Alw,Jan didden, Guido please help with TL431

[kimschips]

Hello all.

I am about to decide how to connect a Tl431 based psu to my new dac project. (3.3V for the digital section)
Cs8416->AD1896-> 2 PCM1798 dacs

First connection

http://peufeu.free.fr/audio/extremi...mplement_2.html

Here we got a choke to separate each TL431 based psu for the digital section in the dac and tL431 is // coupled with small ceramic capacitors+10uF Oscon.

Do you think TL431 is stable when it´s directly coupled with such small values?

Or is it better to do it this way:

LM337->CCS->TL431//22uF Oscon->47uH inductor->ceramic 10nF +100nF?

Or:
LM337->CCS->TL431//22uF Oscon->47uH inductor-> 22uF BG-NX-HQ//100NF Rifa phe426??

Any suggestions for a small local psu and decoupling for the Dacs VD and VA supply?

Best regards
Thanks for all

Kim

 

[Guido Tent]

Hello all.

I am about to decide how to connect a Tl431 based psu to my new dac project. (3.3V for the digital section)
Cs8416->AD1896-> 2 PCM1798 dacs

First connection

http://peufeu.free.fr/audio/extremi...mplement_2.html

Here we got a choke to separate each TL431 based psu for the digital section in the dac and tL431 is // coupled with small ceramic capacitors+10uF Oscon.

Do you think TL431 is stable when it´s directly coupled with such small values?

Or is it better to do it this way:

LM337->CCS->TL431//22uF Oscon->47uH inductor->ceramic 10nF +100nF?

Or:
LM337->CCS->TL431//22uF Oscon->47uH inductor-> 22uF BG-NX-HQ//100NF Rifa phe426??

Any suggestions for a small local psu and decoupling for the Dacs VD and VA supply?

Best regards
Thanks for all

Kim

Hi Kim

The link does not work (today). So far I used the 431 with a variety of caps between 100nF and 47uF.

To separate the regs, us ferrites. Your last config looks OK. I like wima MKS2 (100nF). I would not paralle the BG's with anything else.

Ans last but not least, try and listen what you like.

bes

PS: The equivalent noise is 100nV/SqrrtHz so a 5V ouitput would give 200nV/SqrrtHz, not super quiet.....

 

[peranders]

Notice that Texas TL431 has noise at 48 nV/Hz, best brand so far.

 

[Guido Tent]

Notice that Texas TL431 has noise at 48 nV/Hz, best brand so far.

Ah, didn't compare the all, so thanks for the hint

Still noisy compared to other references, but these need a whole circuit around them, whilst a 431 is an easy and reliable soulution.

Not reccomend for best clock performance........

 

[jcx]

http://www.cirrus.com/en/pubs/appNote/an04.pdf

fig 12 shows tl part as adc ref, table puts its performance in perspective with other good references

can find good bypass decoupling hints in other ref circuits shown

 

[peranders]

Not reccomend for best clock performance........

What do you recommend then?

I checked the document mentioned above and it's interesting to see how noise in the reference influences the S/N, see last page.

 

[kimschips]

Hello all

Thanks for your input

The link is here, should work now, please comment the TL431 setup

http://peufeu.free.fr/audio/extremist_dac/implement_2.html

The setup is thought like this, because LM337+CCS or any other preregs shall be on a separate board, so people who want´s the new Dac board( when it´s finnish) can tweak and change supply after their wishes goals.


Please make some suggestions if you had a other solution for decoupling and local psu by the digital chips

Best regards
Kim

 

[Guido Tent]

What do you recommend then?

lower noise

 

[Guido Tent]

Hello all

Thanks for your input

The link is here, should work now, please comment the TL431 setup

http://peufeu.free.fr/audio/extremist_dac/implement_2.html

Please make some suggestions if you had a other solution for decoupling and local psu by the digital chips

Best regards
Kim

Hi Kim

This link works, indeed

Mr Peufue already refers to my document, so no additional remarks made.....

best

 

[peranders]

How low noise do you recommend achieved by what?

