can anyone make sense of this?

i dropped a song from itunes into logic, and when played it peaks at 0db like it should. so i opened up an instance of the channel eq and turned on the highpass filter and took away a bunch of the bass, i didnt boost any frequencies. and now its peaking at +2.5 or somthing db. this dosent make any sense to me, anybody know whats going on??

thanks

Jack

Well, yes, I can imagine what happens, but I'm afraid it's a little bit abstract. I will try to explain anyway.

First of all, notice a higher dB value related to waveform peaks (dB FS) doesn't mean there is more "energy" in the wave (which is represented by dB RMS). To illustrate this, think of a square wave. One half of the time it remains at its maximum, the rest of the time it is at its minimum. It jumps periodically from one value to the other and back, and the jump's height is its amplitude.

Now send this square wave through a highpass filter (the traditional kind of filter, electronically realised with one capacitor and one resisitor, without any "linear phase" stuff), what will you get? The "jump", the edge remains mostly unchanged and thus keeps its height, but between the jumps the wave always asymptotically returns to zero, depending on the filter's assigned frequency, so the starting points of the jumps up and down are nearer to zero than before, so the jump up goes higher and the jump down lower than before. The distance between the peaks has increased, although there is something (the lower frequencies) missing in the filtered wave!

Unfortunately I could not find appropriate pictures in the manual, but I hope you can use your imagination to see what I tried to explain...

If not: Just believe it. Hope that helps.

Ok, I believe three pictures say more than 134 words, so I took a (real world) function generator, a (real world) oscilloscope and some resistors and a capacitor. Here are some screenshots.

The first picture shows the unfiltered square wave with a frequency of 1kHz. For the second screenshot, the wave was filtered by a combination of a 3.3nF capacitor and a 100k resistor. And the third one shows a strongly filtered wave (3.3nF/10k).

You can watch two things here.

Firstly: The more the wave is filtered, the less area is left between the wave and the zero line.

Secondly: The edges pass the filter unchanged, but because of the filtering the starting points of the jumps go near to zero. While the amplitude (peak to peak) is two units for the unfiltered wave, it is as much as four units with strong filtering.

Be aware that a combination of passice components as resistors and capacitors can never "add energy" - quite the opposite resistors consume energy, i.e. pull it out of a circuit and convert it to heat.

Remark: The three waveforms are mounted into one picture now (2008-1-21).

Second Remark: Uploaded again... (2012-1-17)

Doesn't make any sense to me either. No clear reason overall amplitude would increase after you do nothing but attenuate a certain frequency range.

How about this...

The EQ plug in adds a marginal amount gain just by inserting it on the channel strip, combined with the fact that the output meter is not 100% accurate. It can meter a dB that isn't 100% "true life"

Just curious...what song is this?

Jope is right. You may want to read this earlier discussion on the topic, I had posted a pretty eloquent picture (all the way down the page).

why does decreasing EQ increase the main meter level?

BTW since then Ski and I saw that effect on an Oscilloscope we plugged at the output of Logic's soft synths. You lower the high frequencies a bit, and you end up with a signal that has a higher peak level despite a lower RMS level.

The effect is pure math, and has nothing to do with Logic in particular. You would see the same effect with an analog synth or any sound source and any analog filter.

Jope is right.

Thanks

Um, David, do you know why my uploaded pictures disappeared after a while? Lastly I mounted them together into one large picture to preserve the order. I hope the picture will stay there now.

You may want to read this earlier discussion on the topic, I had posted a pretty eloquent picture (all the way down the page).

 

why does decreasing EQ increase the main meter level?

 

BTW since then Ski and I saw that effect on an Oscilloscope we plugged at the output of Logic's soft synths. You lower the high frequencies a bit, and you end up with a signal that has a higher peak level despite a lower RMS level.

 

The effect is pure math, and has nothing to do with Logic in particular. You would see the same effect with an analog synth or any sound source and any analog filter.

I like those "real world" experiments better than software simulations in some cases as you can be sure there's no cheating by any phase or gain compensation algorithms.

Jope is right.

Yup!

The effect is pure math, and has nothing to do with Logic in particular. You would see the same effect with an analog synth or any sound source and any analog filter.

David's right too!

Just to add to this, most of the digital filters we use in Logic are IIR filters which really do nothing more than emulate their analog counterparts, thus they behave the same. An FIR digital filter won't necessarily exhibit this change in crest factor, but you have to be wary as some FIR linear eq's use an additional algorithm to add this back in to recreate an analog sounding filter while maintaining linear phase.