another eq question.(sorry guys)

[cake1122]

another term i was wondering about is rolling of a frequency. is that when u just take it down to like -5o dbs at a certan place

thanks

 

[SouthSIDE Glen]

"Rolling off" usualy referrs to a downward slope in an EQ curve, where the amount of EQ cut increases at a regular rate as frequency increases or decreases.

For example, if I told you to roll off your highs at around 5kHz, that would usually mean to apply a low-pass filter to the signal were everything below 5kHz gets past unimpeded, but starting at 5kHz the signal is attenuated at a particular increasing rate (e.g. 6dB/octave, 12dB/octave, etc.)

The same can be done on the other end with a high-pass filter where one can roll off the low end below a certain knee frequency as well.

G.

 

[RedStone]

the Roll off describes the shape of the slope on a filter
looking for a picture to help describe it ......

http://www.practicalpc.co.uk/computing/sound/record3.htm

there's a picture on that page that shows different roll offs

 

[cake1122]

thanks guys. so for exaple if i was to roll of my guitar track at 80 hz. i would cut everything 80 and below?

 

[Ford Van]

Yes, at a rate determined by the slope of the filter...ie 6dB, 12dB, 18dB, 24dB The bigger the number, the more extreme the cut.

For instance, 40Hz is one octave below 80Hz. Thus, on a 6dB per octave slope, it would be 6dB lower. 20 Hz would be 12dB lower

On a 12dB per octave slope, 40Hz would be 12dB lower, but 20Hz would be 24dB lower!

60Hz would be down about 3dB with a filter set at 80Hz with a 6dB per octave slope. There IS the taper of the filter too, which may or may not be linear. I believe they generally are.

 

[FALKEN]

For example, if I told you to roll off your highs at around 5kHz, that would usually mean to apply a low-pass filter to the signal were everything below 5kHz gets past unimpeded, but starting at 5kHz the signal is attenuated at a particular increasing rate (e.g. 6dB/octave, 12dB/octave, etc.)

I was always under the impression that the corner frequency of a high or low pass filter was where that frequency was 3 db down. Meaning that if you "applied" a low-pass filter at 5khz the signals below 5khz DO NOT pass "unimpeded" at all.

 

[SouthSIDE Glen]

I was always under the impression that the corner frequency of a high or low pass filter was where that frequency was 3 db down. Meaning that if you "applied" a low-pass filter at 5khz the signals below 5khz DO NOT pass "unimpeded" at all.

On some (many?) filters that can be true, but it depends greatly upon the filter design. There are some hi/lo pass filters that actually throw a bit of a "bump" or boost at or near the knee/corner/cutoff frequency, with the amount of boost inversely proportional to the steepness of the roll-off slope (i.e. the steeper the cut-off slope, the higher the boost around the knee.) Other filters are designed with a soft knee where the beginning of the slope pulls back (or forward) from the knee frequency at a distance proportional to the steepness of the slope.

Those are all details that indeed can change depending upon filter design, and in that light, yeah, you're right that my definition implying a hard knee at the corner frequency is not quite accurate. The same is true of the graphs shown in RedStone's link, which illustrate a hard knee. There are very few hi/low pass filters where the knee is actually that hard.

But the actual character of the slope immediately surrounding the set corner frequency changes not only from filter to filter, but often even within the same filter epending upon roll-off slope setting. So to keep it simple, I just described a simplified explanation of "roll off".

G.