Flying higher and higher with the Doede Dac

[jean-paul]

Non Os Unexperienced Prejudiced Person, a N.O.U.P.P so to speak

 

[ThorstenL]

Konnichiwa,

I can not understand why people built DACs with TDA1543.

Same reasons people build Amp's with LM3875. Good sound (often drastically better than supposedly "good" traditional implementations) and simple, quick and cheap to make.

Look at Doede, this guy build quite a few "serious" DAC's, what does he end up with - 8 X TDA1543 Stacked....

THD is -75dB in the data sheet

A 300B SE Amp has -26db THD at full power. It still sounds good. Your Ears have -10db THD, still, live music does sound unduely distorted. THD is one of the most useless of all audio measurements to qualify sound quality, right up with anechonic speaker frequency response measurements.

By comparison, the 1541 has -95dB.

And sounds better too, but implementation is not for the fainthearted.

Does the distortion noise floor provide tube sound ?

Absolutely no. Tubes too doe not sound the way they do BECAUSE of their distortion, but despite.

Had a CD614 with this chip. Sounded lower middle class.

I have a DVD Player with a super quality 192KHz/24 Bit DAC Chip. It sounds awful. Much worde than anything I ever heard with TDA1543. It is not the Chip, it is the implementation.

Sayonara

 

[Lucas_G]

Proper implemetation of TDA1543 is not that easy either, but almost any of them sounds pretty good. I'm currently prefering 8V and IC regulators. I never expected that it may come to this.
And BG at the output rule

PS: I still can't force myself to build a stack of 8 of them though

Presently I'm using AN8008 (Panasonic, used by Kusunoki in his 3rd DAC) and I'm surprised with a good sound. I will also try LM7808, as Fedde reports it sounds good too.
I'm using 3K for I/V and 1k58 for reference.

Hi Peter,

Did you ever compare 8 versus 8.5 Volt? Or is the 8V merely caused by the use of a fixed voltage regulator?

What type and value BG do you currently use at the output?
Did you try different types?

Why do you use 3K for I/V? I saw that Fedde too uses this value. Doede however states that the optimum lies with 2.2K (for 1 piece of TDA1543). That is why I use 274 Ohm with 8 TDA's.
If 3K would be the optimum, I should try 375 Ohm...

BTW, Soldering a DAC-Tower of 8 pieces TDA isn't that difficult.

Here is a picture of mine:

 

[Lucas_G]

Lucas,

I can not understand why people built DACs with TDA1543.

This chip is declared by Philips as "economy" and used in their el cheapo players.

It is the very worst of all.

THD is -75dB in the data sheet, your measurements are very accurate

By comparison, the 1541 has -95dB.

Does the distortion noise floor provide tube sound ?

Had a CD614 with this chip. Sounded lower middle class.


Bernhard

Oh oh, another spec hunter Trust the Philips data but let your ears be the judge please.

Try the TDA1543 in non os just for the heck of it. It will cost near to nothing so building one DAC with it won't hurt your wallet. Big chance you'll forget about the 4 x TDA1541A thingie.

Your CD614 had oversampling which makes quite a difference. IIRC it had an even more crappy digital filter than SAA7220 too. And the power supply could have been better too and... well you know what I mean.

To be honest I think TDA1543 is one of the better DAC chips soundwise but TDA1541A and TDA1545A I like more *when implemented right*. It is those last words that make building a DAC with 1543 so nice. Only one supply voltage and passive I/V, not much that can be screwed up except for PCB layout and a decent supply. I'll leave the comments about stacking TDA's and running them on near maximum allowable voltages to others.

Hi Bernhard and Jean Paul,

I guess both of you are right in this...

As things stand, I find the Doede Dac much better than one would expect from the ugly spectrum in post 2 of this thread.

When you take into account its low price, one can easily understand why people are raving about this dac.

When comparing its sound to commercial offerings, I would prefer it to most cd-players in the 1000 to 1500 Euro league, including the not so bad sounding Arcams and the like...

Its is also preferable to a stock Philips DVD 963 SA and to most of the other fancy super upsampling DVD-Audio SACD universal players, which all too often have lousy powersupplies and cheap output-configurations. A super DAC-chip cannot help there.

However, a well implemented player, like a Sony SCD XA 777 ES, a TEAC DV 50, or a modded Philips DVD 963 SA DOES SOUND BETTER than the Doede DAC

These modern chips DO give better resolution in the high frequency range, PROVIDED they are made to sing by their surrounding.

That said, the Doede Dac remains a bargain. I will see how much it can still be improved. That is why I hope many will post their findings with different mods.

Regards,

Lucas

 

[Peter Daniel]

Hi Peter,

Did you ever compare 8 versus 8.5 Volt? Or is the 8V merely caused by the use of a fixed voltage regulator?

What type and value BG do you currently use at the output?
Did you try different types?

Why do you use 3K for I/V? I saw that Fedde too uses this value. Doede however states that the optimum lies with 2.2K (for 1 piece of TDA1543). That is why I use 274 Ohm with 8 TDA's.
If 3K would be the optimum, I should try 375 Ohm...

BTW, Soldering a DAC-Tower of 8 pieces TDA isn't that difficult.

Here is a picture of mine:

8V choice is made by use of fix regulator. Fedde suggested that voltage and I really like it now. Comparing to 5V, I had previously, it is more forward, detailed with better soundstage. 5V is more laidback, relaxed, smooth and toned down, but somehow less involving and colorfull. You could listen for hours to 5V version without getting tired, but it doesn't present anything that makes it stand out either. It's almost like comparing inverted and NI GC amp

I made my choice for 8V supply and I'm trying to improve it further. BTW, if I'm not mistaken, 8V and 3K I/V resistors are close to what 47Labs using.

However I'm not opting for a simple supply, like most others do, but using 2 separate transformers with extensive CLRC filtering, and triplets bypassing. Maybe that's the key to good sound I'm getting from that DAC?

The output caps are 4.7 BG N, but I made an arrangement for super pair as well to see how it works.

Also, most other major parts are installed on sockets, so I can easily switch them and choose the ones that sound the best (all bypass electrolytics are done in that way).

 

[Peter Daniel]

Just yesterday somebody brought a CDP to be modified and I compared it to my setup. CD-Pro2 driving outboard TDA1543 DAC was better than that unit, which some claim comes close to Audio Aero.

 

[Lucas_G]

8V choice is made by use of fix regulator. Fedde suggested that voltage and I really like it now. Comparing to 5V, I had previously, it is more forward, detailed with better soundstage. 5V is more laidback, relaxed, smooth and toned down, but somehow less involving and colorfull. You could listen for hours to 5V version without getting tired, but it doesn't present anything that makes it stand out either. It's almost like comparing inverted and NI GC amp

I made my choice for 8V supply and I'm trying to improve it further. BTW, if I'm not mistaken, 8V and 3K I/V resistors are close to what 47Labs using.

However I'm not opting for a simple supply, like most others do, but using 2 separate transformers with extensive CLRC filtering, and triplets bypassing. Maybe that's the key to good sound I'm getting from that DAC?

The output caps are 4.7 BG N, but I made an arrangement for super pair as well to see how it works.

Also, most other major parts are installed on sockets, so I can easily switch them and choose the ones that sound the best (all bypass electrolytics are done in that way).

Hi Peter,

To use ic-sockets also for bypass-caps is a very clever idea indeed! I think I will steal that from you.

I use a Panasonic 12 Ah battery for the Doede Dac, just like Doede himself. I am very convinced of the use of batteries, because they do provide a very clean supply. When using an automatic charger it also becomes quite easy to handle. However, in Doede's implementation, the voltage regulation (here with a LM317) introduces its own noise again...
That way much of the purity of battery-supply is sacrificed again. One can of course try to filter extensively after the LM317, but what is the use of a battery if one has to bring in filtering in the end?

I am thinking of building a second DAC with mains powersupply and extensive filtering. I strongly believe that you are right on the relation between powersupply and soundquality in DAC's.
I am not yet sure whether to choose the Kwak-like regulation (like Fabian describes on his TDA1543 website) or simply with LM7808... Did you do a comparison?

I am curious to know what you think of the BG super-e configuration. I use two BG-N 220uF in such a configuration in my Philips and I like the sound a lot, but I have never compared it with a single BG-N of 470 uF....

Regards,

Lucas

 

[Lucas_G]

Just yesterday somebody brought a CDP to be modified and I compared it to my setup. CD-Pro2 driving outboard TDA1543 DAC was better than that unit, which some claim comes close to Audio Aero.

I think the TDA1543 let you better hear into the dissonances of music...

 

[Peter Daniel]

Did you do a comparison?

One surprising comparison I did here: http://www.diyaudio.com/forums/showthread.php?s=&threadid=26001&perpage=15&highlight=&pagenumber=5

 

[Lucas_G]

One surprising comparison I did here: http://www.diyaudio.com/forums/showthread.php?s=&threadid=26001&perpage=15&highlight=&pagenumber=5

That is a very interesting post! I too have come to think that BG-N sound better as decoupling caps than than most of the expensive and large MKP's. They are simply slower, less transparent and neutral, in my experience...

Is it further true that in the large setup you used Elso Kwak's voltage regulation, and in the small setup you used simple LM7808's?

 

[Peter Daniel]

I used Elso regulator at 5V in bigger DAC and AN8008 Panasonic reg in the smaller DAC. Also smaller DAC had less LC filtering.

 

[Bernhard]

Okay, just what I have read here was about oszillations of the stack, distortions that come and go , and stack sounding worse than single chip.

About implementation:

In my CD304mkII they use ceramic smd chips for the 14 decoupling caps and it sounds good.

But also ground plane on the upper layer of all boards.

Bernhard

 

[Lucas_G]

I used Elso regulator at 5V in bigger DAC and AN8008 Panasonic reg in the smaller DAC. Also smaller DAC had less LC filtering.

Ok, then I will stay with the LM7808 that I already have.

Btw, I noticed that the left and right channel of my 8 piece Dactower are not exactly symmetric. I have used 0.1% 275 Ohm Welwyn resistors for the I/V, and I checked them myself. Still the volume of one channel is a bit louder, and the DC-offset is 0.06 Volt higher on one channel. I guess the TDA 1543 is not so accurate on left/right symmetry...
If one considers the fact that I use 8 of them in paralel, deviations must be even much greater when using one piece of TDA1543. Did you ever check this?

Peter, can you also tell us what caps of what values you are using in the powersupply?
Currently I have only got a 33 uF BG-N at the TDA, just like Doede... For 8 TDA's consuming half an Ampere it seems a little bit low, don't you think?

Thanks for all you valuable info here!

Regards,

Lucas

 

[Lucas_G]

Okay, just what I have read here was about oszillations of the stack, distortions that come and go , and stack sounding worse than single chip.

Bernhard

Hi Bernhard,

It is good to be sceptical when other people seem to rave...

Yet, building such a cheap DAC can be quite funny, and if you are not wholy satisfied yourself, just sell it to some poor little fellow that has to buy the usual overpriced gear that shops have to sell... Ten to one that it will sound better than that crap.

Btw, what DAC or cd-player do you listen to? Are you satisfied with its sound?

Regards,

Lucas

 

[Chris]

Tuning and tweaking

I have not built the doede dac but a Scott Nixon dac. Using only one tda1543. I built it as simple as possible just to get an idea regarding the sound. So far I`m happy with the sound but understanding that there is room for some improvments I have a question:

Have any of you tried the decimation circuits described at
TNT-Audio?

http://www.tnt-audio.com/clinica/convertusdecima_e.html
http://www.tnt-audio.com/clinica/convertusdecdig_e.html

Any comments about these circuits versus reclocking?

