The Best DAC is no DAC

Jump to Latest
#141
You
Pano said:
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.
Click to expand...

Yes, but no. The transitions all fall on clock edges (quantized) whereas class D transitions are movable in continuous time (analog).

This damned tablet keeps 'correcting' my spelling. I only notice once I've posted it.
 
Last edited:
#143
Yes, in the case of constant frequency one transition is clocked, but the other moves in continuous time. This is also called PWM, and needs to be distinguished from digital PWM. I think (I stand to be corrected here) this arrangement is more commonly encountered in SMPS's, which are amplifiers, like any voltage regulator. Similarly constant pulse-width can be employed with continuously variable frequency, and this is a form of Pulse Density Modulation.

I was just trying to draw a quick distinction for Pano, with reference to self-oscillating types, but in general class D amplifiers have a continuously variable control element as opposed to digital PDM, which is quantized. This is why they can achieve good fidelity at oscillation frequencies that would not be high enough for digital, which I think needs to go to 2.8901376 GHz to get adequate resolution (16/44k1) PWM.
 
#145
Thanks Counter Culture, I don't know much about PDM, but am keen to learn.
Still, if the line level device needs only a low pass filter to recover the AF, then wouldn't that be the same with a power device? Of course switching that fast would not be easy, as you point out.
 
#146
DSD power amp

Pano said:
Thanks Counter Culture, I don't know much about PDM, but am keen to learn.
Still, if the line level device needs only a low pass filter to recover the AF, then wouldn't that be the same with a power device? Of course switching that fast would not be easy, as you point out.
Click to expand...

There is an interesting AES paper on this,
"Digital audio power amplifier for DSD data streams" (authors Prime and Hawksford)
It is a digital power amplifier optimized specifically for use with DSD.
 
#148
Below my concept for a output filter/buffer schematic. The values will follow later after testing. It is a 24db/oct LP filter at 34khz, transformer decoppling between digital and analog, the tube buffer can also drive headphones and at the and a relaycontact for muting so that there a no strange noises when there is no music playing.
 

Attachments

  • DSDDac.jpg
    DSDDac.jpg
    50.1 KB · Views: 1,580
Last edited:
#151
Koifarm said:
Below my concept for a output filter/buffer schematic. The values will follow later after testing. It is a 24db/oct LP filter at 34khz, transformer decoppling between digital and analog, the tube buffer can also drive headphones and at the and a relaycontact for muting so that there a no strange noises when there is no music playing.
Click to expand...

Please keep us updated of your progress. A couple of questions if I may;

why 34KHz? are you assuming minimum rate of DSD128?

How about a TVC instead of the Llundahl?

Ray
 
#152
Hello Ray,

the crossingpoint is even a little bit heigher, 36,4khz.(butterworth).At that point the frequencycurce at 20khz was only -0.1db lower.
The primary resistance of the transformer needs to be relative high and must be known. This resistance is part of the filter calculation, in this case 1220 ohm.
Until yet this is just a concept.
Waiting for my amareno, other parts i have already. If the concept is proven i will design also a PCB. This PCB will supply several options. Just passive filter or with tube output or opamp output.

Ronny
 
#153
Koifarm said:
The primary resistance of the transformer needs to be relative high and must be known. This resistance is part of the filter calculation, in this case 1220 ohm.
Click to expand...

I was thinking of the TVCs mentioned at the start of the thread;

SAC Thailand

or these;

SOWTER ATTENUATOR TRANSFORMERS TVC VOLUME CONTROL

Anyway, appreciate your sharing and looking forward to seeing developments.

Today I put together some PC hardware based on a passively cooled Intel Atom processor and a SSD. I've installed AP-Linux and it all seems to be working fine - it seems like a nice package. Now waiting on delivery of a JLSounds USB board, then I'll be able to experiment with a simple filter like Hazard's.

Ray
 
#155
nautibuoy said:
I was thinking of the TVCs mentioned at the start of the thread;

SAC Thailand

or these;

SOWTER ATTENUATOR TRANSFORMERS TVC VOLUME CONTROL

Anyway, appreciate your sharing and looking forward to seeing developments.

Today I put together some PC hardware based on a passively cooled Intel Atom processor and a SSD. I've installed AP-Linux and it all seems to be working fine - it seems like a nice package. Now waiting on delivery of a JLSounds USB board, then I'll be able to experiment with a simple filter like Hazard's.

Ray
Click to expand...


Will be very interested, I have a jlsounds board but cant get it to work in this application.
 
#156
Koifarm said:
The primary resistance of the transformer needs to be relative high and must be known. This resistance is part of the filter calculation, in this case 1220 ohm.
Click to expand...
Not sure I follow you here. Doesn't the low pass filter point depend on the impedance it's driving? Meaning the impedance of the circuit following the transformer divided by the transformer ratio and the addition of DCR?
 
You must log in or register to reply here.
Top Bottom