Vari-Q Bandpass Filter With Constant Gain

[JohnRoberts]

I did grep but I found the same stuff as PRR..

Sadly to get the most out of forum discussions from experienced posters you need to meet them halfway.

If you ask what NF means surely somebody else will answer.

I am with abbey, I probably use lots of jargon myself.

JR 

 

[mojozoom]

Thanks for all of the replies everyone!

This is something I want to do with an analog circuit, as I already have a DSP in the signal chain but it's not capable of all-pass filters. It's only able to handle EQ and time alignment.

The reason I'm looking for a bandpass filter with independently variable Q and Fo is that I can then incorporate it into an all-pass filter using the AP=1-2BP concept. Truly what I want is an all-pass with independently variable Q and Fo, the BP filter is just one way to get there and more commonly addressed than AP. It seems like AP gets a bad rap.

I modeled the state variable using the modification that Abbey advised and it seem to work very well. The circuit does limit the bottom end of the Q however, so as you keep increasing the Bandwidth resistor you don't continue to get a wider Q. Somewhere I thought I read that the State Variable was fixed at a Q of 2? I've read alot lately and have lost track a bit...

The next step would be to sum the input with the bandpass at a 1:2 ratio, but it may need to be inverted first I think. I'll mess with that model tomorrow.

If anyone knows of a simpler way to get a high quality all-pass filter with independently adjustable Q and Fo please let me know as well - I'll try anything.  I'll dig deeper into the DPOTs as well since that may be a good way to get a wider range of Q adjustment.

 

[abbey road d enfer]

I already have a DSP in the signal chain but it's not capable of all-pass filters. It's only able to handle EQ and time alignment.

Why is that? Any DSP should be capable of producting all sorts of AP filters. Is it because of teh user interface?

The reason I'm looking for a bandpass filter with independently variable Q and Fo is that I can then incorporate it into an all-pass filter using the AP=1-2BP concept. Truly what I want is an all-pass with independently variable Q and Fo, the BP filter is just one way to get there and more commonly addressed than AP. It seems like AP gets a bad rap.

I don't really get it. You'll end up with an AP filter anyway. What "bad rap"?

I modeled the state variable using the modification that Abbey advised and it seem to work very well. The circuit does limit the bottom end of the Q however, so as you keep increasing the Bandwidth resistor you don't continue to get a wider Q. Somewhere I thought I read that the State Variable was fixed at a Q of 2? I've read alot lately and have lost track a bit...

Since it's a single biquad response, the asymptotes eventually tail out at 6dB/octave. It may or may not matter, depends if you're interested in what happens when far enough from the characteristic frequency.

If anyone knows of a simpler way to get a high quality all-pass filter with independently adjustable Q and Fo please let me know as well

I don't think there is and I've dabbed with all sorts of EQ circuits.

  I'll dig deeper into the DPOTs as well since that may be a good way to get a wider range of Q adjustment.

I don't think you really have a wider range by using an alternate pot. If you want a wider BW, you have to load the inverting input of the summer opamp, ,as JR already mentioned.

But... what are you trying to achieve? Is it for speaker alignment? Using AP filters for speaker alignment is a tried and proved technique.

 

[efinque]

I think "if Scatman can do it, so can you" is a good monday motivation and a general DIY philosophy.

EDIT : sorry, I'm not really helping here but I really have a hard time understanding why someone would want a constant gain bandpass filter..

EDIT 2 : unless it's a highly specialized application for finding resonant peaks in signals, but then a 3-band EQ with a sweepable midrange and mute switches in the HF/LF would do plus you could use it for a number of other applications.

 

[abbey road d enfer]

EDIT :  I really have a hard time understanding why someone would want a constant gain bandpass filter..

You probably haven't been facing the problems associated with amplitude varying with BW, then.

EDIT 2 : unless it's a highly specialized application for finding resonant peaks in signals, but then a 3-band EQ with a sweepable midrange and mute switches in the HF/LF would do plus you could use it for a number of other applications.

