Hi Eldam, Salas are obviously good, but if you need 6 or 7 for the compete Dac, it gets pretty hard to fit it all in. If I use LM317 I will do the following Using 3-pin regulators off-piste: part 4 (last one) and use TL431 and to92 LM317 as CCS. Between the pre reg and chip reg I will use a Pi filter and maybe ferrite bead ala Ts Addagio.
My jfets are missing in action, I should look for them, they are not graded yet.
My jfets are missing in action, I should look for them, they are not graded yet.
T's classic with update (2012):
"
maybe time for some updates.
If you need more current than the LM334 can produce use an LM317, it makes an okay CCS. If you can, use a discrete C4S, but the difference will be much less in reality than the few 100K of the 317 or 334 vs the GOhm of the C4S suggest.
Nowadays I bypass the upper resistor of the Voltage set in the 431, even in tiny "microshunts", but it is not an absolute requirement.
In those days Cap's where at 0805 for pretty much the limit. Nowadays, if you can manage the layout you can use 0612 Parts or 4 pcs 0603 in parallel.
So make the 10nF X7R 4pcs of 100nF 0603 in X7R, add 4pcs 10uF 0603 in Y5U dump the 15uF Os-Con for several 100uF of the cap you like best.
Depending on application I use nowadays either Sanyo Os-Con 470uF/6.3V, Elna Silmic II 220uF/16V or Nichicon/Sanyo Low-Z up to several 1,000uF...
Around the 317 I now use large ESR output cap's, usually 1,000uF or more, to kill the inductive tendency of the reg. Think of the Cap as a Zobel that flattens the impedance of the 317 and select capacitance and ESR accordingly... I learned this trick indirectly from Jocko's SPDIF input circuit...
In fact, you will still find variations of this regulator circuit (with a variety of actual CCS Circuits, from 317 to C4S and anything inbetween and a variety of shunts from discrete sub nV|/Hz to the very noisy 431 with or without current boost) all over AMR's gear. We have tried others, so far we keep using this if things really matter for performance. We sometimes tried to "shortcut", deleting either CCS for resistors or using more normal regs, they never seem to cut it as well.
Some of new microscopic low power, low noise regulators from AD, TI and others are starting to show promise, especially if you draw some extra current, they are sufficiently better than 317's to warrant using them in lower cost products where you want to beat the 317 or it's SMD cousins...
Ciao T
"
from:
DIYHiFi.org • View topic - How modern "High Performance" DAC's and ADC's are designed
thanks Audiolapdance
For what I know, a non feed-back shunt reg is even better and musical : the TL431 does not output on the load but feed Darlington like a TIP142 or 147 for negative; ok you can filter the resistor of the shunt with a 2 to 4 nano cap. The impedance will be very low and larger, output is the same, the energy came of a big cap before the shunt and the darlington : of course the shunt has its own filter cap also.
Rapid and musical, even if a feed-back could show better results... on the scope only, not to the ears !
Variation exist for the shunt like the Abraxalito one !
I also add a shematic in a thread somewhere on the PS or SS threads
For what I know, a non feed-back shunt reg is even better and musical : the TL431 does not output on the load but feed Darlington like a TIP142 or 147 for negative; ok you can filter the resistor of the shunt with a 2 to 4 nano cap. The impedance will be very low and larger, output is the same, the energy came of a big cap before the shunt and the darlington : of course the shunt has its own filter cap also.
Rapid and musical, even if a feed-back could show better results... on the scope only, not to the ears !
Variation exist for the shunt like the Abraxalito one !
I also add a shematic in a thread somewhere on the PS or SS threads
I agree with the design for the few I know (almost nothing) and I am enable to simulate it ! It could be keeped for the analog outpustage but for the TDA itself : if yes as said Audiolapdance, Alexiis, Mr T : is it because the noise is less than the noise floor of the tda-1541 ???? Because nowadays very good all in one regs exist.
http://www.diyaudio.com/forums/powe...ed-back-emitter-follower-reg.html#post3887530
Achtung bycyclette with the shunt : must be near to the load : complex with our 3 voltages... but ceratinly possible. There are also the Jung regs for tha analog stage.
Andrea knows a lot about regs, maid tests : the link is on the first page of this thread !
Last edited:
read Ts comments in post 22, they use LM317 and CCS then TL431 in many AMR products. I like these regs and have modded 3 cd palyers extensively and like the difference they make. Im sure there are bettter, but for very little money you could build a benchmark psu to use in comparisons while you try more interesting and exciting designs. I also plan to compare some salas shunts at some stage.
It sounds interresting but have a lookt at what ANdrea Mori did already : a great benchmark with powersupplies. I almost sure he already tried those and can answer to you for saving time and money. Peufeu advices could be welcomed also... (if you reading this Peufeu !)
Well have you a shematic of the T' PS : basicly a LM317 feeding a CCS near the shematic Audiolapdance has just given but with a Darlington ? Am I right ?
