I am designing a very simple first-order RC band pass filter with cutoff frequencies \$f_l = 0.5 MHz\$ and \$f_h = 4.5MHz\$. I have it pictured below (I'm using 5Spice):
This filter acts the way I want it to. The half power points are around the appropriate cutoff frequencies. For the HPF side of the filter, \$f_h = 1/(2*\pi*100nF*3\Omega)= 0.53MHz\$ and on the LPF side, \$f_l = 1/(2*\pi*1nF*35\Omega)= 4.55MHz \$. Below is the frequency domain plot:
Now, when I change the capacitor and resistance values to something that should yield the same cutoff frequencies, the frequency domain plot looks very different. Here's the modified circuit:
And the cutoff frequencies should be \$f_h = 1/(2*\pi*31.8pF*10K\Omega)= 0.50MHz\$ and on the LPF side, \$f_l = 1/(2*\pi*3.54pF*10K\Omega)= 4.50MHz \$. However, not only does the peak of the signal exist at less than -20 dB:
-3dB from that peak is not at the correct cutoff frequency:
How can these two configurations yield such different results? They both calculate the correct cutoff values, so why are they different in simulation?
Also, in researching this subject I realize that this is a very poor filter and that I should be using a passive RLC band-pass filter or an active one. I'm new to 5Spice, so please suggest any edits to these pictures and I will make them to make this question more readable.