Low frequency preamplifier

[ritchie888]

I'm trying to design a pre-amplifier for low frequencies. I'm specifically interested in 2-200Hz. I'm fairly sure I have the right design, just the values of the components must be wrong.

Having done some math, I'm under the impression that my circuit should work for my frequencies, but plotting these results using Multisim is giving some strange results.

So, if I want my cutoff to be below 2Hz, I've selected the values of 220uF and 150k Ohm biasing resistors (with a 100x gain). To my understanding this gives an input impedance of 75k Ohm. Punching this into the frequency cut off equation: F = 1 / (2 * pi * R * C) = 1 / (2 * pi * 75000 * (220 * (10^-6))) = 0.0096Hz, far below what I actually need it to be. Anyway, even with this appearing to be correct, I did plot the frequency response (1Hz to 100Hz) using the following equation, and it would appear my values are fine:

Gr = 20 * log10((w * R * C) / (sqrt(1 + (w * R * C)^2)))

where w is the frequency times 2*pi.

However, as you'll see in the attachment, the plot starts off with a lot of jumping and cutting off, but eventually sorts itself out and corrects itself. I need the circuit to respond correctly from the start.

Can someone suggest what I should actually have as values, if these are incorrect. I can plot the frequency response if that will help.

 

[VEBill]

Maybe C3 charging up to 1.5 volts? Perhaps you should employ a bipolar (±V) power supply and keep the signal centered about 0v.

Stand by for others' advice.

 

[LiteYear]

Can't see anything wrong with your values. In fact, when I duplicate the circuit in LTSpice it works as expected with no jumping or cutting off. Looks like a simulation error. Would connecting the negative terminals of the probe inputs to ground help?

There is a transient effect that decays after about a second, but all it does is attenuate the output slightly. As VE1BLL suggests, this is because C3 takes that long to get in phase with the output. Could you scrap C3 and tie the base of R3 between a 50:50 resistive divider across the 3V source instead? That would fix the output relative to the input DC bias right from the start. Can imagine this might be unstable with such a large gain though.

 

[signal2noise]

Try forcing the step size to a small value in the simulation setup. Sometimes the algorithms that try to speed up the simulation get themselves into trouble.
I have found that by setting a fixed small step size and just waiting a little bit fixes a lot of things.

The circuit should work as drawn.

 

[ritchie888]

Thanks, guys.

I was under the impression the values were correct, just a bit of a delay for the capacitor to charge. The simulation in Multisim isn't fantastic, but it'll do (certainly doesn't run in real-time, that's for sure...).