The OP wants to build an All Pass filter. The common topology uses a filter (could be Lp, Hp or Bp) that has unity gain, which is substracted to the input signal in a 2:1 ratio. The result is a filter that has flat amplitude all over, but has variable phase. The consequence of variable phase is variable propagation time. AP filters are often used in a range where the time delay is pretty much constant. Until Bucket Brigade Delays and DSP became practical, AP filters were the only solution to produce delay electronically (tape/disk is electromechanical and acoustic delay lines are...electro-acoustic).
The amount of delay is usually very small (about 5us for a 20kHz BW).
AP filters were used, among other applications, for compensating the 22us delay between the left and right outputs of early stereo DAC's.
They are still used today for alignment of speaker systems, in particular to compensate the phase-shift introduced by cross-overs in multiway systems.

 

[efinque]

You probably haven't been facing the problems associated with amplitude varying with BW, then.
The OP wants to build an All Pass filter. The common topology uses a filter (could be Lp, Hp or Bp) that has unity gain, which is substracted to the input signal in a 2:1 ratio. The result is a filter that has flat amplitude all over, but has variable phase. The consequence of variable phase is variable propagation time. AP filters are often used in a range where the time delay is pretty much constant. Until Bucket Brigade Delays and DSP became practical, AP filters were the only solution to produce delay electronically (tape/disk is electromechanical and acoustic delay lines are...electro-acoustic).
The amount of delay is usually very small (about 5us for a 20kHz BW).
AP filters were used, among other applications, for compensating the 22us delay between the left and right outputs of early stereo DAC's.
They are still used today for alignment of speaker systems, in particular to compensate the phase-shift introduced by cross-overs in multiway systems.

I think the problem comes with the unity gain, ie getting the Q value/amplification ratio right so that it holds 0dB output which I think would be achievable only with an auto-make up gain correction DSP (if one was to design an analog circuit I'd figure it would be nearly impossible)

EDIT : and even so I think it'd be only useful for constant amplitude, not so much for program material..

 

[abbey road d enfer]

I think the problem comes with the unity gain, ie getting the Q value/amplification ratio right so that it holds 0dB output

That is not an issue at all.

  which I think would be achievable only with an auto-make up gain correction DSP

  No need for DSP; AP filters existed way before DSP became practical.

  if one was to design an analog circuit I'd figure it would be nearly impossible

And you'd be wrong. Many pieces of gear based on AP filters exist and work.

EDIT : and even so I think it'd be only useful for constant amplitude, not so much for program material..

Why would it be so? I have proof that AP filters work as well with program than sinewaves or squarewaves.

 

[efinque]

That is not an issue at all.
And you'd be wrong. Many pieces of gear based on AP filters exist and work.

EDIT : and even so I think it'd be only useful for constant amplitude, not so much for program material..
Why would it be so? I have proof that AP filters work as well with program than sinewaves or squarewaves.

A stripped-down multiband compressor? I'm thinking it in terms of a reversed de-esser.. (edit : although they rarely operate in the 100Hz range..)

EDIT : I just tried a Reaktor patch using comparators for the EQ input/output signals feeding a relay to a compressor but for some reason it didn't work..

 

[abbey road d enfer]

A stripped-down multiband compressor? I'm thinking it in terms of a reversed de-esser.. (edit : although they rarely operate in the 100Hz range..)

EDIT : I just tried a Reaktor patch using comparators for the EQ input/output signals feeding a relay to a compressor but for some reason it didn't work..

Please, this is going nowhere; your gibberish is of no interest. Try to turn on the brains before posting.

 

[JohnRoberts]

Please, this is going nowhere; your gibberish is of no interest. Try to turn on the brains before posting.

or just stop answering, expecting a different outcome this time.

It is almost impossible to change other people's behavior on the WWW, I can barely change my own.

JR

 

[bruno2000]

NetEQ w/different BW settings.
Best,
Bruno2000

 

[mjrippe]

To say, yes, to write, no.

If You google "acronym NFB, it's at the third line... How tough is it?

negative feedback = 17 characters including the space.

When in doubt, ask Google first. = 32 characters, nearly twice as much time and effort.

The real point being that folks here are usually friendly and accommodating.

 

[abbey road d enfer]

negative feedback = 17 characters including the space.

When in doubt, ask Google first. = 32 characters, nearly twice as much time and effort.

The real point being that folks here are usually friendly and accommodating.

