• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Direct Drive DSD

Status
Not open for further replies.
Hi Acko, A couple of thoughts on the project, perhaps slightly provocative but not meant to slight in any way🙂.

Having read back through quite a bit of the thread is there a danger that the beautiful simplicity of where the project started is being lost?

And a suggestion for a slightly different approach, without compromising the work you've already done, that would allow simple or complex to suit personal requirements.

As you're already using a motherboard approach, how about separating the 'direct DAC' board into discrete modules that can be 'plugged' into the motherboard (or omitted as required). How about separate plugin modules for;

  • Digital volume
  • LP filter (maybe several varieties)
  • DC Offset Nulling
  • A variety of output buffers to suit different drive requirements
That would make the overall unit a little more complex but very flexible and would facilitate incremental developments.

Of course, I recognise that you the driver...

Ray
 
I made some final tweaks to my Debian Stretch NAA build today and I think I now have it behaving like a true 'black box' appliance; push the button to turn it on and it's ready to use in a couple of seconds then press the button to turn it off. No need for any commands, it has just worked over quite a number of cycles.

Ray
 
  • Digital volume
  • LP filter (maybe several varieties)
  • DC Offset Nulling
  • A variety of output buffers to suit different drive requirements

Can I add one more feature to the wishlist?

Variable digital delay for each channel!

Then we can all dump our Multi-channel DSP crossovers but save their most useful feature. It seems to me that upsampling everything to a chosen fixed frequency makes this sort of delay feasible.

Thoughts?
 
Hi Acko, A couple of thoughts on the project, perhaps slightly provocative but not meant to slight in any way🙂.

Having read back through quite a bit of the thread is there a danger that the beautiful simplicity of where the project started is being lost?

And a suggestion for a slightly different approach, without compromising the work you've already done, that would allow simple or complex to suit personal requirements.

As you're already using a motherboard approach, how about separating the 'direct DAC' board into discrete modules that can be 'plugged' into the motherboard (or omitted as required). How about separate plugin modules for;

  • Digital volume
  • LP filter (maybe several varieties)
  • DC Offset Nulling
  • A variety of output buffers to suit different drive requirements
That would make the overall unit a little more complex but very flexible and would facilitate incremental developments.

Of course, I recognise that you the driver...

Ray

Ray, thanks for the feedback and suggestions. Just to allay your concerns 🙂 there is no project creep as such. The original design is still the reference. What goes on the board or taken off depends on the outcome of the test results. Some functions are core requirements, others are glue logic for features, so you as can see the motherboard design appearing.
Ordinarily, this type of work will be done on breadboards but there are MHz switching for this case and therefore pcbs with HF layouts even for simple tests. In fact the basic modules shown need further HF layout updates for DSD256 @ 12MHz switching.

Then volume control, and this will be done on a separate module altogether. If it does not compare well then it will be discarded. Btw, the software volume control on HQPlayer is not up to scratch and best disabled (directSDM) and use external attenuator. There are no signs of CPU stress with Volume ON so must be implementation issue or settings. 9018 DAC does a better job here!

The rest, where appropriate, as external modules to suit.
 
Ray, thanks for the feedback and suggestions. Just to allay your concerns 🙂 there is no project creep as such. The original design is still the reference. What goes on the board or taken off depends on the outcome of the test results. Some functions are core requirements, others are glue logic for features, so you as can see the motherboard design appearing.

Ordinarily, this type of work will be done on breadboards but there are MHz switching for this case and therefore pcbs with HF layouts even for simple tests. In fact the basic modules shown need further HF layout updates for DSD256 @ 12MHz switching.



Then volume control, and this will be done on a separate module altogether. If it does not compare well then it will be discarded. Btw, the software volume control on HQPlayer is not up to scratch and best disabled (directSDM) and use external attenuator. There are no signs of CPU stress with Volume ON so must be implementation issue or settings. 9018 DAC does a better job here!



The rest, where appropriate, as external modules to suit.


Im not sure if it's the filtering or what with the HQplayer volume control. Compared to SDMdirect, your passing the signal through an extra layer of filtering. I just wish the volume control could be added without that extra layer of filtering. But I think some sort of filter/modulator is required. And may be impossible to make this as transparent at straight bypass.