 

[Guido Tent]

lower noise

Ah, in rereading the post I assume you ask for a practical suggestion.

Most LEDs biased by a current source are pretty clean. They aren't the most precise and temperature stable, but that is not always the requirement either....

best

 

[Guido Tent]

How low noise do you recommend achieved by what?

Hi,

For my clock circuits (XO2/3) I have typically 3nV/SqrrtHz at the supply line (all trimmed for lowest noise). Care has been taken to achieve 1/f corner as low as possible.

In our Grimmaudio AD1 converter we also have taken thermal measures as to reduce the subLF noise on the supply. We can measure 0.1Hz - 10Hz in the nV range......

best

 

[peranders]

Interesting to hear about your good oscillator designs but the question was about the reference voltage for a DAC project and you don't recommend 431 but instead LED's as reference.

 

[Guido Tent]

Interesting to hear about your good oscillator designs but the question was about the reference voltage for a DAC project and you don't recommend 431 but instead LED's as reference.

Hi Per Anders,

I haven't said a word about oscillators, but about their power supplies, upon your request.

You asked about low noise references. I suggested some, independing of application. If you like, you can apply them to a DAC project.

At this end your reply is perceived as being unfriendly.

 

[peranders]

"For my clock circuits (XO2/3)".... oscillators with closely integrated power supplies I'll guess? ... not ment to be perceived as unfriendly but isn't the application important what type of reference voltage source you should choose, especially when you have asumed RF noise around you?

 

[Guido Tent]

... not ment to be perceived as unfriendly but isn't the application important what type of reference voltage source you should choose, especially when you have asumed RF noise around you?

Hello Per Anders

OK, clear, thanks.

Intrinsically, the noise of the reference should meet a certain requirement. Next indeed is the application. Then things like decoupling and board layout start playing a role.

In addition, when buffering (if needed, but please do without whenever you can) the opamp may be affected by some RF noise around.

best

 

[kimschips]

Hello Peranders and Giudo don´t get in fight I just need some advice->my first dac project

Guido you did not tell me if LM337->CCS->Inductor->TL431//22uF Oscon//100nF//10nF ceramic was stable for the digital psu(s)?

What do you think for psu to the Dacship (3.3VD and 5V VA):
A sulzer reg with 220uF BG-FK->47uH inductor//22uF BG-NX-HIQ//0.1uF BG-NX-HQ??

or keep the TL431//220uF BG-NXHQ->47uH->22uF BG-NXQ//0.1uF

Thanks for all
Kim

 

[kimschips]

Just an idea

What about using 2 red leds in serie (3.3V reference) instead off the TL431 by the digital chip?

How much current is needed to sent throught the led to optain low noise?

Has it been done/listening to before?

Kim

 

[ThorstenL]

Konnichiwa,

Do you think TL431 is stable when it´s directly coupled with such small values?

First, stability boundaries can be checked in the various manufacturers datasheets. Had you done this trivial task you would find that depending on current and voltage the various 431's do not appreciate a load capacitance in the 5nF - 3uF. This suggests, in order to be safe either very small load capacitors like 1nF (which BTW is too small) or at least around 4.7uF.

Secondly, I recommend the TL431 and compatible devices for applications that require a good and predictable dynamic performance (in other words audio or "close to audio" circuits). So they are not a "universal" solution. Feeding them from a CCS is recommended.

Sayonara

 

[Guido Tent]

Just an idea

What about using 2 red leds in serie (3.3V reference) instead off the TL431 by the digital chip?

How much current is needed to sent throught the led to optain low noise?

Has it been done/listening to before?

Kim

Hi

If it works depends on the current requirements of the circuit to be fed. The LEDs dynamic impedance is not that low.

I've used 3 red LEDs as a shunt for PCM63 supply. It works, but can be bettered by lower impedances.

best