Why do NOS dacs sound different to oversampling designs?

[rwortman]

I read that review with my b.s. detecter bouncing off the peg. Reverb decay is just another quiet sound. There is no reason for a DAC with decent S/N and linearity to treat reverb any differently than another. The audio writers do much of their listening in sighted tests, often while drinking or using other chemical enhancements (I believe). This theory explains much of the observations. If a DAC really did make reverb decay sound different, it would show up in either a high noise floor or poor low level linearity. R2R DACs were replaced by D/S DACs because of poor linearity. If a ladder DAC made reverb sound different it would likely be a distortion.

 

[Blumlein 88]

Is there a way to measure how well a DAC recreates recorded reverb? I got to thinking about this when reading an article by Herb Reichert in Stereophile about the R-2R Denafrips DACs' handling of reverb in recordings.

The one thing that appears to be consistent about subjective reviews of R-2R DACs is a sense of “soundstage width”, or “space”, or “three-dimensionally”. I interpret this as reverb. I’m not sure how the reproduction of this specific aspect in the recording is reflected (pun intended) in any one measurement.

Humans seem to have a strong attraction to and sensitivity to reverb. I wonder how this translates to the digital realm.

Any thoughts on this?

Herb Reichert was reviewing that DAC using a 300B based SET amp. He liked the modes of operation with no reconstruction filter or a slow over-sampling filter. Both of which allow rather high imaging artifacts above 20 khz. I wonder if that creates low level IMD back into the audio band which gets heard as more reverb or space? He indicated in the modes with steeper filtering the reverb wasn't as apparent and sound was flat. Reads suspiciously like his amp (or the output transformers) may have interacted with the spurious ultrasonic tones that slow or no filtering would cause.

 

[j_j]

There is nothing special about reverb. It's just part of the audio signal.

 

[magicscreen]

Specs are given at 192kHz as mentioned in the sata sheet. Both TDA1543 and TDA1387 are intended to be used oversampled.

Is this means, they are not oversampling DAC chips, only to be fed with oversampled data?
PCM56 as well?

 

[solderdude]

These chips were always used with a separate chip that performed the 4X oversampling, not stand alone as DAC chip and not in external configs either.

 

[j_j]

These chips were always used with a separate chip that performed the 4X oversampling, not stand alone as DAC chip and not in external configs either.

Ah! So the 192 was the '4x'?

 

[solderdude]

These chips were designed to work up to at least 176.4kHz (4x 44.1)
Later DAC chips got onboard oversampling filters which could do 8x or even higher.
Back in the days they could not build accuarate DAC chips with a 16bit resolution but could only manage 14 bit.
The solution Philips came up with was 4x oversampling which had a second advantage which was a much less steep reconstruction/post filter was needed.
This all was cheaper than 16 bit resolution and complex and expensive analog reconstruction filters.
When they could make 16 bits DAC chips the oversampling stuck because of the reconstruction filtering.
That's why all those DAC chips can do at least 192kHz.

Then the audiophool guys and not well versed "designers" happened with a hang to the 'early days where all sounded better' and silly, non compliant, external DACs came into existence.

 

[j_j]

These chips were designed to work up to at least 176.4kHz (4x 44.1)
Later DAC chips got onboard oversampling filters which could do 8x or even higher.
Back in the days they could not build accuarate DAC chips with a 16bit resolution but could only manage 14 bit.
The solution Philips came up with was 4x oversampling which had a second advantage which was a much less steep reconstruction/post filter was needed.
This all was cheaper than 16 bit resolution and complex and expensive analog reconstruction filters.
When they could make 16 bits DAC chips the oversampling stuck because of the reconstruction filtering.
That's why all those DAC chips can do at least 192kHz.

Then the audiophool guys and not well versed "designers" happened with a hang to the 'early days where all sounded better' and silly, non compliant, external DACs came into existence.

I remember the Philips DAC. The one with the stuck LSB It was an interesting example of a small logic error.

But yeah, running at 4x makes things a lot easier to make a reconstruction filter. (Well, there is still one, but it's digital and much better behaved.)

I am afraid I missed the comment where this was explained to be an antique. Pretty much everything now is delta-sigma. r2r really has genuine physics problems in making the ladder itself, and then the reconstruction filter piles on top of everything else.

 

[solderdude]

Antique would actually be DSD as that was invented before PCM

There are folks that like the simplicity of R2R as that works on a more intuitive principle and as these were common in the early days (the antique part)
That is untill Sony quietly, as they did not like to explain the 1bit DACs when the bitrate was booming silently introduced those and Technics came with MASH.
This was the forerunner of DS.

Feels like antiques to me by now

 

[j_j]

Antique would actually be DSD as that was invented before PCM

There are folks that like the simplicity of R2R as that works on a more intuitive principle and as these were common in the early days (the antique part)
That is untill Sony quietly, as they did not like to explain the 1bit DACs when the bitrate was booming silently introduced those and Technics came with MASH.
This was the forerunner of DS.

Feels like antiques to me by now

Well, I think you mean what we usually call "delta modulation" before explicit PCM, yes? Things like, maybe, CVSD? Delta-sigma is what I view as a distinct improvement. DSD itself was simply noise shaped PCM in my view, of course that's true of all delta-sigma, but not of delta-mod, where the integrator is in the feedback loop instead of in front of the ADC. The way delta-sigma works is a bit hard to explain, yes, with the integrator(s) before the ADC instead of in the feedback loop, but it does result in something mathematically equal to noise-shaped PCM, at some very high rate.

But this r2r stuff, well, physics works. r2r is hard as a result, it goes down to actual placement and size of resistors either discrete (really old) or on chip. Then of course there's the filter tolerance problem, which is pretty much horrid in the real world. There is a world of hurt in steep, active analog filters.