why would a highpass filter boost the overall level?

[dobro]

Why would a highpass filter boost the overall level?

I play the track dry, and it doesn't clip.

I apply a highpass filter with the cutoff point at about 90 Hz, and suddenly certain parts of the track clip. Why is that? It seems to me that if I'm cutting low end frequencies, the overall level of the track should be lower, right?

I'm using a plugin, the Ultrafunk EQ.

 

[eanderso13]

Does the plugin have an output level that's too high? Is it all going through a compressor plugin? Compressors tend to trigger on a lot of bass, so if you're cutting that out, the highs may not be trigering it as quickly.

Eric

 

[dobro]

"Does the plugin have an output level that's too high? Is it all going through a compressor plugin? Compressors tend to trigger on a lot of bass, so if you're cutting that out, the highs may not be trigering it as quickly."

There's no output level on the EQ plugin, so I think the answer to the first question is no.

Second question: it doesn't matter whether I put it through a compressor plugin or not, the clipping still happens. It's as if the highpass filter wasn't just cutting the lows, but boosting everything *above* the cutoff point as well. That isn't supposed to happen, is it?

 

[eanderso13]

hmmm

No, that shouldn't happen. But I have encounterred problems with different plugins. I one tried to use a certain multiband compressor plugin that slowed the track it was on down a few milliseconds every sample, so after a few seconds, the one track was totally out of time with the rest of the tracks.

Is this the Ultrafunk EQ in the Sonitus FX pack?(like the pic I attached)

If so, there is an output slider at the bottom. Try pulling that back a few notches and see what happens.

Eric

 

[dobro]

Damn, I don't believe it - that output slider's been there all along, and I've never even noticed it. Duh.

Well, what happens is this: when I cut the output level, it doesn't clip, but the overall level of the track is lower. So, yeah, it looks like the Ultrafunk EQ plug boosts the overall level when I cut the low end. Dumb.

 

[jake-owa]

Or it could be possible that certain low frequencies are being phase masked by the low end of that one track and when you cut it the masked frequency gets freed somewhere else in the mix to peak the stereo mix. Did you try applying the filter while soloing the track?

Not too likely but possible.

 

[BasPer]

Well, the high-pass filter will muck up the phase so previously destructive interference can now have become creative. So it is quite probable that some peaks will be higher even tough you have boosted no individual frequency. This is the way this shit works, a cut in the frequency domain can very well become a boost in the time domain, and vice versa! What do you think happens when you peak limit? Will there be more or less peaks in the frequency domain? Will the energy of the signal be higher or lower?

 

[mixsit]

Damn, I don't believe it - that output slider's been there all along, and I've never even noticed it. Duh.

Well, what happens is this: when I cut the output level, it doesn't clip, but the overall level of the track is lower. So, yeah, it looks like the Ultrafunk EQ plug boosts the overall level when I cut the low end. Dumb.

I was hoping to find out what turned out with this.
Was the gain slider set above 'zero' to cause the boost? How close to full scale was the track befor the eq?

The phase/masking shift idea don't compute. Wouldn't an effect like that have to be 'when combined' with a second path or something? (but from within a single track...?
Thanks all.
Wayne

 

[jake-owa]

Wayne, everything is combined in the stereo buss.


Though, I'm not clear if it was one track alone or a track in a mix...

 

[BasPer]

The phase/masking shift idea don't compute.

It computes very well indeed! Read my post above again a few times, and I think you'll figure it out. (I probably have the advantage of a few years more math schooling so don't be feel bad about it. ) And otherwise ask again, and I'll give a little more in depth explanation.

 

[dachay2tnr]

I find a similar thing happens to me as well when I use EQ (and not just a highpass filter). Heres the deal:

Let's say I have stereo mix that has been compressed and limited to -0.1 db. I can play it back, and the peak meter will show it peaks at -0.1db.

