Explain F Scale in this Video of Multiple feedback Bandpass...

[amdx]

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
> https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek

 

[Tauno Voipio]

On 22.8.22 17.08, amdx wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek

He just changes the frequency scale between radians/s and cycles/s
(which is Hz).

One full cycle is 2 pi radians, so 500 Hz will be 2 pi times as much,
which is about 3140 radians/s.

--

-TV

 

[Ricky]

On Monday, August 22, 2022 at 10:08:57 AM UTC-4, amdx wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

The part you misunderstood is thinking the resistor values calculated up to that point were for 500 Hz. It\'s not, it\'s for 1 rad/s which is around 0.16 Hz. The \"F scaling\" is frequency scaling to the frequency in your design rather than 1 rad/s.

https://youtu.be/UtnRQVeeLwM?list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&t=493

Keep in mind that you also need to \"scale\" the component values to real capacitor values. Not many use 1F capacitors.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

 

[amdx]

On 8/22/2022 9:43 AM, Ricky wrote:

On Monday, August 22, 2022 at 10:08:57 AM UTC-4, amdx wrote:
I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?
The part you misunderstood is thinking the resistor values calculated up to that point were for 500 Hz. It\'s not, it\'s for 1 rad/s which is around 0.16 Hz. The \"F scaling\" is frequency scaling to the frequency in your design rather than 1 rad/s.

 OK, that is helpful.

https://youtu.be/UtnRQVeeLwM?list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&t=493

Keep in mind that you also need to \"scale\" the component values to real capacitor values. Not many use 1F capacitors.

Yes, he does that as the next step, for audio, 10^-7 is a good scaling
value. (100nF cap values)

                                        Thanks, Mikek

 

[Ricky]

On Monday, August 22, 2022 at 10:48:15 AM UTC-4, amdx wrote:

On 8/22/2022 9:43 AM, Ricky wrote:
On Monday, August 22, 2022 at 10:08:57 AM UTC-4, amdx wrote:
I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?
The part you misunderstood is thinking the resistor values calculated up to that point were for 500 Hz. It\'s not, it\'s for 1 rad/s which is around 0.16 Hz. The \"F scaling\" is frequency scaling to the frequency in your design rather than 1 rad/s.
OK, that is helpful.

https://youtu.be/UtnRQVeeLwM?list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&t=493

Keep in mind that you also need to \"scale\" the component values to real capacitor values. Not many use 1F capacitors.

Yes, he does that as the next step, for audio, 10^-7 is a good scaling
value. (100nF cap values)

I didn\'t pay careful attention to the whole video, but it appears he first calculated Q to allow calculating the R values. Then he scaled to the actual frequency of the circuit and the final scaling is to set the capacitors to realistic values. It is typically very easy to get whatever value of resistor you want, so they can be anything that makes the rest of the circuit work ok.

--

Rick C.

+ Get 1,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

 

[amdx]

On 8/22/2022 2:56 PM, Ricky wrote:

On Monday, August 22, 2022 at 10:48:15 AM UTC-4, amdx wrote:
On 8/22/2022 9:43 AM, Ricky wrote:
On Monday, August 22, 2022 at 10:08:57 AM UTC-4, amdx wrote:
I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?
The part you misunderstood is thinking the resistor values calculated up to that point were for 500 Hz. It\'s not, it\'s for 1 rad/s which is around 0.16 Hz. The \"F scaling\" is frequency scaling to the frequency in your design rather than 1 rad/s.
OK, that is helpful.
https://youtu.be/UtnRQVeeLwM?list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&t=493

Keep in mind that you also need to \"scale\" the component values to real capacitor values. Not many use 1F capacitors.

Yes, he does that as the next step, for audio, 10^-7 is a good scaling
value. (100nF cap values)
I didn\'t pay careful attention to the whole video, but it appears he first calculated Q to allow calculating the R values. Then he scaled to the actual frequency of the circuit and the final scaling is to set the capacitors to realistic values. It is typically very easy to get whatever value of resistor you want, so they can be anything that makes the rest of the circuit work ok.

Does a second order multifeedback bandpass have the same roll off rate
as a second order lowpass and highpass in cascade?
                                             Thanks, Mikek

 

[server]

On Mon, 22 Aug 2022 09:08:46 -0500, amdx <amdx@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek

The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

 

[amdx]

On 8/23/2022 9:24 AM, jlarkin@highlandsniptechnology.com wrote:

On Mon, 22 Aug 2022 09:08:46 -0500, amdx <amdx@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

 Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
                                Thanks, Mikek

 

[John Larkin]

On Tue, 23 Aug 2022 09:52:25 -0500, amdx <amdx@knology.net> wrote:

On 8/23/2022 9:24 AM, jlarkin@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <amdx@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

 Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
                                Thanks, Mikek

I think that wideband BPFs are better done with cascaded LPF and HPF.

 

[amdx]

On 8/23/2022 3:54 PM, John Larkin wrote:

On Tue, 23 Aug 2022 09:52:25 -0500, amdx <amdx@knology.net> wrote:

On 8/23/2022 9:24 AM, jlarkin@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <amdx@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

 Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
                                Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
                               Mikek

 

[bitrex]

On 8/23/2022 10:24 AM, jlarkin@highlandsniptechnology.com wrote:

On Mon, 22 Aug 2022 09:08:46 -0500, amdx <amdx@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
 I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

 Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

                                        Thanks, Mikek

The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

Once you do a certain type of math problem one time, if a computer can
do it the other times use a computer. It\'s what they were invented for!

Understanding how the filter equations are derived and solved is good
(and so you can sanity-check the computer\'s results) but there\'s nothing
to be learned from grunting the same type of math repeatedly. It\'s
error-prone, too.

 

[Ricky]

On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:

On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.

He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

--

Rick C.

-- Get 1,000 miles of free Supercharging
-- Tesla referral code - https://ts.la/richard11209

 

[amdx]

On 8/23/2022 6:47 PM, Ricky wrote:

On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:
On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

That wasn\'t my question. I want a 1000Hz narrow bandpass, say 900Hz to
1100Hz, Would a multiple feedback filter
be the choice or would cascaded LP and HP filters be the way to go. And
would there be a difference in roll off?
                                        Thanks, Mikek

 

[server]

On Tue, 23 Aug 2022 19:53:04 -0500, amdx <amdx@knology.net> wrote:

On 8/23/2022 6:47 PM, Ricky wrote:
On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:
On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

That wasn\'t my question. I want a 1000Hz narrow bandpass, say 900Hz to
1100Hz, Would a multiple feedback filter
be the choice or would cascaded LP and HP filters be the way to go. And
would there be a difference in roll off?
                                        Thanks, Mikek

A multiple feedback opamp circuit makes poles and zeroes. You can
build anything with them: lp, hp, bp, bandstop, allpass.

But to make a decent bp filter from 900 to 1100, for a given shape,
cascaded lp and hp filters would each have to be pretty sharp, high q,
complex.

 

[server]

On Tue, 23 Aug 2022 19:53:04 -0500, amdx <amdx@knology.net> wrote:

On 8/23/2022 6:47 PM, Ricky wrote:
On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:
On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

That wasn\'t my question. I want a 1000Hz narrow bandpass, say 900Hz to
1100Hz, Would a multiple feedback filter
be the choice or would cascaded LP and HP filters be the way to go. And
would there be a difference in roll off?
                                        Thanks, Mikek

The TI FilterPro designs your bpf as four mfb sections, basically
instantly.

 

[Ricky]

On Tuesday, August 23, 2022 at 8:53:14 PM UTC-4, amdx wrote:

On 8/23/2022 6:47 PM, Ricky wrote:
On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:
On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

That wasn\'t my question. I want a 1000Hz narrow bandpass, say 900Hz to
1100Hz, Would a multiple feedback filter
be the choice or would cascaded LP and HP filters be the way to go. And
would there be a difference in roll off?

I think the question being answered was, \"I seem to see cascading a lot, why?\"

I think you\'ve been told why a low pass and a high pass filter are cascaded.. Is that not clear? You seem to be asking others to make a decision for you. If you want me to do that, I would say use the single stage band pass filter, since that suits your goals adequately as far as I can tell. You can make the roll of of a filter anything you need. What are your requirements?

Maybe this is one of those times a simulation would offer a lot. Do you use any simulators? I know you don\'t like LTspice, but there are some very adequate online simulators, I just don\'t recall their names.

