Trinity's claims vs. what actually happens in digital

Ken Newton said:

I'm uncertain as to what you refer to by "digital+analog oversampling" with respect to Trinity's reconstruction filter implementation. Are you saying that Trinity also utilize some digital reconstruction filter?

Click to expand...

Yes, see http://www.whatsbestforum.com/showthread.php?12023-Trinity-DAC&p=243849&viewfull=1#post243849

Ken Newton said:

Setting the above question aside for the moment, with oversampled interpolation the ultrasonic image bands will still be present, but shifted up in frequency by a factor equal to the oversampling ratio. So, while the typical analog filter can be designed to have a lessor subjective impact on the audio band frequencies, it should still be present in some form.

Click to expand...

But if the digital oversampling ratio is high enough can you go without the analog filter?

Anatta said:

Yes, see http://www.whatsbestforum.com/showthread.php?12023-Trinity-DAC&p=243849&viewfull=1#post243849

But if the digital oversampling ratio is high enough can you go without the analog filter?

Click to expand...

I had not seen that chart before. It likely has big implications for both the impulse and frequency response of the Trinity DAC, regardless of the use of LIANOTEC. Do you know of any published measurements of those two parameters for the Trinity DAC?

If the oversampling ratio is high enough, the analog filtering will effectively be performed by the parasitic circuit capacitance presented by the interconnects. Aiding this filtering is that the oversampled image bands are also strongly suppressed by a frequency domain masking profile stemming from D/A converter sample-and-hold operation. So, yes, then you wouldn't require a dedicated analog filter. However, it would be good practice to include at least a benign first order low pass analog filter, rather than leave effective filtering to chance. The primary technical concern behind filtering away the undesired ultrasonic image bands is to remove the possibility of their being inter-modulated down to within the audio band.

Ken Newton said:

I had not seen that chart before. It likely has big implications for both the impulse and frequency response of the Trinity DAC, regardless of the use of LIANOTEC. Do you know of any published measurements of those two parameters for the Trinity DAC?

Click to expand...

Check the manual @ p 28-29 for frequency response, image suppression and impulse response for the 2 digital filters available, DF1 & DF2, and LIANOTEC.

www.trinity-ed.de/typo/fileadmin/template/bilder/dac/DAC_Manual_2013.pdf

Ken Newton said:

The primary technical concern behind filtering away the undesired ultrasonic image bands is to remove the possibility of their being inter-modulated down to within the audio band.

Click to expand...

Probably the filtering is more critical with DS DACs due to out of band noise which the R2R DACs don't have (correct me if I'm wrong) and this may be a reason for which Trinity selected the 1704 chip.

If the oversampling pushes the first image bands way beyond the amplifier bandwidth would it still matter? That's what Trinity claims with its 8x digital + 8x analog oversamplig (manual p 16)

The Weiss Medus for example uses the Sabre 9018 which oversamples to 1.536Mhz (as Trinity does to 192kHz signals), but after the conversion it also has a 3rd order analog filter. So maybe the problem with commercially available DS DACs is that they don't accept high enough sample rates to get away only with digital filtering. And then you would also need a custom digital filter that would mimic the response of the LIANOTEC, if that is possible.

Anatta said:

Check the manual @ p 28-29 for frequency response, image suppression and impulse response for the 2 digital filters available, DF1 & DF2, and LIANOTEC...

Click to expand...

What I conclude from my reading of the Trinity DAC manual is, that their sound quality is not primarily due to linear interpolation - at least, not for source sample rates less than 176KHz. The impulse response graphs clearly depict that linear phase brick wall type digital filters are utilized, one featuring a sharp transition and the other a slow, both apparently in a hybrid reconstruction filter scheme utilized in tandem with LIANOTEC. Linear interpolation cannot remove the pre-ringing or post ringing produced by these digital filter functions. I now suspect that much of the Trinity sound character is largely due to other design factors, such as the massive paralleling of D/A converter unit outputs.

Probably the filtering is more critical with DS DACs due to out of band noise which the R2R DACs don't have (correct me if I'm wrong) and this may be a reason for which Trinity selected the 1704 chip.

Click to expand...

While delta-sigma conversion generates out-of-band quantization noise, full resolution R-2R type converters (such as in the PCM1704 chip) will still produce strong ultrasonic image bands which must be effectively filtered for correct signal reconstruction.

If the oversampling pushes the first image bands way beyond the amplifier bandwidth would it still matter? That's what Trinity claims with its 8x digital + 8x analog oversamplig (manual p 16)...

Click to expand...

It may still matter. At high oversampling ratios, the image bands are essentially feeding RF to the next amplification stage. Unfiltered, this may or may not cause inter-modulation, junction rectification, or common-mode signal coupling problems, depending on how resistant that stage is to RF input.