AK4396: best solution for output stage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
While I am waiting for my 1543 DAC from HK, I did some research on my Emu 0404 USB. Sound is not bad but not the crown in audio.

So I found some passive approaches for the output stage just to omit the opamps.

In the datasheet is reported that Aout+ and Aout- is 2.8Vpp each so in addition you'll get 5.6pp or 2Vrms if you're going symmetrical. That's my problem!

I need unbalanced connection and so my question arose: is there a (passive) possibility to keep the 5.6Vpp for unbalanced output?

The XLR solution could be:
 

Attachments

  • ak4396 pass xlr.gif
    ak4396 pass xlr.gif
    5 KB · Views: 7,702
No filter needed

The AK4396 will sound it's best running direct out through the coupling cap without the analog ultra sonic filter. The ultra sonic noise spectrum of this chip is unique in the industry in not needing one. Adding the 1k resistor to the output impedance of the chip tends to choke the sound. Just take the signal out through the blocking cap from one leg as you have shown in the second drawing for 2.8v and omit the filter. No output R or cap to ground. Leave the other leg open, or terminate with a blocking cap and resistor to ground which approximates the load the the + leg will see.
.
Check out the noise graphic of the AK4396 here.
.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=137703
 
You should never use the output of the DAC-chip directly. These are normally very sensitive to the load (Even though the Zout is low).
Also, you should never use onle the 1 of the outputs of a differential DAC, to have a simple solution to the single-ended output. You have to use a differential-to-single-ended converter.

Doing as above, you will never ever discover the potential in a DAC like the AK4396.
Look at page 34 on the AK4396 datasheet:
AK4396 datasheet

This is the simple approch to diff-single-ended conversion, using a single op-amp. I must emphasis, that this is NOT the best solution, but it is way better the above!

The best solution is a discrete design, prefferable with no negative feedback. Search the forum, and you will find many of these.

AK4396 is actually a great chip, and if you give it the right invironment, you will discover that the TDA1543 is like driving an old VW Buggy. Not worth the time, even though it's quite funny.

And be aware, that when you listen to AK4396 (or similiar) based DAC-systems/CD-players that do not sound that good, it's not because of the DAC-chip. Typically it's because the analog stage is made completely wrong.
Funny thing we discovered in the 4 years of developing our DAC is, that most of the thing people call "Digital sound" (As in negative), actually has nothing to do with the digital parts. It's all in the analog stage. The reason that you still hear these issues after 25 years of improvement in DAC-chips is, that most CD-players is still equipped with the same low grade op-amps as in the early 80's.
 
Hi Hurtig

is fig. 17 (3 opamps) so much better than fig.16 (1 opamp)? What are the differences?

I am trying my 0404 USB at the moment with Foobar2000 on my old Dell Notebook. The bad thing are the irregular crackles. Is this normal for EMU 0404 USB or is it a PC problem?
 
Hi Thomas
I don't know anything about the EMU 0404 USB. But I really do not like the idea of connectiong the computer to the stereo. Computers are a noise-nightmare.
If you do connect, galvanic isolation would be wanted. But almost any USB DAC/Soundcare do not have this feature, and will result in noise entering the stereo. This could very well lead to the symptoms that you mention.

About the 2 different schematics in the AKM datasheet. No, the 2nd one is not better. I just gives you balanced output to. But it is still an op-amp, follow by another op-amp. Not that good.
You could get OK results from trying different op-amps. But it's still an op-amp and sounds just like it.
I would go for a discrete analog stage using NO negative feedback.

If you do choose an op-amp, my suggestion is the old fashion NE5534. If you can get the "A" version it's even better.
I have seen many people change from NE5534 (5532) to newer and "better" devices, only because of the better specifications in the datasheet. People who actually run a blind-listening-test, tend to go for the NE5534. For an op-amp it's not that bad, and sonically it will run down most of these new "racing-op-amps".
 
Ever tried it?

Hurtig said:
You should never use the output of the DAC-chip directly. These are normally very sensitive to the load (Even though the Zout is low).
Also, you should never use onle the 1 of the outputs of a differential DAC, to have a simple solution to the single-ended output. You have to use a differential-to-single-ended converter.

Doing as above, you will never ever discover the potential in a DAC like the AK4396.
Look at page 34 on the AK4396 datasheet:
AK4396 datasheet

This is the simple approch to diff-single-ended conversion, using a single op-amp. I must emphasis, that this is NOT the best solution, but it is way better the above!

The best solution is a discrete design, prefferable with no negative feedback. Search the forum, and you will find many of these.

AK4396 is actually a great chip, and if you give it the right invironment, you will discover that the TDA1543 is like driving an old VW Buggy. Not worth the time, even though it's quite funny.

And be aware, that when you listen to AK4396 (or similiar) based DAC-systems/CD-players that do not sound that good, it's not because of the DAC-chip. Typically it's because the analog stage is made completely wrong.
Funny thing we discovered in the 4 years of developing our DAC is, that most of the thing people call "Digital sound" (As in negative), actually has nothing to do with the digital parts. It's all in the analog stage. The reason that you still hear these issues after 25 years of improvement in DAC-chips is, that most CD-players is still equipped with the same low grade op-amps as in the early 80's.
.
Have you ever tried the AK4396 running direct out?
 
I have not tried the AK4396 running directly from the outputs, but I did try on some similiar IC's from AKM and Crystal. They all tend to be quite sensitive to the load, and to have optimum performance, you must use the signal from both + and - outputs, since the signal will contain common mode signals.

If you want to achieve the best performance, there is no shortcut. I know a lot of people say they get extreme results from old Philips DAC's like the TDA1543 often running with no oversampling and very bad analog stages. I still believe it to be a case like people who like the performance of an old VW beatle, just because the did some bodywork on their own.

Face it.... Solutions like that will never get you high end performance. If you like it, it's probably because you like that it destroys the dynamics of the music. Some people actually thinks of this as "Musicality". But it's not.
Campare it to a high end DAC, and you will notice all the things that are missing in the music.

I did listen to a few TDA1541/43 based DAC's, of which at least one of them has been called "High End".
But sorry to say... They all sounded terrible. For background music it's OK, since there is absolutely no dynamics. But again... That's not the idea of High End. Who would call an old taperecorder High End today?? Quite funny, because these old DAC's does remind me of the sound of a taperecorder. He he he... wow... That's High End :D (In the early 80's....)
 
@ Thomas:
I do know a better solution for the AK4396. But since I will launch it in a CS4398 based DAC soon, I really can't reveal to much.
It is the product of about 4 years of developement and listening tests, and may very well set a new standard for digital audio gear.

You could start by cloning the LC Audio ZAPFilter. Maybe upgrade it, by using a push-pull output. This will most likely kick some op-amp ***.
 
Hurtig,

some people like midfrequencies, either for hifi-religious or psycho-acoustic reasons. Not long ago Dynaudio made (mabe still makes) a loudspeaker they advertised as being made for listening to acoustical music. It contained only one of their higher end bass-mid drivers and it's frequency response topped officially at 8 kHz.
 
@ Ola:
Funny.,.. I live about 2km from the Dynaudio Headquarter.

I am surpriced that they launched a speaker that will roll off at 8kHz.
I do not see what this has to do with the AK4396 output stage....

But anyway... try the discrete non feedback design, and you will be surpriced....
In our developement model, we had an arrangement, where we could switch in/out both the discrete design and an op-amp based analog stage in the same DAC-PCB. We have tried MANY op-amps, and yes. They all sound different.... But... Still so much alike. The simply sound like an op-amp! None og them was even near the performance of the discrete design.
But among the op-amps, the NE5534A was a true winner!
 
The bad thing are the irregular crackles. Is this normal for EMU 0404 USB or is it a PC problem?

Your problem with the crackles is almost certainly because of your computer not keeping up with the audio stream (and not because of badness of the analog circuit in the DAC). Fix that first before looking into exotic things like opamps and capacitors.

Probably not because of low computer speed, but is more likely due to a hardware device in the computer interrupting the processor for so long that it can't keep the DAC buffer filled. This happens very commonly with Bluetooth or Wireless LAN devices within the machine (because of bad driver design and particularly often on laptop computers). Sometimes a driver update can fix it, but usually disabling or uninstalling the wireless device is the only fix.

To find out if this is the case (and to check whether your computer can even work effectively with audio), highly recommend that you download and run the DPC Latency checker (it's freeware) from
http://www.thesycon.de/deu/latency_check.shtml
This can make all the difference! (The 0404 is a pretty good DAC)
 
Telstar said:


Why the choice of this chip?


Well... quite a hard choice. Kurt_von_Kubik who is also part of this project, is a hardcore fan of Burr-Brown. So we have been working with the BB/TI chips for a long time during the developement.
I have been a working with Crystal DAC's since the late 90's, and suggested the CS4398. None of us really believed that it would match the BB chips, since these generally have a little better data. But the CS4398 has shown extremely good results, when combined with our discrete non feedback analog stage.

The choice of CS4398 is 100% based on listening test, just like everything else in the design. Thats why we have been woking more than 4 years on this project already!
Currently we are working on implementing an ASRC in our prototype. None of us realle believe in ASRC's, but we have decided to give it another try.
 
Hurtig said:
@ Ola:
Funny.,.. I live about 2km from the Dynaudio Headquarter.

I am surpriced that they launched a speaker that will roll off at 8kHz.
I do not see what this has to do with the AK4396 output stage....

But anyway... try the discrete non feedback design, and you will be surpriced....
In our developement model, we had an arrangement, where we could switch in/out both the discrete design and an op-amp based analog stage in the same DAC-PCB. We have tried MANY op-amps, and yes. They all sound different.... But... Still so much alike. The simply sound like an op-amp! None og them was even near the performance of the discrete design.
But among the op-amps, the NE5534A was a true winner!


Interesting, every time I replace a 553x, I found a big improvement. Is there a specific brand/grade of 5534 that sound very good.
BTW, I found the tda1543 don't lack in dynamics and sound better than the cs4397 dac (P3A). Go figure...:xeye:
 
Bill Fuss said:
Hi, I can't resist adding my 2 cents. I am a recent convert to transformer output with a CS4398. I have been able to directly compare it to a Borbely discrete OP stage and the transformers are clearly superior, but YMMV.

Best, Bill


Transformer... Hhmmmm... I really do not agree that this is the way to go. I guess you feel the sound more soft and maybe a deep black background. Right??

Just be aware, that the softness, is the result of the transformer killing every inch of dynamics.
And about the black backgroud: I know very few CD's recorded with such a black background. If you experience this black background, it's typically because everything below a certain level is gone. I have seen similiar results from a lot of classD amps. Very impressing at first because the details left seems more clear. But after a few days you discover, that much of the small details is totally gone.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.