Thanks Chris

 

[Lucas_G]

Re: Tuning and tweaking

I have not built the doede dac but a Scott Nixon dac. Using only one tda1543. I built it as simple as possible just to get an idea regarding the sound. So far I`m happy with the sound but understanding that there is room for some improvments I have a question:

Have any of you tried the decimation circuit described at
TNT-Audio?

http://www.tnt-audio.com/clinica/convertusdecima_e.html

Any comments about this circuit versus reclocking?

Thanks Chris

Hi Chris,

I had read the article too, but I did not fully understand how it works.
That is also the reason why I forgot about it. Maybe it can do a nice job in the treble-range... I hope someone has experience with this...?

Regards,

Lucas

 

[Peter Daniel]

Peter, can you also tell us what caps of what values you are using in the powersupply?
Currently I have only got a 33 uF BG-N at the TDA, just like Doede...

33 BG N at the DAC seems like a very good choice. I wouldn't go any higher and I had excellent results with just 4.7u at the DAC. Bigger caps vail the sound and make it sound slower.

That said, I have around 14,000u in the DAC and receiver ea. raw PS.

But I'm also tempted to just use a battery and forget about all those inductors and transformers.

 

[Fabian]

Btw, I noticed that the left and right channel of my 8 piece Dactower are not exactly symmetric. I have used 0.1% 275 Ohm Welwyn resistors for the I/V, and I checked them myself. Still the volume of one channel is a bit louder, and the DC-offset is 0.06 Volt higher on one channel. I guess the TDA 1543 is not so accurate on left/right symmetry...
If one considers the fact that I use 8 of them in paralel, deviations must be even much greater when using one piece of TDA1543.

Hi Lucas,

I measured dozens of TDA1543. The right channel output has almost always more DC. The deviation between different chips can be quite large. Here are some results:

VDC L    VDC R

3.84    3.94
3.78    3.79
3.87    3.92
3.68    3.78
3.75    3.84
3.73    3.84
3.75    3.82
3.77    3.77
3.80    3.75
3.78    3.87
4.06    4.02
3.81    3.78
3.90    3.94
3.68    3.85
3.77    3.85


Regards,
Fabian

 

[Peter Daniel]

After some trouble with proper voltage adjustment (it's not that easy in CLRC filter), I got this DAC running http://www.diyaudio.com/forums/attachment.php?s=&postid=327626

As you noticed I mounted most of the caps on sockets, so I can easily swap components and experiment with the sound. Although this is not Doede Dac, it is still TDA1543 based and my observations are as valid to other, similar DACs.

On classical and jazz recordings the first impression was quite positive, but when I switched to contemporary rock and country, it didn't perform that well. Comparing to my other DAC it was very muffled and sounded almost distorted. Well, I decided first to get rid of OsCons. I tried them before, and didn't like them that much, so here they go. The sound opened up a bit, but still not to the point that I'm used to. Another thing that I decided to let go is that famous super-e configuration of output 4.7 BG N. I tried to parallel those BG caps before, and they always sounded better in singles. So now, when super-e setup is gone, the DAC started to sound pretty decent. The fine detail and acoustics of the recordings are back and vocals opend up to the point that thay actually sound real.

I'm using 10u Gold Nichicons on all digital chips and 33u on the DAC. I will be getting type N 33u BG or maybe even Hi Q type (they are better, right? ).

Next thing to try out is different regulators on the DAC and finally reclocking and input buffering effects. Will keep you updated.

 

[Lucas_G]

Hi Lucas,

I measured dozens of TDA1543. The right channel output has almost always more DC. The deviation between different chips can be quite large. Here are some results:

VDC L    VDC R

3.84    3.94
3.78    3.79
3.87    3.92
3.68    3.78
3.75    3.84
3.73    3.84
3.75    3.82
3.77    3.77
3.80    3.75
3.78    3.87
4.06    4.02
3.81    3.78
3.90    3.94
3.68    3.85
3.77    3.85


Regards,
Fabian

Hi Fabian,

I am somewhat relieved to see that you have made the same experience.

Would it be a good idea to do some TDA selecting? Taking only those TDA's that have a very small left/right deviation?

With a price of only 70 Eurocents one could afford to do some selecting...

Regards,

Lucas.