Building the ultimate NOS DAC using TDA1541A

Happy new year everyone,

About a month ago I put together another I2S to Sim PCB but this time I put in a dual inverter chip to invert the data lines so I could try out the 1541 in differential mode. Im really enjoying how it sounds but I have a question I would like to put out there to the much more knowledgeable people than I-

Does the LSB error occur as a result of how I have inverted the DATA? I remember reading something about a 1 LSB DC offset error somewhere - does that apply here? Does this mean at full scale current there will be a 1 LSB error? Could someone help to explain this? Thanks.
 
Last edited:
Happy new year everyone,

About a month ago I put together another I2S to Sim PCB but this time I put in a dual inverter chip to invert the data lines so I could try out the 1541 in differential mode. Im really enjoying how it sounds but I have a question I would like to put out there to the much more knowledgeable people than I-

Does the LSB error occur as a result of how I have inverted the DATA? I remember reading something about a 1 LSB DC offset error somewhere - does that apply here? Does this mean at full scale current there will be a 1 LSB error? Could someone help to explain this? Thanks.
The 1 LSB error only occurs when you use the Signed Magnitude circuit John suggested.
I am so sorry I haven`t had time to assemble the Signed Magnitude you sent to me, but there has been so many other projects, audio and non audio related. I hope at some time in this year I will come to that.
 
Thanks for the reply koldby, ok so when the sign bit (when present depending on the data format) is inverted it causes this 1 LSB error... I think I get it now. So with the setup I have its should be all fine - the only issue I can think of is the propagation delay when inverting the data which may be an issue running at higher sampling rates.

I haven't made a signed magnitude pcb, i think you may be mistaking me for somebody else?
 
Thanks for the reply koldby, ok so when the sign bit (when present depending on the data format) is inverted it causes this 1 LSB error... I think I get it now. So with the setup I have its should be all fine - the only issue I can think of is the propagation delay when inverting the data which may be an issue running at higher sampling rates.

I haven't made a signed magnitude pcb, i think you may be mistaking me for somebody else?

You are right. My bad.Sorry
 
Thanks for the reply koldby, ok so when the sign bit (when present depending on the data format) is inverted it causes this 1 LSB error... I think I get it now. So with the setup I have its should be all fine - the only issue I can think of is the propagation delay when inverting the data which may be an issue running at higher sampling rates.

I haven't made a signed magnitude pcb, i think you may be mistaking me for somebody else?

As I se it, the reason for the 1 bit error has to do with the way two DACs are put together in a signed magnitude operation. One TDA1541 has a zero crossing and are then symmetrical around this: Equal amounts of + bits and - bits. But when you combine two, you do not have this zero crossing bit, hence the LSB error.
IMHO.
 
Disabled Account
Joined 2019
Really time John is drawing a standalone pcb without errors... It's not that we learn nothing it's more than pcbs made by enthusiasr memnbers have errors between the drawing, the devils in the details with ground, impedance, groundloop, etc.... and the export files and the pcb printing given to the factory.

But I do not see any inducement from members but 4 or 5. Fellows would like tbis idea to be realized ?

John, are you ok or not with that idea or not ? Have you changed your mind on the proposal you made about this pcb ?
 
As I se it, the reason for the 1 bit error has to do with the way two DACs are put together in a signed magnitude operation. One TDA1541 has a zero crossing and are then symmetrical around this: Equal amounts of + bits and - bits. But when you combine two, you do not have this zero crossing bit, hence the LSB error.
IMHO.
Makes sense. Thanks for clearing that up.
 
Disabled Account
Joined 2019
Short answer : yes you're right. But...

Some are more efficient, different backgrounds, deepness of language understanding, involvment. You right, we learn a lot thanks to John's dynamic.
But devil again is in the details. There are many good wills but that often fall in individual enthusiast projects as Baron Rouge board, Distinction and so on which have earned their existence have always some dfaults due to some errors ornthe natural lack of knowledge in some areas, sometimes just by lack of time too or because most (sometimes good experienced EE) believes they can master all alone with more success than a team.

