But the PCM1794A is not a complete DAC - so did you add some analog parts beyond it? Those will be the main determiner of the sound, not the DAC itself. Ayre QB-9 allegedly gets good results with it, it appears to be using passive I/V.
Sounds plausible until one notes that no single part can be listened to, we can only hear complete systems.
ThorstenL already hinted at the problem on this very thread, around post #273 - noise modulation.
Yes, ThorstenL spend a long time talking bad about the CS4398. Seems he has got some kind of hidden agenda
But the thing is, that NOBODY had any arguments based on sonic performance. Every single time someone talked bad about the CS4398, they refer to some technical "issue". I'm an engineer to, and can also read the datasheets. The nice specifications on the PCM1794A was one of the reasons, that we decided to try it. But it failed when we powered on the ultimate measurement tool.
And that may very well be the thing that we do different. It is very clear, that you guys talking bad about the CS4398, mainly pick your DAC-chip based on the tech specs....
Some search sonic performance, others search for best measurements.
BTW... will you please inform all the tube-guys, that they must immediately trash all their tube-gear, because op-amp based designs has much better THD performance. Thanks in advance...
Yes, ThorstenL spend a long time talking bad about the CS4398. Seems he has got some kind of hidden agenda
Doesn't seem so to me - he talked bad about CS because he had bad listening experience with those products. What's unreasonable there?
Oh its quite obvious you did not listen to what ThorstenL said. Which makes me not trust your ears (to be more accurate, your perception which is brain/ears) when listening to DACs or anything else. He both said they sounded bad to him AND gave technical reasons for why.
Speaking of failure, this is not the first time you've failed to answer - what I/V conversion and analog stage were you using for this listening?
I'm sure you'd like to think that you're the only guys who listen rather than examine datasheets.
Who am I to break the spell?
If its clear then show me my postings which demonstrate my 'lack of genuine interest'? I am most certainly interested in an open source DAC and plan to make one of my own.
I've asked a few questions - please explain to me how questioning becomes 'attacks' and/or 'battering' ?
Infraction points are for what? Is there an FAQ I can go to to find out how to avoid infraction points in the future? At present I'm unclear how I've incurred them.
@ abraxalito:
I suggest that you will either accept the fact, that this DAC will be based on a CS4398.
Alternatively, you are free to share your own design in here.
--------
I did consider making an alternative DAC, using a low cost Wolfson WM8740 chip. This will save around 10USD on the BOM cost compared to the CS4398. But this is not yet decided.
I suggest that you will either accept the fact, that this DAC will be based on a CS4398.
Alternatively, you are free to share your own design in here.
--------
I did consider making an alternative DAC, using a low cost Wolfson WM8740 chip. This will save around 10USD on the BOM cost compared to the CS4398. But this is not yet decided.
I only discovered this thread yesterday, but have read through all of it. Have to say it's been a frustrating read as certain people seem to have a VERY condecending 'I know better than you' attitude towards this project and its goals. I won't name names but I think a certain poster here has serious ego problems...
Anyways, I applaud the use of the CS4398 chip. I couldn't care LESS about the measurements (beyond a certain point...) as it all gets into the realm of diminishing returns anyway (i.e. you can measure something, but can you hear it, and - of course - vice versa) in terms of specs between different chips. I have been using the Gigaworks 24/192 CS4398 DAC with beefed up PSU and UTC A-20 output transformers and have been extremely happy with this combination for the past 6 months. The size of the Gigaworks thread is testamont to the fact that MANY love the sound of this chip.
I'm all for educated input to make an open source project as good as it can possibly be, but certain people want to stamp their influence all over it and - really - turn it into what THEY would make.
Rant over, and I look forward to seeing what the outcome of this project is - I'd certainly be up for purchasing a board as/when they become available.
- John
Anyways, I applaud the use of the CS4398 chip. I couldn't care LESS about the measurements (beyond a certain point...) as it all gets into the realm of diminishing returns anyway (i.e. you can measure something, but can you hear it, and - of course - vice versa) in terms of specs between different chips. I have been using the Gigaworks 24/192 CS4398 DAC with beefed up PSU and UTC A-20 output transformers and have been extremely happy with this combination for the past 6 months. The size of the Gigaworks thread is testamont to the fact that MANY love the sound of this chip.
I'm all for educated input to make an open source project as good as it can possibly be, but certain people want to stamp their influence all over it and - really - turn it into what THEY would make.
Rant over, and I look forward to seeing what the outcome of this project is - I'd certainly be up for purchasing a board as/when they become available.
- John
Last edited:
yes, well the people running this thread (community project it is not) really have noone to blame but themselves for that IMO, many a thread has seen these 2 come in and trash the use of a component they may be using or whatever (especially chips), while making reference to how amazing their dac sounds. sometimes this was even in paid for threads for other products. this has been going on for years and still this dac has not seen the light of day, aside from a few fabled listening tests. I await the final release of this design, so that real world comparative listening tests from consumers/builders in their own homes can be undertaken and opinions both ways formed. there is zero possibility that this dac will be preferred by all, that is pure fantasy
You seem to have been labouring under a considerable misapprehension, so perhaps a little clarification is in order.