I surmise the shunt party made By Abraxalito to sound better but a micro mosfet to add. In all case, a simulation has to be made with Spice like soft to find the decoupling cap value // to the resitor plot of the TL431 shunt to adjust the outputvoltage. The big quality of a Darlington is it give you a lower and wider impedance, sound depend mostly of the quality of the filtering before (a Pi to a triple Pi) : mostly the big cap of 10 k Uf before and the 0.1 to 0.2 uf after. As it's a non-feedback, TL431 is less involved if fine decoupled (very low esr caps : 1000 to 2000 uf with a // MKT (MKP is a little big here) : test can be done with polymer; some people report good result with non-continued Sanyo Oscon SP (polymer, ancient technology that the newers has not anymore).
If i would have more skill and money I will test myself alsoo the Nazaar's with excelent caps around the LM317... and with some different 0.1 uf at the output like a MIT RTX (polystiren for cap speaker filters), why not a 2 mm pitch Wima MKTS : in bot case the darlington and the MKS2 near the DAC chip (to be benchmarked with a smd 0805 NPO as the Distinction provide the pcb traces...
Well have you a shematic of the T' PS : basicly a LM317 feeding a CCS near the shematic Audiolapdance has just given but with a Darlington ? Am I right ?
I surmise the shunt party made By Abraxalito to sound better but a micro mosfet to add. In all case, a simulation has to be made with Spice like soft to find the decoupling cap value // to the resitor plot of the TL431 shunt to adjust the outputvoltage. The big quality of a Darlington is it give you a lower and wider impedance, sound depend mostly of the quality of the filtering before (a Pi to a triple Pi) : mostly the big cap of 10 k Uf before and the 0.1 to 0.2 uf after. As it's a non-feedback, TL431 is less involved if fine decoupled (very low esr caps : 1000 to 2000 uf with a // MKT (MKP is a little big here) : test can be done with polymer; some people report good result with non-continued Sanyo Oscon SP (polymer, ancient technology that the newers has not anymore).
If i would have more skill and money I will test myself alsoo the Nazaar's with excelent caps around the LM317... and with some different 0.1 uf at the output like a MIT RTX (polystiren for cap speaker filters), why not a 2 mm pitch Wima MKTS : in bot case the darlington and the MKS2 near the DAC chip (to be benchmarked with a smd 0805 NPO as the Distinction provide the pcb traces...
I will probably try Ts then Salas, I'm curious if it makes any difference. T states that the self noise of the TDA1541 chip is greater than that of the TL431 shunt. I wonder if going to great lengths is of any great value, but I hope to try at least two.
On another subject, the PPS caps are crazy prices, dollar per mm they are like gold
Maybe we should pick a cap and all do a GB?
On another subject, the PPS caps are crazy prices, dollar per mm they are like gold
Maybe we should pick a cap and all do a GB?
from the grumpy place:
DIYHiFi.org • View topic - The TDA1541 PCB project at the Pub...
"glen wrote:
Or just use TPS7A4901 and TPS7A3001 regulators instead of fooling with ancient noisy gloried zener TL431's ?
Those new Chips look at best 3dB quieter, so they are not all that horroshow.
The point of using a Shunt regulator and a CCS is however a little different.
I the thread over at the pub I mentioned "think where the current will flow".
If you think that way you will understand why TL431 and CCS.
Ciao T"
Hi Eldam, the dem cap is the one across the two 6k8 resistors right? Which is the timing cap?
from the grumpy place:
DIYHiFi.org • View topic - The TDA1541 PCB project at the Pub...
"glen wrote:
Or just use TPS7A4901 and TPS7A3001 regulators instead of fooling with ancient noisy gloried zener TL431's ?
Those new Chips look at best 3dB quieter, so they are not all that horroshow.
The point of using a Shunt regulator and a CCS is however a little different.
I the thread over at the pub I mentioned "think where the current will flow".
If you think that way you will understand why TL431 and CCS.
Ciao T"
Hi AudioLapDance,
I'm not sure I understand your post, do you agree with my approach and no need to try fancy regs?
Yes my own vocabulary, sorry... :
-100 pf :John value (ECDESIGNS) ; 14 decoupling caps : 0.33 uF
- 470 pf : current value ; 0.1 or 0.2 uF :
here : 5 mm pitch is ok like the Wima FKP2 serie, smd silvered Mica, smd ceramic NPO.... everything not larger than smd 1206 case but 0805 will give you easier way to solder it (or the 5 mm pitch...)
About the distinction & the 14 decoupling caps I just tried what I bought 3 weeks ago at Mouser :
Take care : 1206 case size max ! = no PPS caps will feet (first size for 0.1 uf and above are minimum 1210 : too large for our pcb) but the acrylic cap from Cornel Dublier are perfect (1206 case) not too big.... but difficult (need a good little iron !) : 0805 size will give you more easy way for soldering
Manufacter ref :fca1206c334m-h2 e.g for 0.33 uF
Other solution is no decoupling cap like OSHIFI fellow does : look for it on diayaudio or ask him
Other solution but many report it's not neutral (???? I wil retry to learn myself) : 0.1 to 0.2 uF SMD ceramic NPO (=COG) in 0603 case : if too small for your eyes/skill, use 0805 case: but I sucees in an other project as a beginner in soldering smd with 0603 case....