BTW, 200 Hz band width is not all that narrow at 1,000 Hz. Don\'t confuse the width of the band with the steepness of the transitions.

Just another thought. I seem to recall a twin-T filter acting (a notch filter) used in an op amp circuit to provide high gain within a band while not attenuating the frequencies outside the band. This is like a band pass filter with a fixed floor. Not all filters have an infinite attenuation in the stop band.

--

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

 

[amdx]

On 8/23/2022 10:08 PM, jlarkin@highlandsniptechnology.com wrote:

On Tue, 23 Aug 2022 19:53:04 -0500, amdx <amdx@knology.net> wrote:

On 8/23/2022 6:47 PM, Ricky wrote:
On Tuesday, August 23, 2022 at 5:10:31 PM UTC-4, amdx wrote:
On 8/23/2022 3:54 PM, John Larkin wrote:
On Tue, 23 Aug 2022 09:52:25 -0500, amdx <am...@knology.net> wrote:

On 8/23/2022 9:24 AM, jla...@highlandsniptechnology.com wrote:
On Mon, 22 Aug 2022 09:08:46 -0500, amdx <am...@knology.net> wrote:

I found this a very helpful video for picking values for a Bandpass filter.
But I don\'t understand one step, (it still works, I just don\'t understand).
Here is the video,
https://www.youtube.com/watch?v=UtnRQVeeLwM&list=PLxuejeK2BP_dMJcEapPFavqaItrewjjT1&index=37&t=620s
At 9:43 He does what he calls an F Scale, he describes that the first
template is Omega(o)= 1 rad/per sec.
He \"F Scales\" it to 2pi times frequency of the bandpass.
I don\'t get it because he has a filter that band passes at his
frequency, then reduces the resistor values, so it seems
the bandpass frequency would be different. Then scales both L and C to
normalize values.

Is there a more discriptive term for F Scale? Or, if I understand it,
that is a good term?

Thanks, Mikek
The classic Williams book does something similar to map an LC lowpass
filter to a bandpass. He designs a prototype lowpass filter, cutoff 1
Hz, and then series resonates every L and parallel resonates every C,
both up to the desired bandpass center frequency. That (somehow,
magically) shifts (NOT scales) the lowpass up to becoming a bandpass
of, I think, 2 Hz width about the new center.

After that transformation, one can do the usual de-normalizations of
frequency and impedance.

We don\'t do the heavy math to design filters; we use somebody\'s
software. Or sometimes just fiddle.

For a guy with little math back ground, this was very easy for me to do.
I even ordered his book with the templates for various type of filters,
I\'m most interested in an active chebychev, as I want a strong filter.
I want to isolate a 1000Hz signal in the noisy world to see changes
in the level of the 1000Hz signal. Noisy world, air conditioner,
oscilloscope fan, refrigerator,
son on keyboard, computer fans, vehicles driving by. Thank goodness LEDs
don\'t make audible noise!

Regarding roll off, is there any difference between a 2nd order
multiple feedback bandpass and
cascading a 2nd order low pass and a 2nd order high pass? I seem to see
cascading a lot, why?
Thanks, Mikek
I think that wideband BPFs are better done with cascaded LPF and HPF.

I\'m sorry, I wasn\'t clear, I want a 1000Hz narrow band pass filter.
He\'s explaining why others would use the combination of a LP and a HP filter. Band pass is not always a single frequency. An audio amp or an ADC will typically use a low pass filter and a high pass filter to set the two corner frequencies of the band of interest. Technically, that\'s a band pass. A single band pass filter with the same band width would have a very low Q and so very shallow roll off in the transition band, being nearly worthless.

That wasn\'t my question. I want a 1000Hz narrow bandpass, say 900Hz to
1100Hz, Would a multiple feedback filter
be the choice or would cascaded LP and HP filters be the way to go. And
would there be a difference in roll off?
                                        Thanks, Mikek
The TI FilterPro designs your bpf as four mfb sections, basically
instantly.

Wow!
It took me longer to get registered than it took to design a 6th order
three section multifeedback chebychev filter.
Of course the selected IC in a form I can work with is \'on order\' I\'ll
look at the alternatives.
                                    Thanks, Mikek