Frequency filter question

[mth5044]

The thought came from Rring's delay schematic, but it's applicable to other circuits that use filtering.


`
The MAX7401 is an 8 pole filter and with that 300pF cap, the corner frequency is set at ~1.3kHz with (I think) a -48dB/octave roll off. That's right up to about the root frequency a guitar can do, but all the harmonics and such can go up to 5-6kHz and beyond, no? So all that is just filtered away?

That has to be noticeable? Or maybe not since it's only on the delay'd bits?

[merlinb]

Or maybe not since it's only on the delay'd bits?

That's exactly it- you can apply a lot of filtering to the delayed sound. Natural acoustic spaces simply don't produce a flat echo response across the audio spectrum, so it sounds tinny and unnatural if you try to hang on to the full spectrum in an echo circuit. You have to filter out some highs to get a warm and realistic sound.

[mth5044]

Thanks for the info, Merlin! I appreciate it.

[PRR]

> an 8 pole filter ...set at ~1.3kHz ... -48dB/octave roll off.

Small/thin stuff absorbs highs.

Large/thick stuff absorbs lows.

The way it works, "most" rooms end up with more small-stuff than large-stuff.

So "most" rooms have longer reverb in bass than in treble.

A hasty search found this curve for the South Shore Baptist Church, but the trends are really similar for most useful spaces with significant 'verb.


(Actually this room had problems, and the discussion may be of some interest.)

Note that "natural" reverb decay frequency response has low slope, perhaps 3dB/Oct; and the (very conservative conventional) "ideal" curves are flat above 250hz.

48dB/Oct at 1.3KHz is NOT natural reverb!!

If you walked into a room with such a reverb you would feel funny. Boom and middle but no tizz at all. A smallish concrete room with canvas on all surfaces would approximate this, but canvas would be more like 6dB/Oct. A 48dB/Oct filter is very un-natural.

I assume this is a digital (or BBD) delay, with not-enough bits for good length, so the clock has been pushed well into the audio range and must be cut severely so we do not hear it.

Obviously the right answer is more bits. But that was historically expensive, and still annoying.

As you say, 1.3KHz covers a lot of the musical range. A balance favoring bass is not wrong. While you would not use it to sweeten full orchestra, and you would prefer not to be so abrupt on a delay-only job (radio profanity delay, or syncing distant speakers), as a behind-the-music effect it may be fine.