I've always accepted that its your project and therefore you get to make the choices. I'm totally fine with you choosing the CS part for your project. In any case, why would my lack of acceptance make any difference? I'm not your boss or mentor or anything like that.
My probing questions were because I'm curious to learn of the rationale behind your choices, not to dissuade you from making them yourself.
Any clearer now?
That is true, and has been my intention for some time.
@ abraxlito:
If you really did post all these "anti-CS4398"-posts because of your curiosity, you clearly have not yet read the thread.
As I have staed at least 10 times, the CS4398 was picked ONLY based on listening tests. I would love to use a PCM1794A because of it's better tech specs... If only it would sound as good as the CS4398. But it did not. And again... that really is the difference between this and other projects. We didn't just pick the DAC with the best specs. We made up PCB's for all of the relevant chips, and tested them with different analog stages.
The PCM1794A was tested with both op-amp based and fully discrete non-NFB analog stage (Also discrete I/V). Just like all the other chips we tested.
Anything else your curiosity needs to be updated on???
If you really did post all these "anti-CS4398"-posts because of your curiosity, you clearly have not yet read the thread.
As I have staed at least 10 times, the CS4398 was picked ONLY based on listening tests. I would love to use a PCM1794A because of it's better tech specs... If only it would sound as good as the CS4398. But it did not. And again... that really is the difference between this and other projects. We didn't just pick the DAC with the best specs. We made up PCB's for all of the relevant chips, and tested them with different analog stages.
The PCM1794A was tested with both op-amp based and fully discrete non-NFB analog stage (Also discrete I/V). Just like all the other chips we tested.
Anything else your curiosity needs to be updated on???
Oh that's good then because I have read the thread and I only post up questions because of my curiosity. Sometimes you say things which are in error, like now, so I post to correct your errors.
Yes, I noted that the first time you said it. For me, the sound of a DAC is at least 90% in the implementation, and nobody can listen to a DAC chip absent its implementation. That's why I've been asking you questions about implementation. I have heard the CS4398 but I pretty much discount what I heard because the implementation isn't optimum in the unit I listened to.
I pointed out earlier that you cannot have used the 1794A in isolation, without other circuitry to complete it.
Ah, now you're providing some really useful information. So when you listened to the PCM1794A, you had the chip itself on one PCB and the analog stage on another? Where was the I/V stage - together with the chip or in the analog stage?
So how about posting up the schematic for the opamp based I/V on the PCM1794. Oh and the PCB too, that would be helpful.
Curiosity is not a need. But sure, just keep answering the questions and we'll get somewhere...
@ abraxlito:
We used an app.note style op-amp analog stage. Check the datasheet, and you will find it.
We also used a pure discrete non-NFB analog stage. In this, the I/V was made of a common base bipolar transistor. We also tried a version with the single transistor replaced by a CFP-pair to lower the input resistance.
But none of the compared with our CS4398 version
We made special PCB for all of the DAC chips used, with the analog stage on-board. This means a lot of wasted PCB's!
I don't really understand why you want to keep discussion the DAC choice. But I will check if I still have some of the PCB's in my basement. I'm not sure, since I did some cleaning up. If I still have some, you can have them for free, since they represent no value for me.
This discussion about the DAC chip, has now ended from my side.... This thread is about a DAC using CS4398, and most likely that will not change.
If you want to discuss PCM1794A or other DAC chips, I suggest that you start your own thread. I'm am sure that using your energy in your own thread/DAC design, will be vefry interesting. So please post a link when you do.
We used an app.note style op-amp analog stage. Check the datasheet, and you will find it.
We also used a pure discrete non-NFB analog stage. In this, the I/V was made of a common base bipolar transistor. We also tried a version with the single transistor replaced by a CFP-pair to lower the input resistance.
But none of the compared with our CS4398 version
We made special PCB for all of the DAC chips used, with the analog stage on-board. This means a lot of wasted PCB's!
I don't really understand why you want to keep discussion the DAC choice. But I will check if I still have some of the PCB's in my basement. I'm not sure, since I did some cleaning up. If I still have some, you can have them for free, since they represent no value for me.
This discussion about the DAC chip, has now ended from my side.... This thread is about a DAC using CS4398, and most likely that will not change.
If you want to discuss PCM1794A or other DAC chips, I suggest that you start your own thread. I'm am sure that using your energy in your own thread/DAC design, will be vefry interesting. So please post a link when you do.
Ah, that explains it totally. Now I can understand why the CS sounded better than the PCM. I wonder how many other design wins TI has lost through that recommended circuit. I found that the NE5534 simply isn't fast enough in that particular application, and the relatively huge value of feedback cap squashes the sound. It can be improved quite a bit by reducing that cap but ultimately the 5534 has to go and be replaced by something with video credentials. After all, what comes out of my AD1955 current outputs is far wider band than a baseband video signal. I did a little video work in my former days...
Cool - thanks for sharing the details and indulging my curiosity. I haven't played with discrete op stages yet myself seeing as my first port of call was opamps and I did eventually get a sound I really liked with an LM6172.
Yeah, sounds like you really worked so hard - but stuffed it up by building the TI app circuit... Can't win them all eh?
I was curious simply because you felt the PCM1794 was so inferior to the CS. But now you've answered, I'm pretty much done on this line of questioning.
Its really a kind offer, thanks but as I'm working on my own AD1955 design I doubt I'd be inclined to play with them. So don't spend any time digging around on my account.
Thanks for your support - its taking a long time because I'm really not satisfied with a standalone DAC - cabling is a significant performance issue, so I'm working on a system. That's a really long process.
@ novec:
I have attached the schematic of the discrete analog stage we have been using with the PCM1794A. This is the "CFP" version.
Some of the resistor values may not be correct, since we made a lot of "tweaking" on this particular design, in hope of succes with the PCM1794A. But until now, we did not manage to bring it to the same level as the CS4398.
Be aware, that the PSU regulators is not shown. Also there is no DC-servo.
We know from experience, that it takes a long time to implement a DC-servo that is not in any way audible. That's why we do not implement the servo this early in the process. We use DC adjustment in prototypes.
So if you want to use the schematic, you must implement a proper servo...
I have attached the schematic of the discrete analog stage we have been using with the PCM1794A. This is the "CFP" version.
Some of the resistor values may not be correct, since we made a lot of "tweaking" on this particular design, in hope of succes with the PCM1794A. But until now, we did not manage to bring it to the same level as the CS4398.
Be aware, that the PSU regulators is not shown. Also there is no DC-servo.
We know from experience, that it takes a long time to implement a DC-servo that is not in any way audible. That's why we do not implement the servo this early in the process. We use DC adjustment in prototypes.
So if you want to use the schematic, you must implement a proper servo...
I'm fairly sure it has more potential - plenty of ability to surpass the CS. If I can help you in any way, PM me. If you'd like me to check layouts or the like, I'll be glad to help.
I´ve been part of these experiments and can confirm, that it seems as if it is somewhat more difficult to get the full potential from the BB 1794A, than first expected.
We´ve build a passive I/V, but so far with nogo. But I think there might be some kind of secret to reveal to get the full blown BB sound.
But this also led to thoughts about the need for even better specs. Both the CS 4398, the AD1955 and the BB 1794A does by far exceed the most important data of our analog stage.
So the point is, if it does have any relevance at all. To me the upside of the BB 1794A, beside its nice specs overall, is its freedom of noise, only needing a reconstruction filter.
The downside is the I out design, needing an I/V stage.
In the other hand, the CS4398 does not need an I/V stage, but some filtration instead, but still the filtration needed is a very mild one. We did 2. order @ 100KHz with good results.
So this leaves us talking about THD lower than 100 dB and an even greater dynamic range, and at the same time, we cannot get even close to that in the following stage, unless we use turbo racer op-amps like AD 797 or like.
And that is no longer relevant for this project, I´m afraid.
We did do experiments with amongst a lot of others the AD 844 op-amp, but with nogo at all, and we did that many times, because we actually could save almost 70 components in the analog stage, making it a lot more easy to assemble, and we really would like that. Besides that, op-amps performs superior to anything regarding THD and PSRR, making them easy to create safe and sound surroundings to.
But by God, I do not like the sound of them.
We´ve build a passive I/V, but so far with nogo. But I think there might be some kind of secret to reveal to get the full blown BB sound.
But this also led to thoughts about the need for even better specs. Both the CS 4398, the AD1955 and the BB 1794A does by far exceed the most important data of our analog stage.
So the point is, if it does have any relevance at all. To me the upside of the BB 1794A, beside its nice specs overall, is its freedom of noise, only needing a reconstruction filter.
The downside is the I out design, needing an I/V stage.
In the other hand, the CS4398 does not need an I/V stage, but some filtration instead, but still the filtration needed is a very mild one. We did 2. order @ 100KHz with good results.
So this leaves us talking about THD lower than 100 dB and an even greater dynamic range, and at the same time, we cannot get even close to that in the following stage, unless we use turbo racer op-amps like AD 797 or like.
And that is no longer relevant for this project, I´m afraid.
We did do experiments with amongst a lot of others the AD 844 op-amp, but with nogo at all, and we did that many times, because we actually could save almost 70 components in the analog stage, making it a lot more easy to assemble, and we really would like that. Besides that, op-amps performs superior to anything regarding THD and PSRR, making them easy to create safe and sound surroundings to.
But by God, I do not like the sound of them.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.