Hi,

I'm trying to design a passive IF bandpass filter with the following specifications:
Center frequency: 61.38 MHz
3dB band width: 2.4 MHz

My first choice was a 3rd order butterworth/chebyshev filter but the attenuation isn't good enough, so I increased the order.
However, with the component's Q-factor that I have available, the insertion loss is just too big (10-15 dB).

Q inductor ~ 50
Q capacitor ~ 1000

Any way to design this filter with the components in hand to achieve a better frequency response?

Thanks for the help,
Alex

 

Hi,

I'm trying to design a passive IF bandpass filter with the following specifications:
Center frequency: 61.38 MHz
3dB band width: 2.4 MHz

My first choice was a 3rd order butterworth/chebyshev filter but the attenuation isn't good enough, so I increased the order.
However, with the component's Q-factor that I have available, the insertion loss is just too big (10-15 dB).

Q inductor ~ 50
Q capacitor ~ 1000

Any way to design this filter with the components in hand to achieve a better frequency response?

Thanks for the help,
Alex

You should be able to do this with a simple overcoupled tuned transformer, if passband ripple isn't too critical.

Let me do some modeling and get back to you

 

Hi,

I'm trying to design a passive IF bandpass filter with the following specifications:
Center frequency: 61.38 MHz
3dB band width: 2.4 MHz

My first choice was a 3rd order butterworth/chebyshev filter but the attenuation isn't good enough, so I increased the order.
However, with the component's Q-factor that I have available, the insertion loss is just too big (10-15 dB).

Q inductor ~ 50
Q capacitor ~ 1000

Any way to design this filter with the components in hand to achieve a better frequency response?

Thanks for the help,
Alex

Okay, this could do the trick for you. Here's the netlist:

L1 2 0 7E-9
L2 3 0 7E-9
K1 L1 L2 .04
V1 1 0 AC 10
R1 1 2 500
C1 2 0 1E-9
C2 3 0 1E-9
R2 3 0 50

Keep in mind, this circuit is VERY termination sensitive, both for input and output impedance. If you change those, you'll have to scale it accordingly.

Notice the VERY low coupling coefficient of K1. In the "flesh" this may be accomplished by having the coils nearly in the vicinity of each other! (If you have a sweep generator/tracker, it really helps to align this thing).

Hope this helps. I've been building I.F. strips like this for decades.

eric

 

Hi,

Thanks for the help.
The simulation indeed shows a real improvement in comparison to the previous filters.

Just a few more questions,

First, about the input and output impedance. They are very large at the resonant frequency and loading them with 50 Ohms will "kill the Q of the filter" (as my instructor said). One proposal was not to connect the signal directly but through a transformer primary winding. What do you think about that?

Second, what about capacitive coupling instead of inductive coupling? Is it easier to implement?

And I'm not sure about the "aligning" process. How can I "play" with the distance between the two coils? (Suppose I have a network analyzer)

Thanks again for the help, you really opened a whole new subject to me.

Alex

 

Hi,

Thanks for the help.
The simulation indeed shows a real improvement in comparison to the previous filters.

Just a few more questions,

First, about the input and output impedance. They are very large at the resonant frequency and loading them with 50 Ohms will "kill the Q of the filter" (as my instructor said). One proposal was not to connect the signal directly but through a transformer primary winding. What do you think about that?

Second, what about capacitive coupling instead of inductive coupling? Is it easier to implement?

And I'm not sure about the "aligning" process. How can I "play" with the distance between the two coils? (Suppose I have a network analyzer)

Thanks again for the help, you really opened a whole new subject to me.

Alex

Hi Alex:

Glad I could help...and indeed this is a very deep and fascinating topic.

One could use capacitive coupling, but historically, the coupled transmformer was always easier to implement...at least at lower frequencies. The traditional "I.F. can" consisted of a double tuned transformer with moveable powdered iron slugs. You could twiddle the mutual inductance pretty much at will this way.

Input and output transformers to give you the desired impedance is certainly valid in your case...with a bit more circuit complexity. OR, you could scale the I F filter for different impedances by changing the L/C ratios on both ends. To get the feel if this, change the L's up by a factor of 10 and the C's down by a factor of 10. See what happens! (I'll leave this exercise for you).

eric