What does the DSD waveform look like? I've never looked at it. Class-D is usually a square wave with variable duty cycle.
DSD utilizes a DAC
In a 1-bit digital signal encoding/decoding system, such as DSD, the system only needs to represent two discrete amplitude states. Call those two states high and low, or 1 and 0, or positive going and negative going, or whatever, but only two states need be recorded, and later reproduced. Because of that, there isn't the need for most of the circuitry (resistor networks and such) required for multibit digital signal conversion. A simple logic gate or some other two-state driver circuit could form the DAC circuit for 1-bit decoding. One consequence of 1-bit encoding/decoding is that is creates a great deal of quantization noise, but that can be moved out of the audible band. In short, functionally, there is a DAC circuit even with bare bones DSD playback.
In a 1-bit digital signal encoding/decoding system, such as DSD, the system only needs to represent two discrete amplitude states. Call those two states high and low, or 1 and 0, or positive going and negative going, or whatever, but only two states need be recorded, and later reproduced. Because of that, there isn't the need for most of the circuitry (resistor networks and such) required for multibit digital signal conversion. A simple logic gate or some other two-state driver circuit could form the DAC circuit for 1-bit decoding. One consequence of 1-bit encoding/decoding is that is creates a great deal of quantization noise, but that can be moved out of the audible band. In short, functionally, there is a DAC circuit even with bare bones DSD playback.
Last edited:
You still do have a DAC, just its not called a DAC, its your digital signal's buffer as Ken points out just above.
The downside of using just a digital chip for your DAC is that it has about 6dB PSRR on average - meaning your digital power supply's quality is the limitation. This could easily be the cause of your bass lightness. The other issue is jitter - a 'raw' DSD signal is inordinately sensitive to jitter. It can be ameliorated through using a transversal filter - like Miska has shown in one of his DSD DAC schematics.
The downside of using just a digital chip for your DAC is that it has about 6dB PSRR on average - meaning your digital power supply's quality is the limitation. This could easily be the cause of your bass lightness. The other issue is jitter - a 'raw' DSD signal is inordinately sensitive to jitter. It can be ameliorated through using a transversal filter - like Miska has shown in one of his DSD DAC schematics.
The digital buffer is a buffer. It's input is a digital signal and its output is the same digital signal. So where is the digital to analogue conversion that you are talking about?
The point is this - the digital output from the USB receiver is normally fed to a DAC. I am not feeding it to a DAC. I am trying something different, if it hasn't been tried before. As per Wikipedia (so it must be right!)
"The process of creating a DSD signal is conceptually similar to taking a one-bit delta-sigma analog-to-digital (A/D) converter and removing the decimator, which converts the 1-bit bitstream into multibit PCM. Instead, the 1-bit signal is recorded directly and in theory only requires a lowpass filter to reconstruct the original analog waveform. In reality it is a little more complex, and the analogy is incomplete in that 1-bit sigma-delta converters are these days rather unusual, one reason being that a one-bit signal cannot be dithered properly: most modern sigma-delta converters are multibit."
To re-iterate. I am not using a sigma-delta convertor. Or any other converter. I am feeding a digital signal (DSD, or PDM) into a band-width limited transformer (hence acting as a low pass filter) to reconstruct the original analog waveform. Any conversion (if you want to call it this) happens in the transformer. Same as Class D amp. PWM signal is fed to a low pass filter then to the speaker. Where is the DAC in a Class D amp?
Anyway let's not argue about semantics. Call it a DAC, or not a DAC. Whatever you like. But i am trying to minimise the signal path in my system. Because I feel like it. And I can (well if this works, I can). You point about about power supploies is valid, and I know I need a better p-s. WIll a better p-s improve the bass? That's an interesting point, I can hear noise at the moment that I need to deal with but I hadn't considered the impact of the p-s on bass response.
Also, I don't know what a transversal filter does. Do you have a link?
The D/A conversion occurs when you cease to interpret the signal as a digital one and begin to interpret it as an analog signal. So that happens at the output of your buffer because you're not feeding it into a digital circuit downstream, rather an analog one.
I'll have a hunt around and see where I found Miska's circuit - its an implementation of a 32 tap transversal filter.
I'll have a hunt around and see where I found Miska's circuit - its an implementation of a 32 tap transversal filter.
Here's the link to Miska's zipped design info for his transversal DAC : http://www.signalyst.com/doc/Dsc1-revA.zip
<edit> I'd missed your question about the classD amp. ClassD amps are purely analog so there's no need for a DAC, as they have nothing 'digital' in them.
<edit> I'd missed your question about the classD amp. ClassD amps are purely analog so there's no need for a DAC, as they have nothing 'digital' in them.
Last edited:
As I understand it, a DSD signal is a form of quantised pulse density modulation. Hence the 'digital' signal already contains the analogue signal, which is why a low pass filter is all you need. So the reason why you don't need a DAC is that the signal was never truly 'digital' at all, but merely used a clever way to record and play back an analogue signal which looks like a digital signal.
Note that 'analogue' does not mean 'music signal represented by voltage level' but 'music signal represented by something variable (but not a number - that would be digital)'. Here it is pulse density.
Note that 'analogue' does not mean 'music signal represented by voltage level' but 'music signal represented by something variable (but not a number - that would be digital)'. Here it is pulse density.
"The process of creating a DSD signal is conceptually similar to taking a one-bit delta-sigma analog-to-digital (A/D) converter and removing the decimator, which converts the 1-bit bitstream into multibit PCM. Instead, the 1-bit signal is recorded directly and in theory only requires a lowpass filter to reconstruct the original analog waveform."
Direct Stream Digital - Wikipedia, the free encyclopedia
Assuming all ends well in this thread, maybe we can start discussing which xmos usb card is the best value 😉
XMOS DSD DXD 384kHz high-quality USB to I2S/DSD PCB - DIYINHK
http://www.diyaudio.com/forums/digital-source/236028-xmos-dsd-384-khz-32bit-usb.html
I2SoverUSB - I2S over USB Audio
XMOS DSD DXD 384kHz high-quality USB to I2S/DSD PCB - DIYINHK
http://www.diyaudio.com/forums/digital-source/236028-xmos-dsd-384-khz-32bit-usb.html
I2SoverUSB - I2S over USB Audio
Seems you would mention WAVE IO especially since one of those you did mention looks a awful lot like a copy of WAVE IO ...
http://www.diyaudio.com/forums/digital-source/188902-xmos-based-asynchronous-usb-i2s-interface.html
"LORIEN" is a true gentleman and a great pleasure to do business with.
Sure, keep them coming😉 WaveIO seems like the most expensive of them all. I guess I'll leave it open for discussion which one is the best value.
Hi,
I have an experimental confirmation that you don't need a DAC to convert a DSD signal to analog: http://www.webalice.it/meneghettig/Un%20DSD%20player%20autocostruito.pdf
(sorry, in Italian).
Actually, the DSD signal is discrete in the sense that it can assume only the value high or low, but it is not a numerical representation of a waveform, it is like a PWM signal and for this reason an RC filter can be enough.
I have an experimental confirmation that you don't need a DAC to convert a DSD signal to analog: http://www.webalice.it/meneghettig/Un%20DSD%20player%20autocostruito.pdf
(sorry, in Italian).
Actually, the DSD signal is discrete in the sense that it can assume only the value high or low, but it is not a numerical representation of a waveform, it is like a PWM signal and for this reason an RC filter can be enough.
Well put. Thank you.do you have listed a some bump when put the player in stop and put back in play or changed file directory?
Yes, I do get a burst of noise when starting and stopping playback, or changing tracks. I don't know what is causing this. More experimentation is required.