My limited experience despite I'm found of this TDA chip is it's hard to make an efficient pcb and rounting. Going from scopes to ears and find trade offs. The best sucess is imho what Pedja Rogic did with the ask of some enthusiasts by putting the Aya on the work shop and draw improved pcbs just because he can. Even if I read some good and ancient threads or books here about pcbs layout (a guy here as Marce for instance but some others too like John here)...
For me this last Aya diy kit from Audial is sounding better than others diy tda1541 today. My idea, perhaps to naive or lack of my self lack of confidence is few master this chip in many areas. Thorsten Losch with a team which share different talent had success, EC Design made a huge work as you highlight, so Audial as well... not too much people, proves in the pudding when you often listen the others poorer diy (some make individual good things because the time spent to coock such diy dacs according the rest of their hifi chain...but often lack of some good parangon dac to benchmark their work.)

Anyway, I just asked about something John's said in a post, but John's silence and lack of +1 to that idea (just 4 people seems ok) has aborted. What a pity, all the good dacs here born from people like Pedja Rogic, Soekris, pcm dddd 1794 dac and are improved for the best in an itterativ work of this communitty... I think the best illustration was the Soekris which was sounding not soso but greatly improved by exchanges with the designer who understood after a time he mastered not all the knowledge. I remember the first contributions were just differents caps and more decoupling and gave a much better result to the work of clever people about programming filter for the fgpa for something much btter at the end... great example of coolaboration which should be learned in enginnering schools. And worst are project where the diy enthusiast is closing himself like an oster by too much ego and self confidence, at the best their good ideas are weaked by sides they believes to know or even worst think it doesn't work. But facts are stupborn when you fire a good dac, it sounds far better than others.

I putt John, Rogic, losch above those guys with their mojo and sharing attitude, it's not only self ads...it really works, proves in the pudding when you listen their stuffs vs others.

But maybe you're right rfbrw... in DIYA , maybe the A is for alone...but internet is from my limited point of view a fantastic co working tool. And it's always win-win.

What a pitty, it would had been fun to make a great construction thread around a standalone pcb like the Soekris (which was not enough standalone pcb as he needed I2S or spidf external pcb) but drawed by John's mojo...
Well let close that idea...
Cheers.
 
Last edited:
Putting aside how things sound as there is no single answer to that, two things occur to me.
The first is that you learn by doing. No one is born knowing all this stuff. The second, perhaps more relevant to this situation, is that I have found it easier to get someone to help you if you are prepared to meet them halfway. Rather than blowing smoke up the ar*e of your guru of choice, come up with a design and run it past them. They may or may not choose to offer some tips on how to improve it but at the very least you are moving forward and learning something .
 
Disabled Account
Joined 2019
That's not untrue and we DID learn something... just took back John's idea to make us a board and gave also an incentive idea by buying him in a bundle his usb to toslink thingy (look at EC Design website)...populating a board is not maybe enough diy today after all the tip given I assume if I understand what you write... Well let forgett it.:( and btw you just have no idea of what I did or learnt, anyway I have not John talent and don't want to make a Baron or Distinction board too, despite the tips, not my pudding. Enough bad cakes...time wasted. And all these bad chineese tda boards also illustrate than most don't understand as the beleive they do. I also find this thread is turning into a commercial one for some to sell expensive not so great pcbs or adverts new dacs which are not tda1541 anymore...If you find it better thzn blowing smoke, free to you, we just disagree.
Cheers
 
Last edited:
Disabled Account
Joined 2019
@HK, Hi, well for the moment it is the best of all the ones I heard in all my diy adventures.