While listening to the mix I think it sounds a little harsh. So I want to hear what a 1db cut at 3K sounds like (Q=1.0). However, when I apply the cut, suddenly the mix begins to clip on the peak meter. Despite the fact that I did nothing but apply a cut (output gain btw is set to neutral at 0.0).

I understand the explanation of phase/masking and it's the only thing I've heard so far that makes any sense. But still seems wierd to me. I cut 1db from the mix across a range wider than an octave, and the overall mix increases in volume????

 

[BasPer]

Well you are asking for it when you have your levels up there! Why do you think it sounds harsh in the first place?
When you peak limited you created all sorts of harmonics and if the filter does any kind of analog modeling (and most do cause it sounds better) the filter will tame those harmonics a bit, causing the peaks to become higher. I would not be suprised if you got overs with the filter in the path even if you didn't touch the knobs. And if the filter in any way changes the phase your peak limiting is as good as undone! (Except that it probably sounded better berore.)

Class! Repeat after me: "If I neccessarily have to squeeze the life out of my recordings, I will at least never do any processing afterwards!"

 

[dachay2tnr]

Well you are asking for it when you have your levels up there! Why do you think it sounds harsh in the first place?
When you peak limited you created all sorts of harmonics and if the filter does any kind of analog modeling (and most do cause it sounds better) the filter will tame those harmonics a bit, causing the peaks to become higher. I would not be suprised if you got overs with the filter in the path even if you didn't touch the knobs. And if the filter in any way changes the phase your peak limiting is as good as undone! (Except that it probably sounded better berore.)

Class! Repeat after me: "If I neccessarily have to squeeze the life out of my recordings, I will at least never do any processing afterwards!"

< removes dagger from heart >

In defense of myself, I wasn't looking to do additional processing to that mix. Rather I was looking to see if I possibly wanted to go back and "remaster" the original mix, and, if so, what areas might work to help it out.

The above exercise was sort of a reality check to see what areas might warrant some additional eq work - BEFORE I went back and redid the orginal mix. However, in the process, I happened to notice that further eq cuts to mix resulted in clipping.

It was never my intent to do further processing to that particular mix.

 

[BasPer]

I realized that, but I still thought it would be a good point to make... And I have never been able to stop myself kicking those already on the ground. (Swedish stock phrase, not to be taken literally.)
You are forgiven!

 

[dachay2tnr]

You are forgiven!

Whew! That takes a load off my mind.

 

[barefoot]

dobro,

The filters seem to be giving you some signal gain. Filter Q's can effect this. Q's higher than 0.707 add gain. The higher the Q the more the gain. Some EQ's account for this gain and normalize the output of each band. This one apparently doesn't.

Try setting the Q in each band as low as possible. 0.71 is good. 0.5 is ideal, though this might actually cause an output drop with your plugin. Remember, this is a multi-band EQ. Unless you can disable the other bands, they are all effecting your signal - not just the high-pass section. Lower Q's cause less harm. Also, make sure the frequencies are spread out evenly (on the logarithmic scale) - sort of like those shown in Eric's picture. Closely spaced filters will also cause problems.

Thomas

 

[mixsit]

On steep filters, there is a rise befor the dip on the curve display. Is this the same as the gain Barefoot is refering to?
Wayne

 

[BasPer]

That actually has nothing (or rater not everything) to to with the steepness of the filter. The slope of the filter is affected by both the order and the type. For a given order of filter the slope will eventually (far into the stop-band) become the same, but you can make it more or less steep around the center frequency by choosing different types of filters. And the steeper types (chebylchev-types) ripples - giving peaks before the drop - while the less steep ones (like bessel) do not.

 

[barefoot]

One addition to what BasPer wrote:

Steep slopes give no free lunch even if the Q's are low. Steeper slope filters create more group delay (phase) distortion than shallower slope filters. Always try to use the shallowest slope, lowest Q filter that will work for what you're trying to achieve.

12 dB/octave, Q = 0.5 to 0.6 is an excellent compromise for many high or low pass applications.

Thomas