#1 Computer Music Sans DAC
Posted: Sun Jul 05, 2015 8:11 pm
I've been trying a different approach to delivering 'computer audio' without the use of a conventional DAC.
My interest was piqued by enjoying DSD128 conversions of CD FLAC rips more than the FLACs via my ES9018 equipped renderer, then I came across several online discussion topics exploring a different approach to DSD playback. So the key to the experiment I've been working on is DSD format files.
Unlike PCM formats, which are strictly digital, DSD inhabits a sort of twilight zone between digital and analogue using a HF carrier signal. If you strip away the carrier signal using a low pass filter you end up with an analogue signal. This isn't new, I know the late Allen Wright was upgrading SACD players more than a decade ago by tapping into the DSD signal and passing it through a filter and buffer.
Anyway, I thought I would build a quick 'prototype' to see how it sounds. I've reused an old computer with an Intel Atom motherboard and processor, loaded up with AP-Linux and equipped with a solid state disk. That gave me the basic ability to process DSD files. Playback is output via USB to a JL Sounds USB card and the DSD signal taped from its header pins. I've made a little PCB that holds the filter and a mute circuit (there are thumps between some tracks due to the USB board detecting the required codec). The filter I'm using is a simple first order at around 40KHz; it is simply a resistor in series and a capacitor in parallel. There is around 1.65V of DC present on the header pins so I'm passing the signal after filtering through a cap and then onto a simple buffer. Here are some pictures.
First picture is of the populated filter/mute board;
The PCB is from 'DirtyPCB' is good and cost less than £10 for ten, including shipping. The board is 1.6mm with 1oz copper.
Second picture is an overview: bottom left is obviously the Intel Atom motherboard. The white PCB to the right of the SanDisk SSD holds two 5V regulated supplies for the two parts of the USB board, these use TPS7A4700 devices. Next to the right is the USB board, with my filter board mounted on top via the headers. Next to the right is the JL Sounds JG Buffer board, with its +/-15V power supplies beneath (black PCBs, again using TPS7A4700) and, finally, a couple of small toroids to power it all.
Third picture is a closer picture of the USB board with my filter PCB.
So what does it sound like? Actually incredibly good. I would describe it as rich, smooth and detailed. Spatial separation is very good, you really can pick out individual instruments/voices in their own space.
Downsides? Output is a bit low and a few decibels of gain in the buffer would be useful but the level is still listenable. There is an audible click between tracks and occasionally during tracks. Between tracks will be taken care of by the mute circuit (I've had to remove the relay for now as I got the connections the wrong way round) and I'm pretty sure the noises within tracks is the result of doing things on the computer whilst playback is ongoing (I use a remote terminal session on my laptop to control playback and now just don't touch it while I'm listening). None of the noises are of speaker damaging magnitude.
Greg, Pepperoni and I are having a get together in a week or so and I plan to take the DSD player along to get their view on its performance - maybe they'll give an opinion in due course.
If anyone is interested in having a go I can post some more info and answer any questions. I have a few PCBs I can send out for the cost of the postage. Please note that the PCBs have the relay mistake so you would need to hack a workround or live with the clicks. I have replacement PCBs on order as I do plan to use the mute circuit.
For me, what I've heard so far is sufficient for me to want to take this further and I've started working on a valve based solution using a Broskie Aikido PCB; this will include an active second order filter and some gain.
Ray
My interest was piqued by enjoying DSD128 conversions of CD FLAC rips more than the FLACs via my ES9018 equipped renderer, then I came across several online discussion topics exploring a different approach to DSD playback. So the key to the experiment I've been working on is DSD format files.
Unlike PCM formats, which are strictly digital, DSD inhabits a sort of twilight zone between digital and analogue using a HF carrier signal. If you strip away the carrier signal using a low pass filter you end up with an analogue signal. This isn't new, I know the late Allen Wright was upgrading SACD players more than a decade ago by tapping into the DSD signal and passing it through a filter and buffer.
Anyway, I thought I would build a quick 'prototype' to see how it sounds. I've reused an old computer with an Intel Atom motherboard and processor, loaded up with AP-Linux and equipped with a solid state disk. That gave me the basic ability to process DSD files. Playback is output via USB to a JL Sounds USB card and the DSD signal taped from its header pins. I've made a little PCB that holds the filter and a mute circuit (there are thumps between some tracks due to the USB board detecting the required codec). The filter I'm using is a simple first order at around 40KHz; it is simply a resistor in series and a capacitor in parallel. There is around 1.65V of DC present on the header pins so I'm passing the signal after filtering through a cap and then onto a simple buffer. Here are some pictures.
First picture is of the populated filter/mute board;
The PCB is from 'DirtyPCB' is good and cost less than £10 for ten, including shipping. The board is 1.6mm with 1oz copper.
Second picture is an overview: bottom left is obviously the Intel Atom motherboard. The white PCB to the right of the SanDisk SSD holds two 5V regulated supplies for the two parts of the USB board, these use TPS7A4700 devices. Next to the right is the USB board, with my filter board mounted on top via the headers. Next to the right is the JL Sounds JG Buffer board, with its +/-15V power supplies beneath (black PCBs, again using TPS7A4700) and, finally, a couple of small toroids to power it all.
Third picture is a closer picture of the USB board with my filter PCB.
So what does it sound like? Actually incredibly good. I would describe it as rich, smooth and detailed. Spatial separation is very good, you really can pick out individual instruments/voices in their own space.
Downsides? Output is a bit low and a few decibels of gain in the buffer would be useful but the level is still listenable. There is an audible click between tracks and occasionally during tracks. Between tracks will be taken care of by the mute circuit (I've had to remove the relay for now as I got the connections the wrong way round) and I'm pretty sure the noises within tracks is the result of doing things on the computer whilst playback is ongoing (I use a remote terminal session on my laptop to control playback and now just don't touch it while I'm listening). None of the noises are of speaker damaging magnitude.
Greg, Pepperoni and I are having a get together in a week or so and I plan to take the DSD player along to get their view on its performance - maybe they'll give an opinion in due course.
If anyone is interested in having a go I can post some more info and answer any questions. I have a few PCBs I can send out for the cost of the postage. Please note that the PCBs have the relay mistake so you would need to hack a workround or live with the clicks. I have replacement PCBs on order as I do plan to use the mute circuit.
For me, what I've heard so far is sufficient for me to want to take this further and I've started working on a valve based solution using a Broskie Aikido PCB; this will include an active second order filter and some gain.
Ray