[PD] bandpass or resonant?

Alexandre Torres Porres porres at gmail.com
Tue Jan 13 16:38:42 CET 2015


"*Ultimately I think "resonant" is a general descriptive term for filters,
that shouldn't be interpreted as a detail of implementation.*"

I guess you have a point there, and I was also driving to this conclusion.

"*It sounds like the Resonz UGen in supercollider is exactly what Julius
Smith is talking about in that description of the two-pole filter.*"

Actually, the Resonz UGen is a Two Pole / Two Zero filter, which is very
more closely related to [cyclone/reson~].

A Two-Pole only filter (no zeros) like the one Julius is describing is
actually what the [bp~] object is!

"*But then there's the other supercollider filter UGens with "resonant" in
the name, which seem more like what Martin was describing - RLPF (resonant
low-pass filter) for example is a low-pass filter where you can adjust the
resonance near the cutoff.* "

Yep, and this is also much like the [cyclone/lores~] object in Pd. Check
that patch I sent for more detailed info on these filters.

I've always assumed that a filter, in order to be called a "resonant
filter" or a "resonator" - being it a low pass, a high pass or a band pass
-, needed to boost/add gain to a particular cutoff frequency (in the case
of lowpass and highpas - which is the case for [lores~] or RLPF) or add
gain to some center frequency (in the case of a bandpass) - which is also
called "resonant frequency".

The quote from Julius in that link - where he says "*A resonator is a
recursive filter that boosts signal amplitude at a particular frequency*" -
is in line with my assumption.

But the concept of resonance in physics, in its utmost purity according to
wikipedia, is that it "is the tendency of a system to oscillate with
greater amplitude at some frequencies than at others". Meaning that it
doesn't really have to add gain to something, but only favor a frequency
amongst others... In this context, a bandpass - in general - is a
"resonator"...

But it'd be cool if I could find a definitive word about this in the filter
literature!

Cheers

2015-01-13 13:16 GMT-02:00 Brian Fay <ovaltinevortex at gmail.com>:

