Mic pre 1 Hz to 200kHz - why bother?

[Goddard]

The delay depends upon the digital filter employed (and for FIR filters generally, the number of 'taps') and is characterized by the 'group delay' (specified as the number of samples (or number of sampling periods)). FIR filters exhibiting linear phase delay all frequencies by the same amount (constant group delay).

Filters may exhibit pre- and post-echo, as well as pre- and post-ringing (with 'pre-' signifying an onset before the main impulse).

Btw, analogue filters also impose delay (with 'time constants' equating to their 'corner' frequencies), but such delay is typically so very much shorter (by an order of magnitude) compared with digital filters (micro-seconds as compared with milli-seconds) that it is not noticeable although the alteration in frequency and phase response is.

[Hugh Robjohns]

FIR filters exhibiting linear phase delay all frequencies by the same amount (constant group delay).

Surely, all linear-phase filters delay all frequencies by the same amount. That's what the phrase means and a primary raison d'etre.

H

[Goddard]

FIR filters exhibiting linear phase delay all frequencies by the same amount (constant group delay).

Surely, all linear-phase filters delay all frequencies by the same amount. That's what the phrase means and a primary raison d'etre.

H

Yes, but although linear-phase filters will invariably be FIR filters, not all FIR filters are necessarily linear-phase (symmetric/freq-invariant) filters!

[grab]

although linear-phase filters will invariably be FIR filters

Not necessarily. Bessel filters...

[Hugh Robjohns]

although linear-phase filters will invariably be FIR filters

Not necessarily. Bessel filters...

Bessel describes the characteristics. FIR describes the construction!

[Goddard]

although linear-phase filters will invariably be FIR filters

Not necessarily. Bessel filters...

Yes, but a Bessel doesn't really transform to digital form.

[Forum Admin]

... there's an error (missing non-standard character) in the online version where it discusses Dave Blackmer's thinking behind 200kHz being the required upper limit. Hint, it doesn't mean 5 seconds!

I've inserted the missing micro character where I spotted it was missing. Let me know if there are others (I'll ask for the web article auto-translate script to be updated so that it correctly includes such characters henceforth).

Ian

[Tomás Mulcahy]

Its been handy for me while making my new album 'music for elephants a long way away and for some whales too who are always a long way away'

As Spock said There are other species on earth. Only human arrogance would assume the signal must be meant for mankind.

As a bat I resent that statement.

[Mixedup]

I've inserted the missing micro character where I spotted it was missing. Let me know if there are others (I'll ask for the web article auto-translate script to be updated so that it correctly includes such characters henceforth).

Thanks. Looks good to me. I can see this and other greek/mathematical symbols cropping up quite frequently when discussing outboard gear, so it would be good if the script could handle them all!

[grab]

although linear-phase filters will invariably be FIR filters

Not necessarily. Bessel filters...

Yes, but a Bessel doesn't really transform to digital form.

Yes, it really does. Not very long ago, I spent a couple of weeks crunching numbers to replace 4th-order IIR Butterworth notch filters with 4th-order IIR Bessel notch filters, because we thought it would improve our controller's step response. So I can testify first-hand that it does transform into digital, because I've done it.

(And yes, it did reduce the initial overshoot on steps when we really caned it. We decided not to use it for various other reasons, but hey, that's R&D for you.)

We did consider an FIR equivalent, but we don't have enough DSP grunt left to do a good enough job in FIR.