The Best DAC is no DAC

"If you find this thread silly please feel free to ignore it rather than continuing to add condescending comments. No please I really mean it - ignore this thread."

well i know i'm just a sunday morning diyer, but for me (DSD) is the holy grail of music reproduction (what sony and philips should have introduce in 1982).
So DF be a good catholic chap and start sharing this PHD knowledge of yours on how to precisely implement this smooth cut off at +- 19 khz with tube, output transformer and fancy opamp

hope i'm not asking to much, as you said it's silly, so DF it should be a walk in the park for you

Martin Gagnon

thank you for the DSD explanation

and thank you for everybody else, for putting their grain of salt in this marvellous thread.
 
... You don't need a DAC to convert DSD to analogue, as it was never digital to start with but merely uses pulse density modulation to carry the analogue signal.

Actually, that is not quite right as the original source was digital, so the real DAC is the circuit which converts to DSD. It's just that the signal analogue used is not voltage but pulse density so people think it is still digital.

Finally!

PDM carries Va !

What is the algorithm? [not circuit]



"...the transformers are the DAC..", "...the flip-flop is the DAC...", "the filter is the DAC...", ...

Pfft! :rolleyes:
 
Last edited:
Administrator
Joined 2004
Paid Member
This all seems rather Class-D to me. Just duty cycle modulation of a square wave. Low pass filter it to get the audio signal. The audio signal is already there, you just want to get rid of the RF.

This puts an interesting spin on the Power DAC. If you could get the power output stage to switch fast enough, you could use the DSD signal to run the amplifier output stage. Sort of a Class-D direct.
 
This all seems rather Class-D to me. Just duty cycle modulation of a square wave. Low pass filter it to get the audio signal. The audio signal is already there, you just want to get rid of the RF.

This puts an interesting spin on the Power DAC. If you could get the power output stage to switch fast enough, you could use the DSD signal to run the amplifier output stage. Sort of a Class-D direct.

Just what i was thinking.;)
 
I am not clear how you intend to separate the left channel from the right, or do you plan to leave this in mono.

The DIYINHK USB has the standard I2S outputs - data, LRCK (left right clock) and bit clock. But when playing DSD the e functions of the first 2 pins are reassigned. Data = DSD right channel. LRCK = DSD left cHannel. DSD doesnt need a word clock so its already demuxed. This is shown on my circuit diagram In earlier post.
 
Just what i was thinking.;)

I think everyone's on board this one but the challenges of good Class D sound reproduction are more difficult at higher power levels - perhaps what differentiates the approach in this thread is that the conversion to DSD is done at low power where common Class D issues (dead-time ?) are much easier to deal with and essentiallhy become a non-issue ?????


p.s. I think this is a great thread too.
 
In my understanding the main reason power amps aren't just big DSD DACs is because of the number of switching transitions (well rather the frequency of those switching transitions). A classD (PWM amp) might have a 300-400kHz switching carrier typically whereas DSD64 has in effect a 1.4MHz carrier. DSD128 would double that.

With the newer devices becoming available (GaN particularly) there might be fewer disadvantages in building a 'DSD-direct' poweramp as GaN has lower switching losses meaning higher output switching rates become attractive.

<edit> I forgot that there's also the thorny issue of how to do feedback with DSD. PWM amps are linearized in practice by feedback but I'm not sure how feedback could be incorporated into a 'DSD-direct' amp.
 
Last edited:
In my understanding the main reason power amps aren't just big DSD DACs is because of the number of switching transitions (well rather the frequency of those switching transitions). A classD (PWM amp) might have a 300-400kHz switching carrier typically whereas DSD64 has in effect a 1.4MHz carrier. DSD128 would double that.

With the newer devices becoming available (GaN particularly) there might be fewer disadvantages in building a 'DSD-direct' poweramp as GaN has lower switching losses meaning higher output switching rates become attractive.

<edit> I forgot that there's also the thorny issue of how to do feedback with DSD. PWM amps are linearized in practice by feedback but I'm not sure how feedback could be incorporated into a 'DSD-direct' amp.
A DSD amplifier would be "odd." Yes, the switching frequency is pretty high for power devices.

Changing the volume would require varying the rail voltages. There are ways that can be done, but it's unconventional.

There would be no feedback and it would be "what you hear is what you get," whatever that is. But it might work better than PWM with feedback.

I'm thinking an alternative method would be using a matched current source and sink. These might be switchable more easily than switching voltage sources. Volume would be changed by changing the output current(s).
 
hazard500, have you tried grounding the electrostatic shield of your Lundahl transformers? This will prevent capacitive coupling of primaries to secondaries, so could/should significantly reduce the unwanted HF present on the output. Less to tidy up after that.
Good point. I haven't but I should. If I get time on the weekend I want to put in a proper PS so will attend to it then.
 
Very interesting, Ray. Since you first started talking about Linux I have been reading up on it. I found this interesting distro

AudioPhile Linux | Quality audio on Linux

Looks similar to audiolinux - but its free to download. So definitely worth installing this and checking it out.
OK I have now installed Linux (I installed the OS on a dedicated SSD to avoid any issues with dual boot). So far i would say:
1 DSD files played on Linux, through my USB board sound almost analog. Sound stage is very 3D. I don't think I've heard digital sound so good.
2 Linux is a real pain, if you are used to Windows and everything is drop down menus and drag and drop. You have to go into The Terminal (similar to old DOS command line stuff) to make changes. Not very user friendly. But the sound makes it worthwhile.
3 I can't figure out to convert PCM to DSD on the fly (FooBar can do this in Windows). So I have a lot of hi res files from my needle drops and stuff I've bought from HDTracks, which I can't play through Linux as yet. Hopefully I can work this out.
 
OK, I have an old laptop I can use to try out this direct dsd "D-DSD". Which version of Linux and where do I download it from ? (never used it before)
I have installed AudioPhile Linux.

AudioPhile Linux | Quality audio on Linux

DSD is sounding very nice. I've left a note with the programmer asking for advice on converting PCM to DSD on the fly (I've ripped al my CDs to hard drive, and I've also bought a number of hi res PCM files from HDTracks). If you've never used Linux before - be prepared for a culture shock. Some things look same as Windows, some things are very different. I'm pulling my hair out trying to work it out. But, the guy who runs the web site has been very quick answering a few questions I have left for him. THe software is free, and he gives free support. You can't complain about that!!
 

Attachments

  • image.jpg
    image.jpg
    247.5 KB · Views: 1,244
Last edited:

It's quite impressive what this guy is doing - makes me curious where is the reward, how does he plan to make money for all the time and effort ?

My worry is accessing my music. Most of my music, actually all of it, is in iTunes. Most of my CDs are in the process of being ripped to Apple Lossless. I'm wondering if I'm not better to keep with Windows on the laptop, install iTunes on it so I can access my iTunes library over wi-fi. I assume I would then be unable to run the AudioPhile Linux and instead I need something running under Windows that will work with my iTunes files to convert the PCM files to DSD ???? (I may know transistors and triodes but I'm a bit thick with computers)


A commercial link.
the wiring/construction looks a bit DIY to me
 
Last edited: