Filter specs?

DSP, Plug-in and Host development discussion.
RELATED
PRODUCTS
Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Mon Nov 28, 2022 10:20 pm

mystran wrote: Mon Nov 28, 2022 1:03 pm We see hard clipped resonance, but +40dB is probably too far to extreme to make heads or tails about the actual nature of the distortion, which might not be quite that hard with less extreme settings.. but like you should also try with higher input gain and what not to try to build a more complete picture of how the filters behave in different situations... and then you basically start guesswork.
Can't go over +40db with FX section hehe, its the max. So can't try higher values.

Do you mean "resonance" on a Peak filter due to its gain stage correct?

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Nov 29, 2022 12:41 am

@mystran: if this help, here's the response using FX freq param 700hz and gain param +10db, instead of +40db:

Image

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Nov 29, 2022 3:47 am

Also: can't really find a peak filter "with resonance" (as plugin/filter in general).

Maybe am I misunderstanding?

Usually Peak Filter have freq and gain controls (not q or whatever), and the more you gain, the more freqs around the cutoff point increase.

Any existing which I can try? Do they have a specific name?

matt42
KVRian
1265 posts since 9 Jan, 2006

Post Tue Nov 29, 2022 6:15 am

Derozer wrote: Tue Nov 29, 2022 3:47 amUsually Peak Filter have freq and gain controls (not q or whatever)
Actually they do. A lot of equalizers feature a q control for peak filters. There are also equalizers that hide this parameter from the user, for ease of use, the developer wanting the filter to operate at a q that they consider musical etc

Edit: Some plugin companies marketed expensive EQs modeled on different units - EQ A modeled on unit x, famous for its warm musical curves, and EQ B modeled on unit y with its tight surgical bands. In reality, these were often just generic filters the 'modeling' simply being the internal q settings. Although there does seem to be a lot more serious attempts at modeling these days.

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Nov 29, 2022 7:32 am

matt42 wrote: Tue Nov 29, 2022 6:15 am
Derozer wrote: Tue Nov 29, 2022 3:47 amUsually Peak Filter have freq and gain controls (not q or whatever)
Actually they do. A lot of equalizers feature a q control for peak filters. There are also equalizers that hide this parameter from the user, for ease of use, the developer wanting the filter to operate at a q that they consider musical etc

Edit: Some plugin companies marketed expensive EQs modeled on different units - EQ A modeled on unit x, famous for its warm musical curves, and EQ B modeled on unit y with its tight surgical bands. In reality, these were often just generic filters the 'modeling' simply being the internal q settings. Although there does seem to be a lot more serious attempts at modeling these days.
I expressed myself in a wrong way :)

Yes, Peak Filter has Q of course (to change smaller or larger band around cutoff), but never seen it with resonance (as Low pass for example, which can start to self-oscillate with higher Q).

Usually, q on Peak filter just increase/decrease the bell :) Never seen self-oscillation on a Peak filter (neither which clip over some levels).

matt42
KVRian
1265 posts since 9 Jan, 2006

Post Tue Nov 29, 2022 7:37 am

I guess mostly highly resonant peak filters aren’t considered that useful, but with extreme enough q they can go from resonant to self oscillating to blowing up.

In an typical analog design the poles (which dictate resonance) are the same for a low pass filter and a peak filter

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Nov 29, 2022 7:45 am

matt42 wrote: Tue Nov 29, 2022 7:37 am I guess mostly highly resonant peak filters aren’t considered that useful, but with extreme enough q they can go from resonant to self oscillating to blowing up.

In an typical analog design the poles (which dictate resonance) are the same for a low pass filter and a peak filter
The strange things is that here there isn't any Q param on the FX unit, only gain.
Increasing the gain, it seems Q is not changing (i.e. fixed value, as you said before).

But starting from some gain point (it seems near +15db), it introduce "distortion".

matt42
KVRian
1265 posts since 9 Jan, 2006

Post Tue Nov 29, 2022 7:55 am

Something is clipping at a threshold set around 15 dB, this doesn’t have to have anything to do with resonance.

In a resonant filter increased resonance boosts the gain close to the center frequency which can exceed the clipping threshold and increase distortion. The clipping may be necessary to prevent the filter exploding.

This is why you might associate resonance and distortion. But anything that increases gain can push things into clipping and distort

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Nov 29, 2022 11:40 am

matt42 wrote: Tue Nov 29, 2022 7:55 am Something is clipping at a threshold set around 15 dB, this doesn’t have to have anything to do with resonance.

In a resonant filter increased resonance boosts the gain close to the center frequency which can exceed the clipping threshold and increase distortion. The clipping may be necessary to prevent the filter exploding.

This is why you might associate resonance and distortion. But anything that increases gain can push things into clipping and distort
I'll have some test and let you know ;)

camsr
KVRAF
7252 posts since 17 Feb, 2005

Post Tue Nov 29, 2022 2:02 pm

A peaking filter is actually a composite filter... it is made of more than 1 integrator. That means it is a mixture of responses between a lowpass, a highpass, and the input signal (as well as feedback). A self-oscillating peaking filter might have 2 self-oscillating filters creating that effect... I say might because there is also the allpass filter, which is also made of 2 integration stages. Bandpass filters are composite filters.

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Tue Dec 06, 2022 2:06 am

I've put in a chain a biquad filter (100/10khz, +-40db) and later an hard clip to the path, but the result is very different.

For example, at 100hz +40db, an hard clip doesn't shape the input saw this way:

Image

Instead, I got this (which is very fair of the target):

Image

Maybe some wave folder instead of hard clip?

Z1202
KVRian
1463 posts since 12 Apr, 2002

Post Fri Dec 09, 2022 12:25 am

Try dry/wet mix

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Fri Dec 09, 2022 1:53 am

Z1202 wrote: Fri Dec 09, 2022 12:25 am Try dry/wet mix
Nothing change :( It also increase the whole signal, over the threshold (since the dry signal have not any wall to smash against). And the shape become more "smoothed" mixing signal+peak vs signal+peak+hardclip.

Should be somethings other...

kerfuffle
KVRer
18 posts since 2 Jul, 2021

Post Fri Dec 09, 2022 3:53 am

Have you tried using PluginDoctor to analyze the signal?

Derozer
KVRist
Topic Starter
131 posts since 18 Nov, 2017

Post Fri Dec 09, 2022 4:43 am

kerfuffle wrote: Fri Dec 09, 2022 3:53 am Have you tried using PluginDoctor to analyze the signal?
Its an instrument, not a FX.

Return to “DSP and Plug-in Development”