DirectStream

No problem.

I thought it was released because it was available.

Actually it just popped up when I had to reload the forum webpage because it lost my user login.

@frode Oh, no worries. I just donā€™t want to start sending folks there yet. It is publicly available because we have multiple developers needing access to it. Do you like the general feel of it?

@frode Oh, no worries. I just don't want to start sending folks there yet. It is publicly available because we have multiple developers needing access to it. Do you like the general feel of it?

Absolutely, it is a giant leap forward, still if the current one also appear nice.

Hi @tedsmith

Iā€™m quoting a 5 year old thing here to show that I have done a bit of a search before I ask this next question hehe. And this part quoted is the closest thing I could find in context of my question.

I read the infamous Andreas Koch article which says with DSD64 the noise shaping curve starts at ~20 KHz, DSD128 ~40kHz and DSD256 ~80kHz.

Raising the Sample Rate of DSD - Is There a Sweet Spot? - Positive Feedback

Can you share a bit more about the analogue filtering of the DS?

For example, at what frequency does it start low pass filtering and is it classed as a 2nd order or 3rd order analogue filter or something else?

Cheers!

For DSD64 thereā€™s very little room to play with and still meet the SACD specs (-120dBFS S/N over the audio band, ultrasonic noise below -40dBFS, etc.) With higher rates you can make more choices and still meet those specs. The numbers quote above are what happens if you keep the same filtering in your sigma delta modulator and design the output filter to match it.

Since no one except the output filter of the DS seeā€™s the actual bits Iā€™m free to do what I want, but Iā€™m stuck with the output filter as designed. The DS was designed for a 5th order analog low pass filter with itā€™s -dB point at 80k. We decided that we didnā€™t like the sound as much as we did if we dropped one of the filtering caps and so the DS Sr hardware is now a fourth order filter. Different releases of the software have moved the digital noise shaping filter around, and I certainly didnā€™t move it up by a factor of two when I went to quad rate internal processing. By letting the noise floor rise a little sooner the audio band can be quieter and have fewer anomalies, so I only raised it a little when I went from double rate processing to quad rate processing.

Early SACD players had a switch to set the filter cutoff at a lower frequency or the higher frequency of about 50kHz because some early (expensive) amps that they tried SACDs on blew up with 50kHz cutoff. In the DS I could go higher because I was starting at DSD128 and so I had a little more room to play with.

1 Like

Thanks Ted. Is the low pass filtering part of the output stage composed only of filtering capacitors? Or other electronics also (passive)?

And can you share what components are between the FPGA output and the output RCA output connectors?

I understand thereā€™s the filtering caps, then high-speed class A video amplifiers, then Edcor 1/2 W transformers do some low pass filtering also?

The filter is balanced RCVRCXCR. The V is the balanced digital switches (video opamps), The X is the transformers, they sort of have a 2nd order Bessel shape on both the top and bottom end. The exact format doesnā€™t really matter, I just picked the places I could get the best quality components with reasonable values. The Jr is a scrambled version of that and has an opamp instead of a transformer. The capacitors are 2% film caps and the resistors are 0.05% oversized thin film resistors.

2 Likes