The LPC4320 is also a 32-bit processor but not quite as fast as the XMOS CPU. When you get down to it, initializing and managing a USB data stream for digital audio needs less computing power than that needed for a USB mouse and I don't see many USB mice with 32-bit processors inside.
Tam, thank You very much for this link, I have contacted Member, hope he can help further
Thats is what I call high end- to think outside the box 🙂
Yes, programing is beyound my abilities for sure. A neigboard of mine(almost) have much more knowlege in digital, he designed Less Loss 2004mk2 DAC, i may ask local help from him as well.
As for my system, I would like to play this DAC in both of my systems. The first one is DIY speaker two line arrays (first consist of 5psc accuton c222n bass and MTM 2x c173n midranges with made berylium tweeters, the second is 8x23w scan speaks subs playing up to 45Hz) amplifiers is DIY pass labs xa 200.5 (actualy more like xa250.5) and Hypex ucd700 for scan speaks. Everywhere is passive xovers, even in subwoofer section(custom made 26kg toroidal multi inductor)
Another sytem is comercial- Foca Utopia Grande EM speakers(with some modification to midrange xover section) with array of 6 electromagnetic subwoofers(also Focal) Amplifiers is DIY Pass labs XA200.5 for speakers and 2x Accuphase p7100 + active analog low pass XO for bass units.
with help of friend I have tried in my system(and heard some other systems) active xovers in 2008 and digital xover in 2011, but did not liked (neither sound, neither idea of full house of amplifiers and DAC) -I am believer of high quality passive xovers- they maybe less direct, less detailed and less tight in bass , but when it comes to music its just more realistic. Resolution is not everything in music. For this reason I believe like( and picked for this project) pcm63dac not the es9018 or similar new chip. Most of my record is 16-18bit, and only about 200 albums are are 24/192. Most of those albums I coverted to 24bit /44khz and play trought my wm8804/sm5842aps/4x pcm63p-y DAC. Not sure how much real bits is, but belive 20bit. those records sound much better than same 24/192 on pcm1794 dac . so real 20bits is OK for me(18bit maybe too little)
Hope this helps.
I'm also a believer in high quality passive crossovers, just line level (1st choice) or biamplified at speaker level. I agree also with your choice of PCM63 over ES9018. Going for a cost-no-object system I'd design a PLLXO into the DAC and use as many amps as I had drive units as after the DAC, the amps are the next weakest link.
Sorry for the weak question : in all active amp systems : is there always a big serie passive capacitor to protect the drivers ? Is not here an half active-passive design : I assume the quality of the cap is important ? Teflon...
Does transformer avoid the cap in high efficienty design (because saturation of transformers is not possible with classic low efficienty drivers).
2 cents, but one thing at my level which surprise me with modest project I maid is: change the caps and the PS : same DAC with same layout can sound as diiferent as a change of dac chip !!!!
What is the trade off between the low noise and the Powersupply between passive components and active device in it ? Ist is always good to use as last ps stage near the chip a smt ldo with always low case smt caps for better inductance e.g. ? My point of view is the scope can avoid more musical layout at the price of a noiser design but not so important in regards to the musical result sometimes.... Am I wrong or in the urban myth here ?
Does transformer avoid the cap in high efficienty design (because saturation of transformers is not possible with classic low efficienty drivers).
2 cents, but one thing at my level which surprise me with modest project I maid is: change the caps and the PS : same DAC with same layout can sound as diiferent as a change of dac chip !!!!
What is the trade off between the low noise and the Powersupply between passive components and active device in it ? Ist is always good to use as last ps stage near the chip a smt ldo with always low case smt caps for better inductance e.g. ? My point of view is the scope can avoid more musical layout at the price of a noiser design but not so important in regards to the musical result sometimes.... Am I wrong or in the urban myth here ?
Yes, if you have enough current to produce the desired output voltage with the chosen IV resistor. It's all in finding the right balance. Multibit DAC chips don't like big IV resistors. Do some research to find the largest IV resistor the PCM63 will tolerate. Use Ohm's Law to compute the output voltage from a single DAC using that IV. The result will be too low to drive an amp directly. To get the voltage up you can add active amplification, more DAC chips, a step-up transformer, or a combination of the above. But, to improve sound quality, you must reduce the resistor value, which counteracts what you did in the previous step.
Use fanout clock drivers. They are buffers with one input and multiple outputs designed specifically for clock distribution with low skew and jitter.
There are multibit DAC chips which are fine with large values of I/V resistor. TDA1387 is one such - at its maximum recommended supply (5.5V) it has an output compliance range from 0 - 4V. It therefore could easily be used to drive a poweramp directly if so desired - I'd not recommend this though as poweramps don't perform at their best when driven with fast rise-time step functions.
This looks like a good passive I/V - it´s not just a resistor, it´s an I/V network:
http://members.chello.nl/~m.heijligers/DAChtml/dig_r2b.pdf
Any opinions?
http://members.chello.nl/~m.heijligers/DAChtml/dig_r2b.pdf
Any opinions?
