1-bit dual D/A Converter question

From what I've read delta sigma modulators work at a very high sampling
rate(the pulse train or bit stream). If this is always the case, then why do manufacturers state there one bit converters are 8x oversampling for example.

Next question, aren't one bit converter's used in dvd-audio (and cd players)players the same as one bit converter used in SACD.
And if SACD can get away with filtering can't dvd-audio(or CD players) since they use one bit converters?

What a confused question! Manufacturers can use whatever oversampling rate they like on a D-A converter - Texas even make converters where there is a choice, for heaven's sake...

All oversampling does in this context is to act as a way of avoiding the use of anti-alias filters, by shifting the effective Nyquist frequency so far away from the audio band that it's never a problem - the effective output clock is so high that you can use a simple roll-off to get rid of it. Once again, Texas actually make an oversampling device that they describe as a filter!

So you have to remove the concept of filtering from the issues of 1-bit conversion - except that it is only realistically possible to implement this on a 1-bit system, because of its inherently serial nature. AFAIA, there is no real distinction to be made about what the source is - both DVD-A and SACD can be played perfectly well through the D-A converters on Pioneer players, can't they?

Maybe I am confused!

I think i was meant to say sigma delta conveter and not vice versa.

I quote from K.C. Pohlmann, Principles of Digital Audio, 3rd ed., McGraw-Hill, 1995

As with other low-bit coders, to achieve high resolution, high sampling rates are required; for example, with an audio band 22,1 kHz and 64 times oversampling, the internal sampling frequency rises to 2.8224 MHz, thus quantization noise is spread from dc to 1.4112 MHz

Can you see why I'm confused? 64 is much higher than 8.

I quote from K.C. Pohlmann, Principles of Digital Audio, 3rd ed., McGraw-Hill, 1995

As with other low-bit coders, to achieve high resolution, high sampling rates are required; for example, with an audio band 22,1 kHz and 64 times oversampling, the internal sampling frequency rises to 2.8224 MHz, thus quantization noise is spread from dc to 1.4112 MHz

Can you see why I'm confused? 64 is much higher than 8.

Pohlmann is not wrong at all about the noise - but in the same way that  the Watkinson references do, he assumes a certain knowledge of basic principles in his readers that perhaps he shouldn't...

The higher you raise the effective sampling rate, the wider the bandwidth of the noise - you can't just make magical gains here. You can cheat slightly by shaping the dither of your audio, which has the effect of squashing the signal contribution to the noise level in one part of the band, and letting it rise slightly in another to compensate, but in reality, this effect varies somewhat when played through different systems with different oversampling charactersitics anyway. The difference that any of these techniques actually make to the percieved noise floor is absolutely minimal.

The bottom line is that you get more improvement because the filtering becomes much easier to implement - other factors in the design make more difference to the output than this - like noise in the monotonic steps, and the overall stability of the conversion. But however you look at it, above 4x, most of the advantages are to the marketing department, not the consumer - a well-designed 8x oversampling system will sound better than a 64x one which uses naff components in the output stage, for instance. And quite frankly, you are looking at vanishingly small differences anyway. When it comes to production techniques, you can often make far more of an improvement to the sound that comes out than you will by changing from 8x to 64x oversampling.