non feed-back emitter follower reg

Status
Not open for further replies.
Hi,

Here is a shematic found in a french forum at Elektor. this design is old, more than 7 years I believe but like the simplicity : few parts. Q1 is a darlington TIP142 (for positiv reg). And the sound is known to be excellent.

What do we have here: 2 solutions with C4. If C4 is small like 0.2 uF, filtering and caps C0 & C1 have to be excellent and greater than 4700 uF (10000 uF could be ok with C1 very low esr).

If C4 is bigger, which is the better way for the SQ, e.g 10 000 uF : "audiophile quality" is needed + MKP. Here transitor Q1 manage only a constant voltage and courant for C4; we have a very linear PS with impedance < 0.3 to 0.4 ohms 🙂. C4 has to be very low ESR; e.g. polymer if the load is digital board or mix of "gound sounding caps" (I know caps has no sound of their own!) for analog stages...

One big default Emitter Follower must be near the load to be efficient : how do you manage this on a small core board for feeding a DAC chip and minimize the gnd loop for the local feedback of the T431 shunt ?

The minimum I success for such a pcb is 3cm x 6 cm without the Pi filter & if C4 is not the second scenario with huge cap! And the local feedback with TL431 has to be the smaller in size but the cap C2 is too big for a core board like the one for dac chip : that's why regulated ldo small noise are known to be better on a core board to have smaller ground loop on the contigous gnd layer?
But is it efficient to use the E.F. above just behind those smd ldo regs on the core board in DIY. Or whatever, Salas e.g.

What could be an efficient power supply for a DAC chip please (here I think to the TDA1541 because question was highlighted elsewhere in the digital forum !) regarding the SQ ? Old technics (linear PS or shunt like these) can be efficient for digital better than the small smd regs from Micrel or TI ?

Thank you
 

Attachments

  • EF.jpg
    EF.jpg
    659.7 KB · Views: 452
Hummmm : 🙄 !

Have you maybe an idea about :

Best C4 ? big or small value according to you ?
Any anti noise ldo reg near a digital load like a DAC? Could it be animprovement?

RC for reducing noise of TL431, what value would you use ? // to C2 ?

Please :

Will you use it for digital e.G. with better Pi filtering (diferential mode before C1 & C0 (Double Pi ?).

On the garbadge with the newest shema. At the end I 'm looking for someting compact and soundy for a DAC chip and not too much expensive. Question no accurate enough ?

thank you if any ideas
Eldam
 
I am not quite sure about the goal of your design, but I assume you are searching for a clean output voltage.
For my opinion there is no point in bulky input capacitors as long as your input voltage is pre-regulated.
Like zener-diodes the TL431 is a noise source, and this is not easy filtered by parallel capacitors. A RC-filter between LM431 and the npn-base might help.
 
Thank you for those leads.

This is an "old" (but maybe good enough I don't know) design I found used for supply output buffers in CD players. My goal was to find something compact to populate in a final Powersupply stage for DAC chip.

My understanding is a very low impedance is needed and such a design give with around 0.3 ohms at the output. But maybe it is for analog stages and not for digitals ?

I know there are design with a lot of transistor à la Salas but I'm looking for something compact to populate, I knew e.g T. Loesch used TL431 in him old tweak near the TDA141 dac chip in cdplayer... this is the start of my need.

LM3xx regs seem to be good only as pre reg, bulk reg à la Linear technology seem not to work fine in a powersupply stage for digital.

My question please will be what is the best type of powersupply, easy to populate (compact) but fine for dac chip. Does the little smd low noise low drop à la Micrel, TI are good enough for dac chip ? Advantages : groundloops are small,but is it enough for a gound sound. Is the first need is not to suppress the noise ? A two stages powersupply : regulator and near the dac chip such smd reg ?

Sorry for my lake of thechnical background...
 
In digital, the regulator seems to be there, just to ensure the delicate circuits do not see over-voltage.

The decoupling spread generously around the digital board, controls the supply impedance.

I don't think any regulator can be as fast as the decoupling capacitors in meeting the demands that high speed digital put on the supply.
 
Thank you Ken for this precise hint.

So the way to go for PS a DAC chip is the less noisy design from the beginning (just after the diode rectifier bridge), very good local decoupling near the dac chip (stable and fast, big enough = ceramic, film, small inductance (SMALL case size as noticed elswhere in Digital Line thread).

Is the precision of voltage swing has an influence to the noise ?

What would you advice nowadays for a dac chip dedicated powersupply design ?

Super reg à la walter jung, Salas bulk transistor design, other integrated ones like the very anti noise integrated chip à la TI AKR47VH (high speed, ultra low noise). For a noob like me I can't just understand the better way for the sound. My final understanding is than for a DAC chip :

- digital part of it : the less noisy, the better : integrated device welcomed.
- analog part of it (output analog stage with integrated oap or more rare output current for external i/V) : design with integrated reg give less Sound quamity results ?

Is it a too big simplification coming from my missunderstanding and lake of technical background ? I'm looking for something to go with most of DAC chip...
 
Last edited:
Thank you Ken for this precise hint.

So the way to go for PS a DAC chip is the less noisy design from the beginning (just after the diode rectifier bridge), very good local decoupling near the dac chip (stable and fast, big enough = ceramic, film, small inductance (SMALL case size as noticed elswhere in Digital Line thread).

Yes, and also for the critical clock circuitry. Not all DAC chips utilize the same clock signal to control the data conversion instant. For example, the PCM1704 uses the bit-clock signal, while the AD1865 uses the word-clock signal, and the sigma-delta based converters I know use the system-clock signal. The critical clock signal in different in each case.

Is the precision of voltage swing has an influence to the noise ?

I'll assume you mean, the precise voltage level of a binary low or high. So long as the two voltage levels remain sufficiently clear of the invalid logic state zone that exists between them, then their precision doesn't matter for correctly determining data. However, power supply noise will modulate the timing of the transition instants between between binary logic states, which will manifest as jitter if the timing of the data conversion instant is therefore modulated.

What would you advice nowadays for a dac chip dedicated powersupply design ?

Super reg à la walter jung, Salas bulk transistor design, other integrated ones like the very anti noise integrated chip à la TI AKR47VH (high speed, ultra low noise). For a noob like me I can't just understand the better way for the sound...

You will find many opinions on this. I can only tell you what I currently use, which is a function not only of performance but also of ease of implementation. I use integrated regulators in a single primary (using a typical 3-terminal part) feeding multiple low-noise, low-dropout voltage integrated regulators. As you correctly surmise, the truly effective means of controlling noise on fast switching circuitry is via low inductance capacitive bypass filtering. The active regulators mostly serve to reject line noise and establish stable D.C. voltage levels.
 
Status
Not open for further replies.