Hi, dear all, thanks again very much for your inputs, They are all very valuable.
So now its clear direction in the output stage- just a resistor will not provide enought current to convert to 6V balanced output (with reasonable qty of dac chips- unfortunately i am sure i can not gahter genue 50-60 psc pcm63p-y chips in near future ) Alexandre alternative
So question - if i use JUST transformer output (1:20 or other ratio) how many chips I need to get 6V balanced output? if 16 chips ( 4in paralel) will be enought?
if not, is there any REAL benefit in DAC design when 2 vs 3 vs 4 paraleled chips? I believe if active gain stage is used a single (or maximum two chips paraleled ,like in wadia 9 or esoterid d-30 is needed.? ) Somehow I thought that 16psc of pcm63 can drive direct powe amp.
so if none of above mentioned implementation (inluding Alexandre) will be achieved with 16psc -cm63, what exotic/extraordinaty active output can be done? some time ago I played around krell cd player- disconected its signal path from pcm1702u-k and used another output from 4xpcm63p-k - used active stage of Krell ( about 200psc transitors in cd player output stage!!) and really liked result- very dynamyc sound, hughe soundstage. Very oposite of most tube stages I have heard /tried. I have never heard any tube stage which I really liked. OPA stages can sound good in medium project, but my interest in discrete transitor stgage. any exotic opinions?
As for filtering- which filtering is more nessesary - digital or analog? Any opinions on passive filtering?- I had Sonic Frontiers DAC 10years ago and I really liked its passive filtering(but absolutely did not like its tube output stage , which had 9V RMS but in reality can't drive any 26db gain amp with 20-30K input impedance even for half swing)
So now its clear direction in the output stage- just a resistor will not provide enought current to convert to 6V balanced output (with reasonable qty of dac chips- unfortunately i am sure i can not gahter genue 50-60 psc pcm63p-y chips in near future ) Alexandre alternative
is interesting, but we need to know how many chips are needed to reach 6V output ( to drive directly power amp)? Even if I am not digital expert , but really doubt that passive parts can amplify signal.
So question - if i use JUST transformer output (1:20 or other ratio) how many chips I need to get 6V balanced output? if 16 chips ( 4in paralel) will be enought?
if not, is there any REAL benefit in DAC design when 2 vs 3 vs 4 paraleled chips? I believe if active gain stage is used a single (or maximum two chips paraleled ,like in wadia 9 or esoterid d-30 is needed.? ) Somehow I thought that 16psc of pcm63 can drive direct powe amp.
so if none of above mentioned implementation (inluding Alexandre) will be achieved with 16psc -cm63, what exotic/extraordinaty active output can be done? some time ago I played around krell cd player- disconected its signal path from pcm1702u-k and used another output from 4xpcm63p-k - used active stage of Krell ( about 200psc transitors in cd player output stage!!) and really liked result- very dynamyc sound, hughe soundstage. Very oposite of most tube stages I have heard /tried. I have never heard any tube stage which I really liked. OPA stages can sound good in medium project, but my interest in discrete transitor stgage. any exotic opinions?
As for filtering- which filtering is more nessesary - digital or analog? Any opinions on passive filtering?- I had Sonic Frontiers DAC 10years ago and I really liked its passive filtering(but absolutely did not like its tube output stage , which had 9V RMS but in reality can't drive any 26db gain amp with 20-30K input impedance even for half swing)
Last edited:
Hi
Yes, the one I posted will definitely need a gain stage.
I wouldn´t start too complicated (many parallel dacs is complicated imo). Start simple and go from there! Though that might not be as exotic as you´re looking for🙂
I have been playing with tda154x chips for years and I´m sure I didn´t reach their full potential yet. That´s the reason my AD1862 chips are still in the drawer. No idea if I will be surprised, or disappointed once I build something with the pricy chips.
Yes, the one I posted will definitely need a gain stage.
I wouldn´t start too complicated (many parallel dacs is complicated imo). Start simple and go from there! Though that might not be as exotic as you´re looking for🙂
I have been playing with tda154x chips for years and I´m sure I didn´t reach their full potential yet. That´s the reason my AD1862 chips are still in the drawer. No idea if I will be surprised, or disappointed once I build something with the pricy chips.
I gather that you are NOT looking for budget solutions like the tda chips (though they can be as good as any other)...
So, for a dac that can drive power amps directly, and has a neat volume control integrated in it, I would look at the pass D1 schematic (at least as a starting point). Did you look at that?
So, for a dac that can drive power amps directly, and has a neat volume control integrated in it, I would look at the pass D1 schematic (at least as a starting point). Did you look at that?
Somewhere in between, I guess, is a good answer? The best of both worlds... A bit of digital filtering, a bit of analog. Maybe switchable between non-oversampling and 2x oversampling? I especially like 2x oversampling, which I do in my source PC, because it doesn´t have the non-oversampling roll off (you will be only 1dB down or so at 20KHz) and allows for less steep analog post filter. Minimizes two problems at once. Without going to extremes like 8x oversampling.
