low pass filter circuit and pspice simulation

Thread Starter

McMurry

Joined Jul 5, 2008
9
I have calculated the value of the resistors and capacitor of the Sallen-key low pass filter in order to meet some specifications i.e.

low freq gain = 22db (OK!)
cutoff freq = 11khz (OK!)
less than -70dB at freq higher than 110khz (NOT OK!)

i am currently using sallen-key (K= 3 - 1/Q) configuration.

i cant simply change the value of capacitance or resistance. i need to support it with calculation.

is there any "trick" to meet all the specification above?
 

Attachments

Audioguru

Joined Dec 20, 2007
11,248
Your problem is the lousy old LM324 opamp. It barely makes it up to 6kHz and you need it to perform above 500kHz.

The TL071, TL072 and TL074 opamps have good response up to 100kHz but might be poor above 600kHz.

I don't know any opamps that work well up to very high radio frequencies.
 

Thread Starter

McMurry

Joined Jul 5, 2008
9
unfortunately i can only use either LM324, or UA741 for this assignment.

is there any component value (resistance/capacitance) that i can change to meet the third specification in pspice (not actual hardware)
i.e. gain less than -70dB at freq higher than 110khz
without affecting the previous two specification?

i've tried for days, and still couldn't figure it out.
 

Ron H

Joined Apr 14, 2005
7,063
If you really want greater than -90dB rolloff (+22dB to -70dB) at one decade from the corner, you won't get it with a simple 4-pole Butterworth. Each pole will give you -20dB/decade, ideally. You will need a 5th order filter if you want to stick to Butterworth.
You can get some relief from that nasty bump near 1MHz by using larger resistor and smaller cap values. This will minimize the effect of the high open-loop output resistance of the op amps, but it won't completely solve your problem.
This article might be of interest.
 

Audioguru

Joined Dec 20, 2007
11,248
The opamps do not have any gain at high frequencies so their output impedance is high.
Then the signal from the feedback capacitor is coupled directly through to the output.

So reduce the value of the capacitors (and reduce the value of the resistors to keep the same cutoff frequency) then less signal will feed through to the output.
 

Ron H

Joined Apr 14, 2005
7,063
The opamps do not have any gain at high frequencies so their output impedance is high.
Then the signal from the feedback capacitor is coupled directly through to the output.

So reduce the value of the capacitors (and reduce the value of the resistors to keep the same cutoff frequency) then less signal will feed through to the output.
When the caps get smaller, the resistors have to get bigger.
 

The Electrician

Joined Oct 9, 2007
2,971
I have calculated the value of the resistors and capacitor of the Sallen-key low pass filter in order to meet some specifications i.e.

low freq gain = 22db (OK!)
cutoff freq = 11khz (OK!)
less than -70dB at freq higher than 110khz (NOT OK!)

i am currently using sallen-key (K= 3 - 1/Q) configuration.

i cant simply change the value of capacitance or resistance. i need to support it with calculation.

is there any "trick" to meet all the specification above?
There are a number of things you haven't said.

Must you use the topology shown in your first image? In other words, must you use 3 opamps, with the first being only a buffer (no reactive components), and the 2nd and 3rd being Sallen-Key filters with only 2 capacitors each?

Can the opamps be ideal, or must you simulate with real opamp models?

Must the passband be maximally flat, or can you have some ripple there?

If you can add a capacitor to the first opamp circuit, you could increase the order of the filter to 5th, and then meet the attenuation spec (ignoring the feed-forward leakage problem Sallen-Key filters have with some opamps) with plenty of margin. If you can't do this, and if you can't add components to the existing topologies of opamps 2 and 3, then you will probably have to allow some ripple in the passband. I checked, and it appears that a type 1 Chebychev filter with 1.1 dB ripple in the passband will just do it.
 
Top