The best transcient is the result of the lowest inductance : so the lower pin pitch, gap, case size : the better inductance ! Here we would choose ceramic class I NPO because few or none piezzo electric effect with the pcb (= no noise instead of class II ceramic : X7R e.g. - the best clas II)
For the main decoupling polymer power caps :
United Chemicon : APSA160ELL.... in 16 V for the -15 V feet perfectly in pitch like in width for the pcb
For the +/- 5 V main decoupling cap : I advise Manuf part : Nichicon PLE0J471MD01 470 uf/6.3 V... yes 470 uF) : pitch is a little too large but I advise it : my prefered polymer with the old Sanyo OSCON SP serie... (I don't tried it I have none left ! )
In each case try first 15 days of listening without the smd 0.1 uF ceramic NPO grade ; then try with it then choose and keep the last in case of no hearable changement or subjective better one).
Last edited:
At Mouser it's free here for 65 euros of Bill ?????
Is it really 200 USD or AUD with Mouser in Australia.... Too big & too far Island with so few people (just 23 millions ! and the kangorous....).
The problem with the GB : always some parts shipped by Mouser are missing and they were on stock during the buy ! So the shipments could cost you a lot or your are to wait 3 months than the people who grup for you wait the missing parts before sent to GB buyers...
Unluckily, I didn't find good tPS trafo on Mouser for the project... better is to go with C-Core because with all the CSS reg trafo need no to be so far from the pcb (around 10 cm max... and they radiate : so if IE : need shielding, torroids not advised because wide range (noise for HF = no good), C-Core is the good candidate, In each case a shielding with iron is advised !
Is it really 200 USD or AUD with Mouser in Australia.... Too big & too far Island with so few people (just 23 millions ! and the kangorous....).
The problem with the GB : always some parts shipped by Mouser are missing and they were on stock during the buy ! So the shipments could cost you a lot or your are to wait 3 months than the people who grup for you wait the missing parts before sent to GB buyers...
Unluckily, I didn't find good tPS trafo on Mouser for the project... better is to go with C-Core because with all the CSS reg trafo need no to be so far from the pcb (around 10 cm max... and they radiate : so if IE : need shielding, torroids not advised because wide range (noise for HF = no good), C-Core is the good candidate, In each case a shielding with iron is advised !
Last edited:
Im in NZ. Our flag is a lot like Australia's, but not the same, we have red stars and only four of them.
Mouser quote in the local currency, NZ$200.00 is the same as $126.00 Euro. Unfortunately we and Australia pay far too much for many things. There were recently court cases I think in Australia with computer manufacturers for this price gouging.
I am going to use potted EI transformers in plastic housing, I have them handy so will use them, and they are quiet.
Mouser quote in the local currency, NZ$200.00 is the same as $126.00 Euro. Unfortunately we and Australia pay far too much for many things. There were recently court cases I think in Australia with computer manufacturers for this price gouging.
I am going to use potted EI transformers in plastic housing, I have them handy so will use them, and they are quiet.
any ref for those EI ?
Hey NZ is worst than Australia... you have no kagourous but golem, dwarfs, hobbits, magicians with white beards everywhere (here our Young people have black beards but by fame or exactly by Oedipus complex... and at the same time shave all elswhere... not so goods for hobbits, wookies, bears... do all the sheeps in NZ are shaved but keep a beard? Here in France it's only the pigeons (= suckers in french). https://www.youtube.com/watch?v=B6p-Act9DhU.... The other day I talked with a Young in a family diner... he believed John Lenon singed "Free as a beard !"....I only answered it was a song from ZZ Top and he had not to believe everything ! .... You know the frenchs and the English speaking....
I think Audiolapdance says as T. it's better to go for shunts with tl431 as he said in the anygood TDA1541 kit thread. But the layout has to be good for the flowing of current... we hope it's good enough with the Distinction but we had no testimonies of T. with our result but hope we applied the best we can his kind and free advices in the best traditions of DIY (thanks T. if read us)
Hey NZ is worst than Australia... you have no kagourous but golem, dwarfs, hobbits, magicians with white beards everywhere (here our Young people have black beards but by fame or exactly by Oedipus complex... and at the same time shave all elswhere... not so goods for hobbits, wookies, bears... do all the sheeps in NZ are shaved but keep a beard? Here in France it's only the pigeons (= suckers in french). https://www.youtube.com/watch?v=B6p-Act9DhU.... The other day I talked with a Young in a family diner... he believed John Lenon singed "Free as a beard !"....I only answered it was a song from ZZ Top and he had not to believe everything ! .... You know the frenchs and the English speaking....
I think Audiolapdance says as T. it's better to go for shunts with tl431 as he said in the anygood TDA1541 kit thread. But the layout has to be good for the flowing of current... we hope it's good enough with the Distinction but we had no testimonies of T. with our result but hope we applied the best we can his kind and free advices in the best traditions of DIY (thanks T. if read us)
Last edited:
- Status
- Not open for further replies.