Computer Audio Asylum Music servers and other computer based digital audio technologies.

---

[ Follow Ups ] Thread: [ Display  All  Email ] [ Computer Audio Asylum ]

---

[ Alert Moderator ]

---

Re: "passive" vs "active"

Posted by Christine Tham (R) on November 22, 2006 at 12:11:41

In Reply to: Re: "passive" vs "active" posted by Presto on November 21, 2006 at 23:48:21:

*** Yes, I do see how the crossover and the driver "add up" (vectorally) to provide a net impedance as seen by the amp. ***

This was the main point I ws trying to make, so I'm glad we both agree.

*** An amp in an active system only "sees" the impedance of the connected driver that are NOT in the stop-band (aka only the pass-band and crossover regions.) ***

I think you will find I have already aknowledged this earlier. To quote one of my earlier posts: "There is a counter argument that says filtering the amp used to drive the tweeter achieves better transparency."

I noticed you are mainly using the tweeter in your argument, and I seem to recall that research does suggest that filtering at the preamp primarily benefits the tweeter.

But a tweeter is very easy to amplify, typically only a few watts at the most. So the benefit of the approach that you suggest is minimal. Arguably filtering post amp is almost as good (but you do waste a lot of the power of the amp, admittedly).

Whereas it can be shown a woofer's impedance does change significantly with the addition of a filter post amp, and woofers require hundreds of watts to drive at high decibels (power required is inverse to frequency).

Anyway, I don't mean to say that there are no benefits to digital crossovers. And certainly you can potentially do digital crossovers, plus equalization, plus room correction more accurately than an analog only solution. I just don't like the "haze" that that imparts.

If you look at 32-bit floating point processing (which is what most software and DSPs do), it's not good enough to preserve 24-bit resolution. A 32-bit floating point number has 24-bit mantissa and 8-bit exponent. Numerical theory suggests you need approximately twice the resolution of your inputs when doing calculations in order to preserve the precision of the inputs. So 32-bit fp is barely adequate for 16-bit signals (the 24-bit mantissa has about 8 bit headroom for intermediate results), but clearly inadequate for 24-bit signals.

Ideally, you want to do all your calculations in 64 bit floating point (or simply use IEEE double precision which holds intermediate results to 80 bits), and then dither the final result to 24-bit fixed point before passing to the DAC.

I don't know any player or DSP that currently does this in real time. When that day comes, I'll be very keen to try it out. TI for example is starting to manufacture DSPs capable of 64-bit processing. And 64-bit VST and DirectX effects should hopefully be more prevalent when Vista is released. Sonar already has a 64-bit mixing engine, and I seem to recall Perfect Space is a 64-bit convolution reverb (but I could be wrong).

This post is made possible by the generous support of people like you and our sponsors:

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---