1 Bit D/A ?

[John Smith]

Hi,

I am confused when I see high end audio manufacturers advertise 1 bit D/A
converter in their equipment. This makes no sense to me. As an Electrical
Engineer I know that the more bits you have when converting analog to
digital the better it is. I am sure I don't quite understand what they mean
by 1 bit D/A. Can someone please explain to me what is meant by it? I did a
search on the web and come up with a lot of irrelevant stuff.

Thanks!

[chung]

John Smith wrote:
> Hi,
>
> I am confused when I see high end audio manufacturers advertise 1 bit D/A
> converter in their equipment. This makes no sense to me. As an Electrical
> Engineer I know that the more bits you have when converting analog to
> digital the better it is. I am sure I don't quite understand what they mean
> by 1 bit D/A. Can someone please explain to me what is meant by it? I did a
> search on the web and come up with a lot of irrelevant stuff.
>
> Thanks!

Perhaps one of the things you need to be aware of is that in 1-bit
DAC's, the sampling rate is much higher than the original 44.1KHz. You
can visualize that a 1-bit DAC can generate *average* levels finer than
1 bit, if the number of samples used for averaging is larger than 1. For
instance, the 4 bit sequence 0, 1, 1, 1 will give an average level of
0.75 over that 4-bit interval, a value that is not realizable with a
1-bit DAC if no averaging is used.

Then you need to have some understanding of oversampling digital filters
and the effects of dither on the spectra of quantization noise. If you
do searches on those keywords, there should be plenty of references on
the web.

BTW, some high-end manufacturers use 1-bit DAC's, but a lot more, in
fact several orders of magnitude more, of those DAC's go into consumer
portable and other applications that are commonly considered low-end or
mid-fi.

[stany]

maybe it should be 1 byte DAC ? (not BIT)
(1 byte = 8 bits)
but also not used ind high-end equipment

generally number of bits declare how many voltage levels can be
recognized by an DAC or ADC. when you have 8-bits DAC you have 2^8 = 256
voltage levels. so in standart 0.7V input level smallest recognisable
level change is approx. 2,7mV (of ocurse depens on many other electrical
features and construction of DAC or ADC).

1 byte DAC ? strange...

chung wrote:

> John Smith wrote:
>
>> Hi,
>>
>> I am confused when I see high end audio manufacturers advertise 1 bit D/A
>> converter in their equipment. This makes no sense to me. As an Electrical
>> Engineer I know that the more bits you have when converting analog to
>> digital the better it is. I am sure I don't quite understand what they
>> mean
>> by 1 bit D/A. Can someone please explain to me what is meant by it? I
>> did a
>> search on the web and come up with a lot of irrelevant stuff.
>>
>> Thanks!
>
>
> Perhaps one of the things you need to be aware of is that in 1-bit
> DAC's, the sampling rate is much higher than the original 44.1KHz. You
> can visualize that a 1-bit DAC can generate *average* levels finer than
> 1 bit, if the number of samples used for averaging is larger than 1. For
> instance, the 4 bit sequence 0, 1, 1, 1 will give an average level of
> 0.75 over that 4-bit interval, a value that is not realizable with a
> 1-bit DAC if no averaging is used.
>
> Then you need to have some understanding of oversampling digital filters
> and the effects of dither on the spectra of quantization noise. If you
> do searches on those keywords, there should be plenty of references on
> the web.
>
> BTW, some high-end manufacturers use 1-bit DAC's, but a lot more, in
> fact several orders of magnitude more, of those DAC's go into consumer
> portable and other applications that are commonly considered low-end or
> mid-fi.

[Richard D Pierce]

In article <j8o7b.407804$uu5.74358@sccrnsc04>, stany > wrote:
>maybe it should be 1 byte DAC ? (not BIT)
>(1 byte = 8 bits)
>but also not used ind high-end equipment
>
>generally number of bits declare how many voltage levels can be
>recognized by an DAC or ADC. when you have 8-bits DAC you have 2^8 = 256
>voltage levels. so in standart 0.7V input level smallest recognisable
>level change is approx. 2,7mV (of ocurse depens on many other electrical
>features and construction of DAC or ADC).
>
>1 byte DAC ? strange...

1-bit DAC? Strange? In a high-end product?

Most certainly.

Consider the following: a 16 bit stream at 44.1 kHz. Now, take
that, resample it at 88.2 kHz at 15 bits, but filter it to 20
Khz, just like before. Do the math, and you'll find that you
haven't lost any data: the signal coming out (assuming it's
implemented properly) will be the same as the 44.1 kHz/16 bit
version.

Fine, resample again at 176.4 kHz. You'll find you only need 14
bits to get the same signal, again filtered to 20 kHz.

Do it again, 352.8 kHz/13 bit, filter to 20 kHz. Same signal, no
loss of resolution, no increase in noise.

Again, only at 705.6 kHz/12 bit...

Again, 1411.2 kHz/11 bit..

and so on.

The point being is that it is NOT a 1-bit, 44.1 kHz DAC. It's
sampling at a MUCH higher sample rate.

It's a wee bit more complicated than that, but, basically, as
long as you maintain the information contents, 1 bit highly
oversampled DACS are theoretically as accurate as base-band
sampled, full-width DACs.

--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| |

[stany]

> It's a wee bit more complicated than that, but, basically, as
> long as you maintain the information contents, 1 bit highly
> oversampled DACS are theoretically as accurate as base-band
> sampled, full-width DACs.

i undrestand now
so, you mean about 16-bits (considering input) DACS using high rate
oversampling.
so, user that started this discussion made the mistake writing about
1-bit DACS :)

it's sth like ADCS delta-sigma

and it's better solution than weight-compensation DACS