My real point being: "Give a man a fish, you feed him for a day, teach him how to fish, you feed him for a lifetime".
Now if you think that inciting people to improve their searching abilities is un-friendly and un-accomodating...

 

[mjrippe]

My real point being: "Give a man a fish, you feed him for a day, teach him how to fish, you feed him for a lifetime".
Now if you think that inciting people to improve their searching abilities is un-friendly and un-accomodating...

Telling someone "just google it" is not teaching.  If you told them the acronym stood for negative feedback and then explained what that concept meant, that would be teaching.  "Just google it" is dismissive and sometimes considered rude.  I have seen some very knowledgeable members of this forum ask about acronyms in a thread and nobody told them to google it. 

Sorry to make this a public thing, but I would expect moderators to set an example for the forum.  "Try to turn on the brains before posting.", really?

 

[abbey road d enfer]

Sorry to make this a public thing, but I would expect moderators to set an example for the forum. 

I believe telling someone to do some research on his own is a good example. We're supposed to be (almost) adults.

  "Try to turn on the brains before posting.", really?

If you read the posts I was referring to, you may understand that my duty, as a moderator, was to stop them; some kind of an ultimatum before blocking his posts.

 

[mojozoom]

I was thinking that the DPOTs could be used as John was saying to spread the two resistances. But in looking back at the previous all-pass filters I've used it looks like I only need a Q of about 1 to 5, and a frequency range of 50 to 1000 hz. This circuit seems to meet those ranges with 10k pots for the BW and (ganged) for the frequency.

I use the all-pass filters to compensate for the phase response of door mounted 6.5" midbass drivers in my car. I'll post some measurements in a bit so you fellows can see the issues. The all-pass filters work, but there are some limitations. What I'm trying to do now is build a circuit that allows me to adjust the all-pass on the fly, and bypass it completely for comparison purposes too.

 

[mjrippe]

I believe telling someone to do some research on his own is a good example. We're supposed to be (almost) adults.
If you read the posts I was referring to, you may understand that my duty, as a moderator, was to stop them; some kind of an ultimatum before blocking his posts.

I am not arguing with the idea of "do your homework before asking questions", but rather the harsh tone.  A member posting rambling thoughts does not need to be told to shut up, nor do they need their posts blocked.  If you are using words like "incite" and "ultimatum" then you might want to take a step back and reconsider why you want to be a moderator.

 

[CJ]

well, if everybody knew everything there would be no need for a forum, which is a place to help people.  which is also our main purpose in life,

forum burnout, waiting for the questions to stop, but it never happens, usually means time for a break.

our job, as spelled out by the great spirit, is to love everybody, especially the ones we do not like. if you can think of a more noble purpose for life then i would sure like to hear it,

now where is that trump thread located?  ;D

 

[JohnRoberts]

I am not arguing with the idea of "do your homework before asking questions", but rather the harsh tone.  A member posting rambling thoughts does not need to be told to shut up, nor do they need their posts blocked.  If you are using words like "incite" and "ultimatum" then you might want to take a step back and reconsider why you want to be a moderator.

Abbey is a major contributor around here. 

Being a mod is not a hand holding teaching position, but more like a janitor or dog catcher. 

--------

@ CJ Sadly the Trump thread is where it always was...

JR

 

[abbey road d enfer]

I was thinking that the DPOTs could be used as John was saying to spread the two resistances. But in looking back at the previous all-pass filters I've used it looks like I only need a Q of about 1 to 5, and a frequency range of 50 to 1000 hz. This circuit seems to meet those ranges with 10k pots for the BW and (ganged) for the frequency.

I use the all-pass filters to compensate for the phase response of door mounted 6.5" midbass drivers in my car. I'll post some measurements in a bit so you fellows can see the issues. The all-pass filters work, but there are some limitations. What I'm trying to do now is build a circuit that allows me to adjust the all-pass on the fly, and bypass it completely for comparison purposes too.

Is it the phase response of the speakers, or the phase response of the speakers + cross-overs? It is important to figure out what causes the phase-shift. Very often, cascaded 1st-order APF are preferred to higher-order, because they are unconditinally stable and they are very easy to align, using a single pot.