BTW with further listening, DSD mode with Sonny's DAC is stunning. Comparing it with passing through the modulator section of the chip is night and day. The AK4490 is as close as it gets to chipless in DSD direct mode as a chip can be.
 
I've loaded Debian Stretch onto the eMMC of a Beaglebone Black so it shouldn't be a problem, especially if you just load the core. I can point you in the right direction but you'll need to see what architecture your device has to make sure you get the correct version. Remind me what an SBC is?

Ray


It's the NUC 3815. Very nice little board. Sounds great.

http://ark.intel.com/m/products/78577/Intel-NUC-Kit-DE3815TYKHE#@product/specifications

Driven by linear supply, it sounds much better than my Mac mini used as an NAA. And that's with Windows on SSD. I'm thinking with Linux running on the eMMC, it will be even better. Plus won't need the SSD.
 
Then volume control, and this will be done on a separate module altogether. If it does not compare well then it will be discarded. Btw, the software volume control on HQPlayer is not up to scratch and best disabled (directSDM) and use external attenuator. There are no signs of CPU stress with Volume ON so must be implementation issue or settings. 9018 DAC does a better job here!

For info;

HQ Player - Page 156

I'm gonna have to try the HQPlayer volume control.

Ray
 
I don't know about you but I can't get the same level of sound quality with SDMdirect disabled so the volume control can be used. It doesn't matter which settings it's on. There's an extra veil over the sound, and resolution is lost.

Same here, during early days of testing this DirectDrive DSD/HQPlayer without external attenuator. Thought this was even more direct... Almost gave up, getting less than favourable sound, until I added this simple ALPS pot, disabled HQPlayer Vol and then bingo!
 
Last edited:
Same here, during early days of testing this DirectDrive DSD/HQPlayer without external attenuator. Thought this was even more direct... Almost gave up, getting less than favourable sound, until I added this simple ALPS pot, disabled HQPlayer Vol and then bingo!


Upon further comparison of Audiolinux running the networkaudiod daemon in command line mode in both DoP and native DSD with my fine tuned Windows 8 embedded OS on the same computer, there's definitely a difference in sound.

I'm not sure if I'm liking the Linux better at this point. At first it seemed more raw and direct. But after comparing further, it also may be a bit too raw. Stringed instruments have more edge to them and seem a bit too metallic. Where on Windows they are smoother and more delicate.

It's gonna take some more time to evaluate. I'll update after some more listening.
 
Upon further comparison of Audiolinux running the networkaudiod daemon in command line mode in both DoP and native DSD with my fine tuned Windows 8 embedded OS on the same computer, there's definitely a difference in sound.

I'm not sure if I'm liking the Linux better at this point. At first it seemed more raw and direct. But after comparing further, it also may be a bit too raw. Stringed instruments have more edge to them and seem a bit too metallic. Where on Windows they are smoother and more delicate.

It's gonna take some more time to evaluate. I'll update after some more listening.


I think what's going on here may be that it's a bit louder with Linux, and it's causing my weak output stage to work harder. I'll have Sonny's preamp Friday. Better testing can be performed after that.
 
Ok guys I take back my last comments about the Audiolinux based NAA. I realized after the problem was I was running the USB stick the installation was on from a USB hub while doing the comparison with Windows 8 embedded. The Windows was installed on a SATA SSD. So I unplugged the SATA SSD, disabled the SATA bus in the bios, disabled the mPCI-E slot, and the onboard eMMC drive. Then I plugged the stick into the USB 3 port and booted into command mode.

Wow! I can't believe the difference that made! The sound coming out of this unit is stunning. After hearing this I feel stupid for using Windows and OSX for all these years.

I also compared to running the NAA daemon on my Mac mini. Absolutely horrible. Unlistenable after hearing this. Not bad for a $99 SBC running 29 Euro OS off of a $10 USB stick!

Next I'm going to try running my HQplayer server off this OS as well. I can only see it making an even bigger difference. Although it's hard to imagine it getting much better!
 
Status
Not open for further replies.