Attenuation circuit. What is ideal location?

[albertNL]

Originally Posted by AndrewT

Have you read the system gain article?

Have you done any research on gain and/or attenuation?

No to both questions.

My background is mathematics and IT, not engineering. I am used to modelling and designing first (and remodelling and redesigning second, and third and fourth and fifth and ... you get my point ), with building much, much later.
I always thought that good sound was the result of superior design. What the last year, reading diyaudio, has taught me is that good sound is much more about part selection/interaction and optimising a design for those parts than I had thought.

It has also taught me never to give an example when I ask a hypothetical question

Which system gain article did you have in mind?

[DF96]

Maybe the mods should prohibit use of words like 'best', 'ideal' (except in a theoretical sense), 'ultimate' etc. from the forum? 'Best' and 'ideal' always mean that an optimisation problem must be be posed and solved, which means that detailed goodness criteria must be stated (but they never are). 'Ultimate' means impossible to better, which is a silly idea for anything made by humans.

[albertNL]

Originally Posted by OllBoll

Isn't the ideal to have no attenuation circut at all but to do the attenuation digitally before sending it to the DAC?

At 0dB you can use all 16 (or 24 or however many you have) bits to send the signal. If you attenuate, you need some of those bits to send a lot of zeroes. You cannot use those bits for the actual signal. Information will thus be lost if you use digital attenuation.

[DF96]

Originally Posted by albertNL

I always thought that good sound was the result of superior design.

It is, provided a good implementation is done too. Parts selection is far less important, although it cannot be ignored.

[AndrewT]

On the front page of diyaudio under articles.

It might do you good to read all of them, so that you then know where to go before asking ambiguous questions.

BTW,
I get the impression that you think you are at a disadvantage in not being an electronics trained Member.
You will find if you look that the vast majority here are non electronics and non electrical. They are here because they have electronics as a hobby and as a result they are self taught.

[AndrewT]

I agree with DF, design and implementation leads to good performing product.

[Jay]

Originally Posted by albertNL

My background is mathematics and IT, not engineering...
I always thought that good sound was the result of superior design. What the last year, reading diyaudio, has taught me is that good sound is much more about part selection/interaction and optimising a design for those parts than I had thought.

This audio discipline is not as exact as Math and IT, but far from it. Many experts spend their whole life in audio but still cannot create the "best" system. So never assume that you can fully understand this discipline simply by reading one or two books or spend one or two years in audio projects.

A good sound is a result of a complete system design from source to speaker. They are interdependent. We need to know IT in digital processing. We need to know mechanic and acoustics in speaker design and worse we need to know psychology and biology

Because audio system is usually designed partially, it makes part selection/interaction and optimization become more critical. You may have the "best" amplifier design and the "best" speaker design but if your speaker doesn't want to cooperate with the amp, then you will not get the "best" sound.

[albertNL]

Originally Posted by AndrewT

On the front page of diyaudio under articles.

Thanks.

I get the impression that you think you are at a disadvantage in not being an electronics trained Member.

No, not so much at a disadvantage, just speaking a different language

[DF96]

My first career was in IT, and the language of serious audio is not that different; both are branches of engineering. As I said, 'best' should make you think 'optimisation problem' which then leads to 'criteria?'.

[Jay]

Originally Posted by albertNL

At 0dB you can use all 16 (or 24 or however many you have) bits to send the signal. If you attenuate, you need some of those bits to send a lot of zeroes. You cannot use those bits for the actual signal. Information will thus be lost if you use digital attenuation.

Are you following the digital technology/development in audio? In audio we have amplifier experts, speaker experts and may be "digital" experts. Sometimes an expert in one area doesn't understand the other areas. Cool, isn't it?