6213	2011-04-26 20:42:11	David	audio filter design rules of thumb
Message	Date	From	Subject
Is there a rule of thumb for maximum Delay and Phase shift that is tolerable? In some other designs I've seen somewhere in the order of 1 to 1.5 ms of delay response. Invariably there is at least one 180 phase shift near the corner frequency.
6217	2011-04-27 07:02:34	Tim	Re: audio filter design rules of thumb
In EMRFD there's some rules of thumb oriented towards crystal filters sounding "good" but I see no reason that the same rules shouldn't apply to audio filters. I think my advice and the EMRFD advice is very much oriented towards CW use but also applicable to other digital modes. It seems to me (having built a lot of audio filters with high Q over the decades that sound like crap in the presence of noise) that it's not necessarily the absolute amount of delay or phase shift, but it's the steepness of change in phase shift that induces the ringing or psycho-acoustical effects that I personally don't like. ("Psycho-acoustical" in that static crashes start sounding like morse dits and dashes after enough time. It's almost a psychosis.) 1 to 1.5 ms of absolute delay implies a bandwidth of a kHz I think and most of my complaints and failures were with narrower filters. But I think if you follow the EMRFD advice that you can build a kHz wide filter that sounds "better" or "brighter" than a kHz wide filter built not following that advice. Designs that purposely put the regions of greatest phase shift past the edge of the passband sound exceptionally good compared to brick wall filters that put a lot of phase shift right at the edge of the passband. Look especially at Gaussian-to-12dB. I think NatSemi's webbench still allows you to design audio filters with this good characteristic. (I hope that doesn't disappear with TI's acquisition). Others here might tell you they never designed a bad audio filter, but I'll tell you up front that I've built a lot of audio filters that I was unhappy with, and I've also heard a lot of crystal filters and DSP processors that had wonderful shape factors but I thought sounded awful
6222	2011-04-27 09:09:45	Fernando Krouwel	Re: audio filter design rules of thumb
Dan Tayloe, N7VE has a document showing several tips about audio filtering design, as well as a sugestion to use Texas Instruments free software for it. Personally I have designed an audio filter for SSB receiving with excellent results based
6224	2011-04-28 03:46:51	Vojtech	Re: audio filter design rules of thumb
Tim, > It seems to me (having built a lot of audio filters with high Q over the decades that sound like crap in the presence of noise) that it's not necessarily the absolute amount of delay or phase shift, but it's the steepness of change in phase shift that induces the ringing or psycho-acoustical effects that I personally don't like. ("Psycho-acoustical" in that static crashes start sounding like morse dits and dashes after enough time. It's almost a psychosis.) Absolute amount of phase shift is irrelevant as long as it is linear or nearly linear with frequency. Group delay is irrelevant as long as it is constant or nearly constant (group delay is derivation of phase shift by frequency). There is nothing psycho-acoustical on filter ringing. It is very real. It modifies time response of the filter. Noise spike comes out as ringing waveform, which masks the signal. Many of 2x NE602 + narrowband crystal ladder filter superhets sound like crap in presence of impulse noise, just because the crystal filter has wild phase response near the filter edges. 73, Vojtech OK1IAK, AB2ZA