Calculating values for input coupling capacitor

[officeboy]

I haven’t found a discussion on this really. The few things I found while searching was that really anything will work, as long as it's not to small, and too large will take longer to become stable. Not really any math or reasoning behind that.
To my understanding an input coupling capacitor seems to look a bit like a first level crossover. But that can’t be right since the resistance is also in the signal path, and not to ground. So could this be like a ½ level crossover? How do you calculate the minimum value of the capacitor, or calculate what the crossover point will be for a given capacitor?

I was just going to use the standard 4.7uf, but that bothered me. I need to know why that value is recommended or used.

[Nisbeth]

An input coupling capacitor acts as a high pass filter together with the input impedance.
The formula for the calculation is C = 1/(2pi*f*Zin) where C is the capacitance, f is the corner frequency and Zin is the input impedance. You should aim for a corner frequency of 5 Hz or below to avoid any phase shift in the lowest bass region.


/U.

[officeboy]

So in the case of an lm3875, or 3886, is Zin just the value of the Rb resistor? Or is there resistance in the chip and it's load calculated into that figure? And is the C value in uF?
And I come up with values then that are in the .00X range for capacitors. Why are values as high as 4.7uF recommended?

[officeboy]

So after a few hours of playing with LTspice and excel. I've found that the calculation "C = 1/(2pi*f*Zin)" gives C in F, so multiply it by 10,000 to get uF.
Using for example Peter Daniels simplified gainclone (22K feedback, 220 input, 680 feedback to ground and 22K ground ref) an input cap of .0001 (1uf) gives a -3dB of 7.23hz. The "magic" 4.7uf gives 1.53hz.

I'm not sure how to calculate if there is no resistance between the capacitor and the op-amp (like shown on page 13 of AN-1192.pdf) Would 7.12hz be right for that, or is it 0.07hz?

[Nisbeth]

In that figure, Zin is = Rin = 47.5 k and so f = ~7,1 Hz (meaning that using a 1uF cap here wouldn't hurt )



/U.

[sreten]

Farads x 1,000,000 (10^6) gives you uF's.

0.0001 = 100uF.

sreten.

[Nisbeth]

Sorry I forgot to mention that in my original post


/U.

[officeboy]

Ok, I think it got it.

Thanks guys..

Now to fix the amp I just put together.
My subamp seems to have a 71hz highpass filter.. and I don't think that's gonna work.

[Nisbeth]

Naahh, that's probably a little on the high side


/U.

[NaftaliG]

Farads x 1,000,000 (10^6) gives you uF's.

I believe that microfarads is 10^-6 farads, one millionth of a farad, not a million farads.