@rfrbw, you refuse to get my point. No, the best chance for better and more music is the process I described according to me, again solid and professional basement then diy communauty for itterativ enhancement... or you fall in the Red Baron, Distinction, Diyiggy or what you want from a standalone enthusiast: a collossus with clay feet... Hallas each way a ritical listeninga always show it's lack of time. Sucess came from a concrete basement then embrassed by the whole as it was done by Soekris, Pedja... so in fact the oppositt of what you think even if true from a human individual to individual relation point of view. I assume it's ALSO diy attitude to ask such things from who you call guru and from who I call a professionnal with experience in a win win attitude. After all it worked already, Aya, Damn boards. Well you will not convince me nore I will and only John can answer. For the moment I just notice you're the only one telling it's a bad idea. Anyway, it's John 's thread, idea was launched and it's my freedom to ask whatever you like it or not. More music for people, less BS...
 
diyiggy.. another thought would be to buy John's current production DAC. I think it is fair to say he tried for years to reach perfection on the TDA1541a chip and ultimately abandoned it. His shared journey has been a phenomenal resource to us all, and some pretty good TDA1541A PCB's have been spawned from this thread. But maybe while they are good enough for some of us, the answer to next step performance if not satisfied is to support John by buying his gear and using it as a base to tweak and add your own special sauce. Max appears to be on that path and enjoying it.
 
Disabled Account
Joined 2019
:) yes it's true and thank you to John, it s a book than this thread, just fantastic. Lucky guy you have tested his last R2R off shelves, am I understanding you ?
I think some here bought his last off shelves tda1541 some years ago too, certainly great collector.

I'm still confident with this tda 1541 chip, I believed moderner chips had more resolution but I discover clearness is not all and for us we always listen to redbook reccording it stills make sense. But yes I find damnly hard to tame by non proffessionnals too isolated, it always fall despite the tips given by something not good enough for my own tastes and ok it's maybe good enough for most people and by the way I am in this category after tons of dacs with an Aya board. I maybe putt in this chip too much hopes (and no monney in the new technology ec design is selling a democratic price btw. But when I see what happened to the Soekris, I think the way this Demning wheel evoluted is a good story, it is highlighted in AudioStream for instance. My limitex experience is tda1541 and ad1862 sing with better aesthtic trade offs imho (at least for the red book format) but evil is Really in the details despite all the tips given already. A standalone finely drawed board from my point of view would go for a better next step, maybe I'm wrong. Or maybe an official one in the ec design website, sort of enhanced diy offer for the people having the tda chip already ? As far I know Rogic did it this way, chips source non exist anymore (iirc John sold the last as commercial seller).

I forgett it, let's continue and sorry for the disgress, I had to try as I love this chip that sweet our lives with music...
 
Last edited:
Im in the same boat as Diyiggy. Gone through my fair share of commercial dac as well as diying a couple of dac's including the 1541. The 1541 has something special in it. I always feel that at least for Red Book playback there's still life in it which has yet to be fully capitalized. For now yes AyA is about the best kit out there but at the back of my head I've always wondered what it would sound like if the design is implemented by another Guru like John. Audio seems to be like a merry go round like SS back to Tubes & Digital back to Vinyl, who knows one day this might happen for 16 bit playback. Lol
 
If Audial is no longer producing 1541A DACs, use it as best available for SOTA core PCB?.

Perhaps Pedja would make them available (true limit seems availability of the chip).

If not .. u.fl I2S/PCM input (select), 1541A, 14x decouple pads SMD or thru hole (for lower f.DEM), local regulators sans feedback, active or passive DEM (select) open PCB trace for synch to TDA input, or ext.

Simples!
HK
 

Attachments

  • Captured.JPG
    Captured.JPG
    21.7 KB · Views: 510
Last edited:
Sumotan,

To my mind, previous assertions are correct; John gave up on 1541A.
There must be reason for this: Others have had long term commercial success with it as bespoke and high-end product, he has not. Join the dots.

Interesting to me how so many of us have the same Pedja board :).. probably for a reason.

I suggest the improvement is elsewhere; input and output 1541A, DEM circuit. Subjective opinion (and bias/ego) are the real obstacles. So far as 1541A layout, I think its done (AYA), a cut-down PCB with direct input and outputs would provide a useful platform for experimentation and there really is much to like about that.

But I have one already..
HK
 
Last edited: