DAC project completed

Status
Not open for further replies.
Bernhard said:


Well, I find that a bit questionable, somehow.

What are the bandwith and impedance specs of your shunt regulators ?


I fully understand your doubt about this. Anyway.... PSpice simulations has shown an output impedance of approx 7milliohm from DC to 20MHz.
Real life measurements has shown something around 12-13milliohm from DC to 10MHz.

I would believe tata is sufficient.

The thing that we really like about this, is the absolutely constant output impedanse. You newer get this from capacitors in paralell.

And since you have no decoupling capacitors to achieve this low impedance, you don't have to fight the different sounds from different capacitors.

The idea of the shunt, has come from turning the problem head down. Normally you use decoupling capacitors, because your regulator do not have low enough impedance, espiceially at high feequence. You know capacitors is emergency solution, but you need this.

We took a different approch... What do we need?? We need low impedance. Not just at DC and 100kHz... We need low impedance from DC to let's say 1MHz, and we need this impedance to be as constant as possible in this ferquence range. How to do that?? Well... Constant impedance over frequency exclude capacitors. Therefor we decided to go for a regulator good enough, to not need the capacitors to decouple at high frequency.

T´he shunt itself isn't any secret... It has been used by various designers. But in combination with a true current source, we have achieved extreme PSRR in the regulator.
PSpice simulations say around 160-180dB. I have not been able to measure it yet, but my guess real life is around 120dB.
 
Kurt von Kubik said:
Thanks a lot!
The choice of reciever is practically done by the industri and not by us, as the CS8416 simply is the industri standard.
The only real alternative is CS8420 which is the same but with internal ASRC, which we did not want to use in our design.
The DAC chip itself is chosen because it is a state of the art component. We experimentet with several others from BB, but these need an I/V conversion stage, which complicates the design further. And by all means the DAC already contains more than 270 components on the PCB. An additional I/V stage is already designed, but we considered the pro´s of Burr Brown PCM1794A to be minor to the drawbacks of the I/V stage.
Therefore the Crystal chip was chosen, as we do not see any competitive alternative to that on the market by now.
The Burr Brown though has a very nice digital filter on board, and it is practically quantization noiseless. The CS 4398 does need a bit of filtering as there is some quantization noise above about 350 KHz.

The CS8416 might be the industri standard, but it is also quite old, and there are far better options out there. For the current state of the art you should look at the WM8804. The DIR9001 is also a good choice, although it is limited to 96 kHz.

Your choice not to use a current output DAC such as the PCM1794 is one I can understand, however I disagree with your reasoning as to why the CS4398 is the best voltage output DAC. I do not want to turn this into a measurment vs sound quality discussion, but I believe the CS4398 does not sound as good as the WM8740 or AK4396, despite having somewhat lower distortion. I also believe that most people will agree with me here.

But still, good job. :)
 
Kurt von Kubik said:


Real world use of the fine specs of i.e. PCM1794A is very difficult as the output current is more than 15mA in monomode, which calls for i.e. NE5534AN opamps for I/V conversion, and we definately did not want to do that.


But I/V conversion does not equate opamps. There are so many other choices.


Why would diyers be interested in buying a complete dac though? You stand a better chance marketing it to a different crowd. And you may be ten years too late. How many would seriously consider buying a dac with only an spdif input interface?
 
Cauhtemoc said:


The CS8416 might be the industri standard, but it is also quite old, and there are far better options out there. For the current state of the art you should look at the WM8804. The DIR9001 is also a good choice, although it is limited to 96 kHz.

...however I disagree with your reasoning as to why the CS4398 is the best voltage output DAC. I do not want to turn this into a measurment vs sound quality discussion, but I believe the CS4398 does not sound as good as the WM8740 or AK4396, despite having somewhat lower distortion. I also believe that most people will agree with me here.

But still, good job. :)

I agree with the above quote.

Also, the cs4398 is not as detailed and much more relaxed than the AKM4396. The AKM4396 is an outstanding voltage out chip. This is my experience working with the respective company's eval boards. The cs4398 is clean and clear sounding, but very mellow and relaxed. It's the least musically involving chip sigma-delta chip I've heard.

AKM is also a easier company to do business with as well, from my experiences.

-David
 
analog_sa said:



But I/V conversion does not equate opamps. There are so many other choices.


Why would diyers be interested in buying a complete dac though? You stand a better chance marketing it to a different crowd. And you may be ten years too late. How many would seriously consider buying a dac with only an spdif input interface?

Well we are not marketing anything to anyone.
We finished our project and we both felt free to be a bit proud of the result.
We could keep it as a secret or we could disclose it to fellownerds.
Anyways! if you think of it as a kind of show of :smash:
feel free to do so.
We are just enthusiasts with one single goal - sound reproduction in its purest and most direct understanding and only that.
SPDIF is in our opinion the way to transfer digital information, USB is not in question, as this is far to noisy and jittery for this kind of circuitry, the only alternative I could think of would be I2S, but that is not standard between components nowadays.
Also firewire seems to be decending out of the standards, and I can recall only about 3 or 4 transports supporting 1394 anyway. So SPDIF seems to be the only way ahead.
Computer stereo or any other equipment with SMPS on board , should not be connected to a DAC like ours. This simply wont work properly, optical isolation is mandatory if that kind of connection is to be made, otherwise noise is going to find its way throug the analog section and at last end up in your speakers.
For that kind of use, more heavily filtratet gear is to be used.

Alternative I/V conversion does exist, but we will only settle for an active and discrete solution without any feedback - in other words a very low impedance input stage that does not complicate or deteriorate the signal path unnecessarily.
 
Cauhtemoc said:


The CS8416 might be the industri standard, but it is also quite old, and there are far better options out there. For the current state of the art you should look at the WM8804. The DIR9001 is also a good choice, although it is limited to 96 kHz.

Your choice not to use a current output DAC such as the PCM1794 is one I can understand, however I disagree with your reasoning as to why the CS4398 is the best voltage output DAC. I do not want to turn this into a measurment vs sound quality discussion, but I believe the CS4398 does not sound as good as the WM8740 or AK4396, despite having somewhat lower distortion. I also believe that most people will agree with me here.

But still, good job. :)


I am aware of the DIR's from Wolfson and TI, but I do not agree that they are better.
They both have better jitter spec's, at least in the datasheet. 100ps where the CS has 150ps max.
But in real life, jitter performance is also a matter of a well designed PCB. And in pratice I don't think you wil achieve better jitter performance than we do on the CS8416.
On the other side, the Crystal version is and has always been ultimately stabil. Look back in time, and you will find that at least the old TI DIR's was not. CS8416 will lock on just about anything.

About the WM8740 and AK4396, you may be right... But still. How did you test this?? Most people test different DAC's by placing them in the same setup, using the same analog stage etc. And that would be fine... Or that's what you think! But the problem is (And that's what often goes wrong). The circuit that works well on one DAC, will not automatically work well on the other. In other words: Comparing DAC-chips this way, will mostly give you an idea of how the DAC-chips works with analog stage and PSU you use.
Try to evaluate a set of F1 wheels against a pair of original VW Golf wheels, on a VW Golf. I guess the original will be the best solution... But that does only refelct the fact, that the F1 wheels does not fit the VW Golf.

We have found that the CS4398 is quite picky, when it comes analog stage and PSU. But when you take your time, it willgive you extreme performance.

If you want optimum performance in Audio, you have to realize, that every part must be carefully matched to the actual application. And that's what took us 5 years!

A good examble is the capacitors we use in the power supply and for decoupling in the digital stage. These were evaluated about 2 years ago, and just did not work! They sounded terrible compared to others. 3 months ago and after a lot of changes, we tried these capacitors again, and found them to be the absolute best performer in the new version of the DAC.
 
analog_sa said:



But I/V conversion does not equate opamps. There are so many other choices.


Why would diyers be interested in buying a complete dac though? You stand a better chance marketing it to a different crowd. And you may be ten years too late. How many would seriously consider buying a dac with only an spdif input interface?

Right you are... There are many other ways to do I/V conversion with no op-amp. I believe the best way is the commen base coupled transistor, since it has a very high frequence range and low input impedance. We have tested using different versions of this with different DAC's (Like the top specified PCM1794A), but did not get the same performance as with the CS4398.
 
Re: Impressing Work - schematics

suchtgutenklang said:
Very impressing result;
to get an impression how the machine works, how it handles time and finally how it may sound: will you share the schematics?

Maybe it would be interesting to compare your work on the CS4398 in its concept to that one: http://hoer-wege.de/dac4398.htm

for sure the ladder is a fully commercial project of an experienced DAC-Designer; but often the little punch of enthusiasm comes to a better result, because there was maybe not the necessity to get early bucks for the work.

In such a comparison your concept may build an interesting contrast to the well known OP-Amps driven analogue stage....

....no OP-Amps sounds quite interesting!

So are you willing to share your schematics?

so far
suchtgutenklang

Yeah, it could be fun comparing with the "Hoerwege" CS4398 DAC, using a op-amp followed by another op-amp. My guess is, that it will sound like: An op-amp ;)

Funny you mention the Hoerwege as a DAC made by an experienced DAC-designer :D . It would say, that it doesn't take a experienced designer to make an op-amp based design. It's almost like following the evaluation-board from Crystal... As I see it, using op-amps is the easy way out, when you do not want to spend time making a discrete design. Using op-amps you never fail... But also... You will never really succed. It's like eating Meat Loaf. Not bad at all... But also not that good.

By the way.... I have a bit of experience in designing myself. I specialised in audio DAC's when graduating as an electronics engineer. I have been working on DAC's for more than 10 years, and has been working as a designer in the danish audio business. Don't know if that counts :confused:
 
Algar_emi said:
Hi Kurt von Kubik. What are the part number, manufacturer and distributor and these nice PC mount RCA connectors? Thanks...

You will not be able to buy from this manufacturer as a private person. And anyway, the MOQ is 1000USD.
So if you only need a few, it's not possible.

I have some i stock, and if you need some, just say the word :cool:
 
hifimaker said:


I agree with the above quote.

Also, the cs4398 is not as detailed and much more relaxed than the AKM4396. The AKM4396 is an outstanding voltage out chip. This is my experience working with the respective company's eval boards. The cs4398 is clean and clear sounding, but very mellow and relaxed. It's the least musically involving chip sigma-delta chip I've heard.


As mentioned before, this very much depends on the actual application. I guess you will be surpriced of the musicality, details and easy of sound from our implementation of the CS4398.

hifimaker said:

AKM is also a easier company to do business with as well, from my experiences.

-David


My guess is, that this is why DIY-people like "the sound" of AKM-chips ;)
 
USB support

We have been talking about USB, since this seems to be very hot these days. But we decided to say NO WAY!!

USB is just about the least suited interface for digital audio, since it will transmit in packages = jitter!
Also as KvK mentioned, connecting a pc electrically to this DAC, will be very bad. The pc will conduct extreme amounts of noise into the circuits. This is also the case in known USB supporting DAC's, but normally these use op-amps. And then the sound may very well be to compressed and free of music, to hear the draw backs of connecting to the pc.

We have discussed hove to solve this, and we agreed that "We do not want any pc interface in the DAC".
The solution may be to develop an external USB to SPDIF converter, with built in jitter reduction and optical isolation. This way we can keep the noisy pc away from the DAC circuits.
 
hifimaker said:


I agree with the above quote.

Also, the cs4398 is not as detailed and much more relaxed than the AKM4396. The AKM4396 is an outstanding voltage out chip. This is my experience working with the respective company's eval boards. The cs4398 is clean and clear sounding, but very mellow and relaxed. It's the least musically involving chip sigma-delta chip I've heard.

AKM is also a easier company to do business with as well, from my experiences.

-David
I will not doubt any of your observations, but one implementation of i.e WM8740 I´ve heard is the Rega Apollo, which I find a lot more mellow and overrelaxed as I do the 4398 chip.
I think it has a lot to do with implementation.
With different supplies and surrounding circuitry we´ve had the 4398 both singing and snorring :D
I don´t think it would be very different with other makes.
 
Hurtig said:
I am aware of the DIR's from Wolfson and TI, but I do not agree that they are better.
They both have better jitter spec's, at least in the datasheet. 100ps where the CS has 150ps max.
But in real life, jitter performance is also a matter of a well designed PCB. And in pratice I don't think you wil achieve better jitter performance than we do on the CS8416.
On the other side, the Crystal version is and has always been ultimately stabil. Look back in time, and you will find that at least the old TI DIR's was not. CS8416 will lock on just about anything.

Please read the following presentations and perhaps you will see why the WM8804 is so clearly superior to the CS8416.

http://www.wolfsonmicro.com/uploads/documents/en/SPDIF_Paper_v1.2 May 2007.pdf
http://www.wolfsonmicro.com/uploads/documents/en/A high performance SPDIF receiver_Oct 2006.pdf

I have been told that the competitive part in this paper is the CS8416. And yes, I remember the issues TI had with their earlier receivers.

Hurtig said:
About the WM8740 and AK4396, you may be right... But still. How did you test this?? Most people test different DAC's by placing them in the same setup, using the same analog stage etc. And that would be fine... Or that's what you think! But the problem is (And that's what often goes wrong). The circuit that works well on one DAC, will not automatically work well on the other. In other words: Comparing DAC-chips this way, will mostly give you an idea of how the DAC-chips works with analog stage and PSU you use.
Try to evaluate a set of F1 wheels against a pair of original VW Golf wheels, on a VW Golf. I guess the original will be the best solution... But that does only refelct the fact, that the F1 wheels does not fit the VW Golf.

I think the question is, have you tried the WM8740 or AK4396? You have so far done exceptional work with the CS4398, and you have probably reached very close to its maximum potential (except for the receiver, see above). Now you need to move on and try other DACs.

Just so this post doesn't come out all negative, the CS4398 is by no means a slouch, and you have done an exceptional job. Keep it up. :)
 
Re: volume control of the rotary encoder

Danilo Almeida said:
hello friends .. sorry bothered you.
I am from Brazil and here is the end of the world when it comes to rotary encoders and LCD display.
ae I seek you aqui.desculpa the mistakes most do not speak the language of you turn to me here with a translator online.

wanted a project to control audio
which has a display on LCD display and volume control of the rotary encoder

ae someone have any schema.
and the code of the microcontroller for me.
may be a link.


Danilo,

Are you looking for a control for a preamp for volume and source selection?

Perhaps we can help with more information.

-David
 
Cauhtemoc said:


Please read the following presentations and perhaps you will see why the WM8804 is so clearly superior to the CS8416.

http://www.wolfsonmicro.com/uploads/documents/en/SPDIF_Paper_v1.2 May 2007.pdf
http://www.wolfsonmicro.com/uploads/documents/en/A high performance SPDIF receiver_Oct 2006.pdf

I have been told that the competitive part in this paper is the CS8416. And yes, I remember the issues TI had with their earlier receivers.



I think the question is, have you tried the WM8740 or AK4396? You have so far done exceptional work with the CS4398, and you have probably reached very close to its maximum potential (except for the receiver, see above). Now you need to move on and try other DACs.

Just so this post doesn't come out all negative, the CS4398 is by no means a slouch, and you have done an exceptional job. Keep it up. :)


We clearly have a different approach to the art of developing High End Audio gear.
If you read the documentation on a typical op-amp, it is clear to most people, that it is way better than our discrete non-feedback analog desing. But if you listen, the world is somehow different.

I have not tested Wolfson DIR, since I have very good experience with the CS8416.
My best guess is, that if you ask Crystal, you will get a quite different story, than if you ask Wolfson. I do not argue the fact, that the Wolfson have better jitter specifications. But what are they in real world applications?

About the DAC, CS4398. You state very clearly, that now we have to try another DAC-chip. But why?? We have achieved a performance with the CS4398, that is far ahead from everything we have ever heard before. And as mentioned several times: The DAC chip is not the only place you achieve the performance. Experiments has shown, that the analog stage and power supplys are much more important.

Anyway, I will contact Wolfson, and ask for some samples for evaluation.
Which Wolfson DAC is the best these days?? WM8740, WM8741 or WM8742??
 
hifimaker hello friend ..

hifimaker hello friend .. this is exactly what I try .. a sketch of a project to control the volume with digital LCD and rotary encoder can someone pass me a diagram or a link where I can find e bothered me stop you is that aki in brasil ta hard to find any such scheme.
excuse the errors more to using a translator to communicate with you
 
what is the intent of this discussion? schematics, schematics, schematics

Dear Hurtig and Kurt von Kubik,

sorry but I have to ask for another time:
please share your technical approach not only with describing words but with schematics or clearifying photos...to get transparency into the whole thing...

you may stress for hundred times again and again that the miracle of good sound is at least the (your) optimization of each step within the relationship of all the parts to each other....without information this is 'idle talk'...

...your work seems to be impressive enough for all of us ...so that we keep being courious onto your approach....but without a real basis for the discussion.... it seriously do not help furthermore!

by the way: I think a lot of the people in this forum are aware of the point, that the optimization/adaption of the parts to each other is the most important thing in creating 'High End'...and also the more sophisticated massproducing companies are even aware of this point: just the Marantz devellopment/producing of the CD 7 in 1998/1999 showed to the whole 'stunning' world that you can get outstanding results even out of 12 years old dacs (1541A emerged around 1987 I guess) and this was in that time absulteley not 'up to date'--they did something that was absoluteley unexpected--- but they did it...

...in every case the point is to invest the utmost passion onto your technical approach...


...so that is not the point...!!

...the point is: schematics , schematics, schematics (at least good pictures)

...or how did you produce/work out your professionally looking PCB ???

so far
suchtgutenklang
 
Status
Not open for further replies.