Last edited:
Passive filtering is definitely a good idea, however Abraxalito has made me rethink my strategy with regard to this. His approach with inductors might have a better sonical result than the more common cap to ground. I have to experiment more and listen before I can say anything about it.
R-L-C-L-R could be a good passive I/V network.
R-L-C-L-R could be a good passive I/V network.
Actualy one of my dac's is kind of this dac (at the begining it was pass D1 copy with cs8412>SM5842APT>>4x pcm63> and more modern J fet output stage instead of mosfet) but upgraded to more modern chip (wm 8804) and better parts - wima MKP, naked vishay foil resistors, Mundorf Supreme oil output caps and black gates. Its one of my favourites dac, which I listen most(and with recently purchuase Joro xmos directly to sm5842 or to pcm 63 I believeI will like it even more)
But This time I want to build something much more exotic, more complex (very high quality power suplies, maybe computer based oversampling) and at the same time more simple(passive if any filtering in analog stage ) As ,I told earlier, my primary goal - not neutral dac - very hughe soundstage with weighty and dynamic sound. Rolled off and slow ( like many tube dac), or thin/etched detailed sound( like benchmark dac ) or refined , analog like but polite in dynamyc (like Ayre qb9) is no go for this price segment and not for single vocal or small intimate jazz sound. Its for large clasical orchestra, wind band marches etc.
I like when dac reproduced sound with lot of weight and authority - when music "breathe" and throw energy to face(not midrange energy but that energy recreation of live ambience and energy flowing,sale venue when orchestra go from forte fortissiom to piano pianissimo- a silent which can be heard) Very deep ,powerfull bass and lower midrange is absolute nesessity and very top octave as well (for my ears those two extremes 20-30Hz and 13-17kHz create that live factor)
hope this helps.
So any interesting active non tube and not OPA stage (which have drive ability like krell discrete or of wadia hammer OPA- i never use preamp on my systems since 2003 I heard some nice dac, but when I conect them directly to 2 gain stage amps they sound very bad) for pcm 63p?, would like avoid coupling caps if possible.
maximum limit of dac weight 30kg, maximum dissipation 100W
Thanks for input, guys.
R.
Last edited:
You say, "thanks for the input", but do you read them? You keep asking the same questions over and over again. In the circuit Alexandre presented, R60 is the I/V resistor; all the rest is a low pass filter, which you don't want if you resample in the PC. Sixteen DAC chips per channel with a 150 ohm I/V resistor will give around 6V RMS output. Do you have 32 PCM63? Is 150 ohms ideal for the PCM63? Will the resulting output impedance match your amp? If the answer to any is no, you will have to make adjustments. Adding a step-up trannny gives you leeway to make those adjustments. With the proper adjustments, eight chips per channel should work but will be far from 'ultimate' which is what you say is your goal. If you add optional linear interpolation you will have the PCM63 equivalent of the 39,900 Euro Trinity DAC. If you do a modular design it should be easy to add more or different DAC chips, as you choose.
That will not work because it exceeds the output voltage compliance of the pcm63. It has internal protection diodes that will clip the signal with any load above 150 ohms or so.
The best advice I can give to Elviukai is to learn from the best, try to build upon something that is already very good, and compatible with his budget... my budget is low so my experience is limited to TDA dacs and gainclone amps, and the results are not bad at all. But I cannot give much advice here haha!
Just for clarification: that circuit is from Guido Tent et al
The full project can be found here: Welcome to the Audio DAC page
The full project can be found here: Welcome to the Audio DAC page
Yes I read them, but you must understand that i have never build dac from scratch, and it will be conected to at least three diferent systems) there is lot of thing to consider , which I can not know without actual digits. 6V rms(I assume its single ended mode -16 paraleld dac?) is good value, unfortunatley i have no such qty of pcm63p-y chips (16psc per whole DAC is my max limit) I may buy simple pcm63p-k even up to 100psc , but they are not the same as p-y versions sonicaly. my amps have average 24-26db gain with 15 to 30K ohms input impedance. I am not using any preamp. all this leads to using 16psc (or even less if there is no point for many chips in paralel) with some active stage i think.. 🙁
By the way, prior to posting here and asking for help I searched for few month any interesting dac with pcm 63 or pcm 1702/ 1704 which can be replaced with pcm63, and found basicaly nothing except this one in japaneese- http://easyaudiokit.hobby-web.net/kit-room2/DACManual/DAC63-4SManual.pdf which output stage is far away with what i consider (with my very limited knowledge) as exotic. I would gladly repeat someones project , but there is no such projects on the net.. So here is my only hope.
What will work though, and quite well I am told, is a common base or common gate transimpedance amplifier, also known as current-steering I/V. But you guys probably know this already. It´s basic form is already known to Elviukai for he has experience with a jfet variant of the D1. However, that jfet version is unsuitable for anything more than 4mA current, unless you parallel a lot of jfets (and there would be drawbacks). You´ll need a much beefier and higher bias circuit.