> It sounds like the Resonz UGen in supercollider is exactly what Julius
> Smith is talking about in that description of the two-pole filter.
>
> But then there's the other supercollider filter UGens with "resonant" in
> the name, which seem more like what Martin was describing - RLPF (resonant
> low-pass filter) for example is a low-pass filter where you can adjust the
> resonance near the cutoff. I haven't played with this too much myself, but
> I'm guessing with the right Q value you could drive this to
> self-oscillation?
>
> Ultimately I think "resonant" is a general descriptive term for filters,
> that shouldn't be interpreted as a detail of implementation.
>
> On Tue, Jan 13, 2015 at 9:28 AM, Alexandre Torres Porres <porres at gmail.com
> > wrote:
>
>> I'm pending to say that there is no real distinction between "Resonant
>> filter" and a "resonator", and a "bandpass" can be implicitly thought of as
>> a resonator. Here's what I also found in Julius' website
>>
>>
>> https://ccrma.stanford.edu/~jos/filters/First_Order_Complex_Resonators.html
>>
>> Pass the mouse cursor over the "Resonator" over the title "First-Order
>> Complex Resonators
>> <https://ccrma.stanford.edu/~jos/filters/Two_Pole.html>" to see the
>> popup (also attached).
>>
>> cheers
>>
>> 2015-01-13 1:20 GMT-02:00 Martin Peach <chakekatzil at gmail.com>:
>>
>> I was looking at circuit diagrams for analog synthesizers recently and
>>> noticed that the "resonance" control is nothing more than feeding some
>>> fraction of the output back to the input. With more feedback oscillation
>>> occurs at the cutoff frequency for any type of filter, highpass, bandpass
>>> or lowpass.
>>>
>>> Martin
>>>
>>>
>>> On Mon, Jan 12, 2015 at 9:09 PM, Alexandre Torres Porres <
>>> porres at gmail.com> wrote:
>>>
>>>> Nice I give an impression to be an expert, but filters is just
>>>> something I've actually recently started studying :)
>>>>
>>>> > I'm wondering if by "resonant" filter you mean the
>>>> > same thing as "resonator" filter?
>>>>
>>>> Now you got me... good question, and I'm not sure, haha. The link looks
>>>> nice btw, will definitely check it. Thanks.
>>>>
>>>> So now I'm even more confused. Is "resonant filter" and "resonator" two
>>>> different concepts? Maybe I'm having trouble with the english nomenclature
>>>> and everything.
>>>>
>>>> To be honest and more detailed about the issues I'm encountering, I ask
>>>> this based on another topic I was discussing with Julius Smith in the
>>>> Supercollider list, but it went dead and I got no replies. In it I was
>>>> asking if the object "Resonz" should really be called a "Resonant filter",
>>>> because it was just a bandpass filter in my opinion. Then Julius was
>>>> mentioning how "*A resonator is a special case of a passband filter
>>>> having a nearly zero-width passband.*"
>>>>
>>>> I see he used the term "resonator" and not "Resonant Filter" (as Resonz
>>>> is described). So yeah, now I'm more confused... is resonator the same as
>>>> resonant or what?
>>>>
>>>> But anyway, we can bring the discussion into the Pd world, and talk
>>>> about the [reson~] object, as I will do later on.
>>>>
>>>> I was googling and saw how the term resonant filter could be used to
>>>> describe a regular bandpass filter. And how the bandpass' center frequency
>>>> could also be called "resonant frequency". So they might be used in the
>>>> same way... (accurately or not).
>>>>
>>>> Now here is my opinion. Just like a resonant low pass filter (the
>>>> [lores~] object in Pd), the concept of resonance in a filter relates to how
>>>> it adds gain around the resonant frequency.
>>>>
>>>> In the Audio-EQ-Cookbook (link:
>>>> http://www.musicdsp.org/files/Audio-EQ-Cookbook.txt ) that presents
>>>> formulas for biquad coeficients you have two different bandpass filters,
>>>> lets call them BPF1 and BPF2. So, BPF2 has "constant 0 dB peak gain",
>>>> meaning it doesn't affect anything arounf the center frequency. Now BPF1
>>>> says it has "constant skirt gain, peak gain = Q", meaning that the Q
>>>> or bandwidth controls the gain of the filter. I consider BPF2 to be a
>>>> regular bandpass filter, whereas BPF1, which adds gain for narrower
>>>> bandwidths, seems to be a "resonant" one... (which makes me think Resonz
>>>> shouldn't be described as resonant filter, as it's just a bandpass, or
>>>> "BPF2").
>>>>
>>>> Oh, there's another term around, the "ringing" filter, which seems to
>>>> be another term for resonant filter. In SuperCollider they have Ringz,
>>>> which was supposed to be the same as Resonz object (or a resonant filter
>>>> for that matter), but they are different like the two different kinds of
>>>> bandpass in the EQ Cookbook (Ringz = BPF1 / Resonz = BPF2).
>>>>
>>>> Coming into the Pd world we can talk about the [reson~] object. As the
>>>> name implies, it is a resonant filter. But the helpfile says it is a "Bandpass
>>>> filter" (damn). Funny enough, in Max, the [reson~] object is said to be
>>>> indded a *Resonant Bandpass Filter*. So maybe we should update [reson~]'s
>>>> help file in Pd... But the deal is: [reson~] is actually a bandpass like
>>>> BFP2 or Resonz, but it has a separate parameter for the gain. Meaning it
>>>> works basically as a bandpass filter, where changing the Q doesn't affect
>>>> the gain. But you can also give it a boost or a cut with the gain
>>>> parameter. By giving it a boost it would behave in a way that I'm
>>>> considering to be an actual "resonant" filter.
>>>>
>>>> Now let me go ahead and share a patch that I'm writing for my computer
>>>> music classes. It's about several filters that can be obtained with biquad.
>>>> So I present Pd's vanilla filters such as [lop~], [hip~] and [bp~]. I also
>>>> present externals like [lores~] and [reson~] and I do present all the
>>>> filters from the Audio Eq Cookbook as well. It's in portuguese, and part of
>>>> a big series of examples, but what the hell...
>>>>
>>>> By the way, I was also able to implement Resonz and Ringz as [biquad~]
>>>> in Pd, but I don't have it on this example (but to hell with supercollider
>>>> already, hehe).
>>>>
>>>> So there you can check the behaviour and differences that I've pointed.
>>>> For [reson~], I have two separate parameters, one is for "Q", and the other
>>>> one, which is originally "gain", I'm calling the "resonance" parameter.
>>>> Much like the resonance parameter in [lores~]...
>>>>
>>>> As for the Cookbook filters, BPF2 is what I'm calling a regular
>>>> "bandpass", and it has a parameter of "Q". As for BPF1, I'm calling it a
>>>> "resonant filter", and it has a parameter of "resonance" instead of "Q"
>>>> (but the idea is that more Q gives more gain/resonance).
>>>>
>>>> I hope I'm clear in the midst of so much detail and information. It's
>>>> just a stupid doubt on the nomenclature of filters, but this kind of issue
>>>> can be quite a pain in the ass, and the subject of many confusions.
>>>>
>>>> So, in short, I still keep my original question: Are bandpass and
>>>> resonant filters the same? Or is there a difference between calling one a
>>>> resonant and not a bandpass? Moreover, what about a resonator? Is my
>>>> assumption on how to call one a bandpass and a resonant correct or it
>>>> doesn't have anything to do with the official literature? And how about
>>>> what Julius Smith had to say?
>>>>
>>>> Cheers
>>>>
>>>> 2015-01-12 20:03 GMT-02:00 Brian Fay <ovaltinevortex at gmail.com>:
>>>>
>>>> Based on your posts in this group, you definitely know more about
>>>>> filters than I do, so I don't really have an answer to this question - but
>>>>> I'm wondering if by "resonant" filter you mean the same thing as
>>>>> "resonator" filter?
>>>>>
>>>>> I saw something interesting earlier today about using resonator
>>>>> filters as control sources for FM synthesis:
>>>>> http://tai-studio.org/index.php/projects/sound-programming/complexres/
>>>>>
>>>>> The paper that is linked in the article has some details on the
>>>>> mathematical implementation of the filter, but that's mostly mumbo-jumbo to
>>>>> me.
>>>>>
>>>>> On Mon, Jan 12, 2015 at 4:24 PM, Alexandre Torres Porres <
>>>>> porres at gmail.com> wrote:
>>>>>
>>>>>> Hi, are bandpass and resonant filters the same? Or is there a
>>>>>> difference between calling one a resonant and not a bandpass?
>>>>>>
>>>>>> thanks
>>>>>>
>>>>>> _______________________________________________
>>>>>> Pd-list at lists.iem.at mailing list
>>>>>> UNSUBSCRIBE and account-management ->
>>>>>> http://lists.puredata.info/listinfo/pd-list
>>>>>>
>>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> Pd-list at lists.iem.at mailing list
>>>> UNSUBSCRIBE and account-management ->
>>>> http://lists.puredata.info/listinfo/pd-list
>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20150113/f71d273e/attachment-0001.html>


More information about the Pd-list mailing list