[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Date Index][
Thread Index][
Author Index]
Re: Frequencies was Re: AW: AW: OT: new Macbook wíthOUT Firewire :(
Stefan Tiedje wrote:
>
> The main reason why higher sampling rates sound better, is avoided
> aliasing. It's almost impossible to create an analog filter which will
> pass all until 20 kHz and cutting all above 22.05 kHz. To create a
> filter that has a complete octave to do that is much easier to build.
Yes - in ye olde days when antialiazing filters were purely analog.
With oversampling and digital filters on-chip, this no longer holds
true. The remaining analog filter serves mainly as a reconstruction
filter. See further below..
> The bad reputation of digital versus analog is due to bad implementation
> of converters in the beginning of the digital era btw...
Totally agree. Back in '84, I build a digital delay for audio as my
graduation project. 44 Khz sample rate (IIRC), 18Khz bandwidth, 11th
order Sallen-Key Bessel filters, Analog Devices ADC & DAC, Burr Brown
sample-hold amplifier, metal film resistors and the best film capacitors
I could get hold on. Sounded gorgeously clean.
What a job matching 1% resistors and caps to ˝ % tolerance for those
high order filters. Fortunately technology has evolved beyond that ;)
Most filters back then were low-order Butterworth types, which has kinda
acceptable ripple in the passband and not-too-severe phase issues, but
elliptical filters, like audio-wise shameless Chebychef and Legendre
filters, has been used too, because they quickly reach a steep flank,
and thus *seemingly* require less complicated filters.
Seemingly, because such filters do not retain the steep flanks, but
starts to level out even before reaching the buttom of the dynamic range
of the device. So, not only did they sound awful, but aliasing products
got introduced, which, even at low levels, were still audible.
> Stefan
--
rgds,
van Sinn