would vary slightly caused by varying transistor base currents. The output voltage divider would also dump extra current into L1 through b-e junction of the connected transistor. This leads to reduced output voltage stability with varying input voltages and / or load currents.
 
Hi Everyone

Just been doing some comparison between over-sampling and NOS; and i came across this at lessloss audio, which john refered me to coincidently.

A minimalist design such as the use of no upsampling or oversampling yields the best sound. The industry needs to return to the roots of simple design. Some companies are doing this and they achieve better results that way.


LessLoss made extensive test devices which allowed the comparison of several filtering techniques including the now ever popular non-filtering solution. The method chosen for production is to implement the following oversampling: 8x oversampling at 44.1 and 48 kHz = 352.8 and 384 kHz; 4x oversampling at 88.2 and 96 kHz = 352.8 and 384 kHz; 4x oversampling at 176.4 and 192 kHz = 705.6 and 768 kHz.


Imaging is better, soundstage is more defined, there is less distortion of subtleties in the music signal, and you hear a reproduction which is unrivalled by all non-oversampling solutions.


In my own opinion, with NOS sound is somehow more natural, and sound stage is deeper, higher, and wider. Even without IS2 attenuation.

Yes, i admit that NOS without IS2 attenuation, highs are somewhat dirty and inaccurate, but still i think id prefer it than oversampling, which produces a sound that is smeared or smoothed out in some way.

I just cant see how Lessloss can think that sound stage and imaging is more accurate with oversampling?

Maybe they know something we dont. (John excluded)

Just wanted to give me 2 cents, so to speak.

What do you think John?


Ryan
 
I listen to John's MK7 DAC and had a lot of listening experience with OS Dac's. I suppose Lossless haven't heard John's MK7 DAC approach with Double Crown chip ;-)

My own listening experience is that OS dac's can sound impressive for about 15 minutes, then i get tired and bored, Up-sampling creates little phase-shifts (faults) which, i suppose, makes the brain nervous. Maybe not all human ears are sensitive for phase-shift...
 
Hi roger57

Could you please let us know which manufacturer is on the tunable coils in the schematic? And to confirm, are they 100uH?
I've checked on Coilcraft's website, I cannot find a 100uH in this style.

These are custom made, multi-segment air chokes (non tunable). They consist of two dual half core formers glued together (intended for use with ferrite pot cores). Turns are equally divided over these 2 formers (4 x 50 windings). These chokes are then screened using a brass IF transformer screen.

The photograph shows an early prototype.
 

Attachments

  • airchokes.jpg
    airchokes.jpg
    56.6 KB · Views: 2,215
Hi Brubeck,

Can you tell us a little more of your listening impression of Johns MK7 DAC compared to other digital or analogue sources that you have heard?

Thanks,

Peter

What I can say more about this DAC that it just sounds ‘right’ in all aspects. You can listen –even with a very, very critical ear- to this DAC without detecting any flaw or fault in the jitter, noise en phase area. That makes the sound truly natural, undisturbed and most close to the master recording. No other DAC i heard comes even close to this.

The way of implementation (amp) and the type/year of TDA1541 can give little variations in the perceived sound character.

Beside of phase issues, a major issue with DAC design is the maximum bitdepth they can reach on the analog output. Even with MK7 it will be hard to get the very last LSB’s as an purified analog differentiation on the output. Here we talking in the microvolts area. I am sure that The MK7 is technically the closest possible approach to get the most of the bits on the output.

My experience is that this micro information of the LSB’s is the next step to make a DAC a real musical performer. There are several ways to achieve this after the DAC. Using an very, very clean AMP like John’s Circlotron to keep the bits alive. If that isn’t enough it is possible to create some white noise (signal correlated), which simulates the micro information of the missing bits. I am are experimenting with graphite from a pencil or a carbon composite resistor after the DAC. Graphite gives remarkable results on this.
 
I'd like to pop in my thoughts. Graphite resistors exhibit different characteristics, but none that I could see would be detrimental - other than size. Carbon resistors are typically large, used in higher power rating applications such as speaker crossovers because of their temperature vs power curve. At 500 ohms, the only option may be to make them. I haven't been able to locate any manufacturers in the USA (yet) Maybe someone knows?
As for sound quality, they should be great, with no inductance to speak of, and none of the harshness of metal types.
 
Last edited: