Hello,
I went the easy and the best way!
If you preamp cannot cope with DC on its input you will need a Sowter.
To use a Sowter you need 4 boards or more.
I think a GOOD transformer wont degrade the signal.
I/V R is easy. You can kind of tune the sound by using a different resistor.
I am sure Doede checked other possibilities during fine tuning.
When i started with my DDDAC i chose to use chokes right away. Now Doede has done some test and it is up to you if you spend some money on a choke. My idea is that a choke will usually give less of an '' electronic footprint '' like most active elements do. I like the idea that it works now and it will do the same for decades to come!
Greetings, Eduard
I went the easy and the best way!
If you preamp cannot cope with DC on its input you will need a Sowter.
To use a Sowter you need 4 boards or more.
I think a GOOD transformer wont degrade the signal.
I/V R is easy. You can kind of tune the sound by using a different resistor.
I am sure Doede checked other possibilities during fine tuning.
When i started with my DDDAC i chose to use chokes right away. Now Doede has done some test and it is up to you if you spend some money on a choke. My idea is that a choke will usually give less of an '' electronic footprint '' like most active elements do. I like the idea that it works now and it will do the same for decades to come!
Greetings, Eduard
I wrote down the schematic for such type of experiment several times, but after that each time I calculated the price of this experiment. Too expensive for experiment with unpredictable results. A lot of top-grade parts are needed. In any other case it's not clear what we are really comparing.
Which is true Yuri, it might be the power supply for example....
The only thing you could do to find a character, is to add a certain Opamp buffer topology behind the normal DDDAC I/V output and than re-wire it for Opamp I/V and listen what the difference in character is.
Well, lot's of work and still I feel not compelled to do it ( = I do not think it will bring anything)
The only thing you could do to find a character, is to add a certain Opamp buffer topology behind the normal DDDAC I/V output and than re-wire it for Opamp I/V and listen what the difference in character is.
Well, lot's of work and still I feel not compelled to do it ( = I do not think it will bring anything)
Doede, I completely agree with you.
Idea of zero impedance vs 133R for current output DAC looks great before the moment when the schematic was drawn with supposed sources of distortion, and internal schematic of OPAMP in hand. A lot of Rhopoint resistors, hq polystyrene/teflon capacitors, several complex "ideal PSU" and so on. Complexity increased by a factor of 100. And all these things on the assumption that close to ideal OPAMP can be found.
Well, I think that at least with discrete OPAMPs and all the things above, the same results as with 4 resistors are achievable. So, only a few thousand EURO to achieve the same results And we really know what the same money spent on increasing the DAC stack will definitely increase the sound quality. In my point of view, it's a very interesting idea, it can help to understand a lot of things on paper, but for me it looks slightly insane
Idea of zero impedance vs 133R for current output DAC looks great before the moment when the schematic was drawn with supposed sources of distortion, and internal schematic of OPAMP in hand. A lot of Rhopoint resistors, hq polystyrene/teflon capacitors, several complex "ideal PSU" and so on. Complexity increased by a factor of 100. And all these things on the assumption that close to ideal OPAMP can be found.
Well, I think that at least with discrete OPAMPs and all the things above, the same results as with 4 resistors are achievable. So, only a few thousand EURO to achieve the same results
And we really know what the same money spent on increasing the DAC stack will definitely increase the sound quality. In my point of view, it's a very interesting idea, it can help to understand a lot of things on paper, but for me it looks slightly insaneHello,
Doede designed his DDDAC to get one that would have a '' sonic signature '' close to a turntable.
I think adding an opamp will always mess up the original intentions. If you want optimum results from it you need a nice power supply. There are just to many things to take care of.
Better spend time and money to get 4 resistors to get optimum results.
Greetings, Eduard
Doede designed his DDDAC to get one that would have a '' sonic signature '' close to a turntable.
I think adding an opamp will always mess up the original intentions. If you want optimum results from it you need a nice power supply. There are just to many things to take care of.
Better spend time and money to get 4 resistors to get optimum results.
Greetings, Eduard
I have been following the development of the DAC on Doede's website. I have a basic question because I am considering either building a NOS DAC (but balanced with 2x AD1865 or 4x PCM1704) or going the Soekris route. My question is actually about NOS itself. I can rationally see why someone would pursue NOS if they are using files that go 96kHz or higher. Then an analog output filter acting at half the frequency might be good. I myself have done experiments with Sox using DSF SACD files that have signal in the spectrum above 40kHz. An excellently effective "analog" lowpass filter with Sox should always be applied in the range of 30kHz and 12dB slope. A 6dB filter correspondingly before. However, the latter does not look so good. I haven't read this whole thread, but I don't know yet (there's nothing on Doede's page either) how you handle 44.1kHz files. Are they upsampled and filtered in the PC player beforehand? I can hardly imagine that a pure untouched 44.1 kHz recording can only sound competitive with an analog 6dB filter. The measurements - especially sinusoidal measurements - look extremely poor, to say the least. I'm also following with great interest the thread "Filter brewing for the Soekris R2R", where an optimization of digital filters is progressing. Because I'm not ideologically biased, I'd be very grateful for any pertinent advice. And/or also for circuit tips.
TJF, I see where you are coming from. On paper, NOS of a 44.1 kHz signal does look like it's a problem for audio. However, ears do not read papers.
I have tried upsampling 44.1 kHz audio with different methods, and then listen to the upsampled audio via my DDDAC. On paper, this makes a huge difference. However, listening to the original and the upsampled audio made a surprisingly small difference, and I actually preferred the original over the upsampled audio.
My conclusion is that upsampling and interpolation ("oversampling") may seem like a good idea on paper, but listening tests indicate something else. I guess this is just another example of where we tend to look a the wrong things when designing audio gear. It's a bit like a super-duper low-distortion amplifier does not necessarily sound better than a very simple amp with 0.1 % THD.
I have tried upsampling 44.1 kHz audio with different methods, and then listen to the upsampled audio via my DDDAC. On paper, this makes a huge difference. However, listening to the original and the upsampled audio made a surprisingly small difference, and I actually preferred the original over the upsampled audio.
My conclusion is that upsampling and interpolation ("oversampling") may seem like a good idea on paper, but listening tests indicate something else. I guess this is just another example of where we tend to look a the wrong things when designing audio gear. It's a bit like a super-duper low-distortion amplifier does not necessarily sound better than a very simple amp with 0.1 % THD.
Without papers no clocks would exist, which you are using. I think you are using the best clocks on paper
May be you are right. Some days ago, somebody told me the opposite (with similar words and a BuffaloDAC III).
Did you do upsampling with "analog" (sox?)filter?
In my long life so far I have made the experience that there is no contradiction between theory and practice if you are on the right track. But what is theoretically correct can only be the foundation on which the truly excellent can be built. By this I mean here also the excellent sounding result.
I guess I just have to build one ...
Thank you very much for your opinion!
Hi TJF,
your points are dangerously at the edge of the 20 years old discussion on NOS versus etc. etc. Also the types of DAC But it is a valid point and will never be debated till the end. As there is more to it than only topology:
My experience is also that Matthias (MWBRENNA) is 100% right and that the best sounding DACs are also about implementation and not what chip you use alone.
I would only add to this and then want to stop this discussion, that the tweeter in your speaker is also a 6-12 dB filter > 20kHz and together with your ears you are well served. Your brain does (surprisingly) the rest
Just be practical as you already suggest, just build something and see / listen for your self. And whenever on route or in the neighborhood of Wiesbaden, feel free to stop buy and compare what you have built with the DDDAC implementation of the PCM1794 No shoot out stuff, just listening and observing and gaining experience (for both of us)
.
your points are dangerously at the edge of the 20 years old discussion on NOS versus etc. etc. Also the types of DAC
But it is a valid point and will never be debated till the end. As there is more to it than only topology:My experience is also that Matthias (MWBRENNA) is 100% right and that the best sounding DACs are also about implementation and not what chip you use alone.
I would only add to this and then want to stop this discussion, that the tweeter in your speaker is also a 6-12 dB filter > 20kHz and together with your ears you are well served. Your brain does (surprisingly) the rest
Just be practical as you already suggest, just build something and see / listen for your self. And whenever on route or in the neighborhood of Wiesbaden, feel free to stop buy and compare what you have built with the DDDAC implementation of the PCM1794
No shoot out stuff, just listening and observing and gaining experience (for both of us) .
I prefer to listen at 352.8 or 384kHz. I don't make upsampled files, but I let the player do everything "on the fly". Maybe there are better ways, but I'm too lazy.
I'm just avoiding making 384kHz from 44.1kHz and that's it. I only use multiplication by 2x, 4x or 8x (88,2 or 176,4 or 352,8kHz from 44.1kHz).
I'm just avoiding making 384kHz from 44.1kHz and that's it. I only use multiplication by 2x, 4x or 8x (88,2 or 176,4 or 352,8kHz from 44.1kHz).
Last edited:
Sure! It does not matter if the upsampled data is save to a file or not.
Do you prefer the sound from the original or from the upsampled data?
We use close to ideal clocks in real devices tooI think you are using the best clocks on paper
Unfortunately, after experiments with resamplers, filters, and reading books in psychoacoustics, I end up with a tough decision: the only way to achieve better results is upgrade all personal collection to hi-res, any other solution is a wasted time and money. No resampler is the best resampler for CD quality records. I agree with Matthias, in the best case, resampling and filtering can give almost the same results. In most cases it’s worse. So, no gain, but increased complexity.
Well, there are some interesting new solutions on market with oversampling and digital filtering, say from Grimm audio|Tentlabs ($10k if I remember correctly). For most of the people it can dramatically improve 44.1k sound quality, but there are two reasons of it:
1. It uses a really good clock and reclocker. As it's two steps better than widely used here Aquasilicon + FIFO, even DDDAC users can achieve great improvements here. But if you already have the same quality in the DDDAC clock chain -- no magic, no improvements, of course.
2. 99.9% of the audio market uses configuration with complex digital and analog filtering. DAC IC is not powerful enough for complex calculations. Upgrading to one-step filtering with more complex algorithms and high precision calculations inside powerful FPGA can give improvements here for most DACs. Improvements for DDDAC users are questionable.
I made the analysis of this expensive solution to show one simple thing: if something brings improvements to other DACs (several solutions mentioned above in this thread), it does not mean that you can achieve improvements of DDDAC in the same way. There are a lot of details in implementation.
All I wrote here is about PCM, 64fs DSD have a different noise spectrum, depending on implementation, maybe resampling can really bring something here.
Off-topic. So, in my point of view the only possible reason to process something in FPGA (signal correction, not simple synchronization of clock domains like FIFO does) for PCM NOS DAC may be speakers/room correction. Matthias, your works are really interesting.
From the upsampled. Not a big difference, a little more fluid presentation.
Not sure what you are referring to here. I am not into room correction. Speakers are my thing though (see my signature and here).
Well I said there was little difference. It takes effort to spot the difference. I have ribbon tweeters maybe that's why upsampling sounds nicer. I don't bother with that, I can listen at any sampling frequency without fatigue.
Last edited:
I think I know the answer here - but just want to check....
I bought an AC coupled headphone amp (connected via XLR) and was hoping that connecting direct to the DDDAC with no output caps or transformers would work satisfactorily. However, trying with some cheap headphones there is a very annoying buzzing in the headphones. I thought it might be a ground loop, but the buzzing changes along with the volume and / or gain settings ie. it gets louder as you turn up the volume. I presume then that this means its not a ground issue? And I presume the best way to deal with this is introduce caps / transformer output on the XLRs?
I bought an AC coupled headphone amp (connected via XLR) and was hoping that connecting direct to the DDDAC with no output caps or transformers would work satisfactorily. However, trying with some cheap headphones there is a very annoying buzzing in the headphones. I thought it might be a ground loop, but the buzzing changes along with the volume and / or gain settings ie. it gets louder as you turn up the volume. I presume then that this means its not a ground issue? And I presume the best way to deal with this is introduce caps / transformer output on the XLRs?