# Multiple Feedback Bandpass Filter: Simulation question.

Discussion in 'General Electronics Chat' started by protobit, Dec 26, 2012.

1. ### protobit Thread Starter New Member

Dec 18, 2012
2
0
I'm trying to design a RF receiver to pick up AM 60khz and what I'm thinking is to make a two-stage BJT amplifiers and an active opamp filter.

For the active op-amp filter I'm trying to use a multiple feedback bandpass filter with middle frequency of 60khz, middle band gain of 1 and bandwidth of 5khz.

To solve for the component values I'm using matlab:
Code ( (Unknown Language)):
1. %Multi-Feedback BandPass OpAmp Filter
2. syms R1 R2 R3
3. C1 = 1e-9;
4. w_m = (1/(C1^2*R3) * (1/R1+1/R2))^0.5;
5. A_m = R3/(2*R1);
6. B = 1/(pi*C1*R3);
7. % Q = 0.5*(1+R1/R2)^0.5*(R3/R1)^0.5;
8.
9. solu = solve(w_m == 2*pi*60e3, A_m == 1, B == 5e3);
10.
11. sol_R1 = double(solu.R1);
12. sol_R2 = double(solu.R2);
13. sol_R3 = double(solu.R3);
14.
15. fprintf('\nC1 = %.3e\nR1 = %.3e\nR2 = %.3e\nR3 = %.3e\n\n', C1, sol_R1, sol_R2, sol_R3)
Results:
Code ( (Unknown Language)):
1. C1 = 1.000e-09
2. R1 = 3.183e+04
3. R2 = 1.109e+02
4. R3 = 6.366e+04
Which produces the expected Bode plot with the following matlab code:
Code ( (Unknown Language)):
1. R1 = sol_R1;
2. R2 = sol_R2;
3. R3 = sol_R3;
4. C1 = 1e-9;
5.
6. num = [1/(R1*C1), 0];
7. den = [1,...
8.         2/(R3*C1),...
9.         1/(C1^2*R3)*(1/R1+1/R2)];
10.
11. h = tf(num, den);
12. bode(h);
Using these results I am simulating the circuit using LTSpice:

Netlist:
Code ( (Unknown Language)):
1. XU1 0 N003 pos neg V_out LM741
2. V1 pos 0 9
3. V2 0 neg 9
4. V3 N001 0 AC 1
5. R1 N002 N001 31.8k
6. C2 N003 N002 1n
7. C1 N002 V_out 1n
8. R3 N003 V_out 63.66k
9. R2 N002 0 110
10. .lib lm741.sub
11. ;tran 0 10m 0 .01
12. .ac oct 1000 1k 1MEG
13. .backanno
14. .end
Which produces a frequency response centered at 40kHz (not 60kHz):

I understand that the matlab and SPICE wont match usually because one is ideal while the other is not. My question is why is their such a discrepitancy between the designed middle frequency of 60khz and the simulated middle frequency of 40khz. How can I modify and/or model my circuit to work for 60khz?

2. ### Audioguru Expert

Dec 20, 2007
10,729
1,201
A lousy old 741 opamp has trouble above only 9kHz. It cannot go as high as 60kHz and I am surprised to see 40kHz.
Most opamps can go up to 100kHz. Try one.

3. ### protobit Thread Starter New Member

Dec 18, 2012
2
0
So I'm somewhat new to electronics. I just simualted a RC4558 and the center frequency is now 50kHz. What would be a good suggestion for a better opamp? Also what specs on a datasheet would be telling of this?

Edit: This would be slew rate correct?

Last edited: Dec 26, 2012
4. ### Audioguru Expert

Dec 20, 2007
10,729
1,201
A 741 is about 45 years old. A 4558 is about 42 years old. Use a much better newer opamp.

The Gain Bandwidth Product and the slew rate should be very high for a peak at 60kHz.

protobit likes this.
5. ### TheGuy New Member

Dec 27, 2012
2
0
Just curious, what kind of op amp would you recommend for something like this?

6. ### Audioguru Expert

Dec 20, 2007
10,729
1,201
I am an audio guy. I use low noise 100kHz audio opamps. You need something with a very wide bandwidth.

7. ### TheGuy New Member

Dec 27, 2012
2
0
Right, so what kind of models specifically? Is there a particular line that you prefer over others? What kind of values would be preferable for something like this?

8. ### Audioguru Expert

Dec 20, 2007
10,729
1,201
Is the bandpass what you want? Its peak is extremely narrow but its skirts are very gradual and wide. The opposite of a "normal" radio.

9. ### Veracohr Distinguished Member

Jan 3, 2011
665
101
Gain Bandwidth Product is a good one to look for. Go for higher. Also slew rate (depending on desired output level).