parallel capacitor values

Hi all,

Is it good practise to parallel up capapacitors if the values i require are unavailable???? The application is for active filters such as butterworth etc.

some of the capacitor values i require are not common values and it becomes even harder when i have to limit myself to certain types of capacitor to give good results for filter circuits.

andy257 said:

Hi all,

Is it good practise to parallel up capapacitors if the values i require are unavailable???? The application is for active filters such as butterworth etc.

some of the capacitor values i require are not common values and it becomes even harder when i have to limit myself to certain types of capacitor to give good results for filter circuits.

Click to expand...

Not sure about filters. BUt at least for decoupling, very stringent applications (like decoupling microprocessors) dictates you should use parallel the SAME capacitor rather than mix different values. THe reasoning is that capacitors of different values have different parasitic inductances and resistances that work together to form RLC tank circuits and impedance spikes within the frequency range that is overlapped by the various individual capacitors (ie. probably the region you are trying to bypass when paralleling different capacitor sizes in hopes their frequency responses will overlap).

So in short...I don't see a problem as long as the capacitors are all the same size.

But practically now...don't you pick the capacitor sizes for a filter before you pick the resistor values? THe reason is that there are far more acurrate resistors than there are accurate capacitors. And also, how strange are these capacitances? Because if these are just op-amp regular filters, etc. the frequency response is pretty crummy (try simulating it and see how bad it is). The gain slope drops so gradually that there's not much difference between different cut-off frequencies a lot of the time. I don't see why it would be a terrible problem unless you had some very very stringent application, which in that case you would also probably be using a much more complicated filter like a Bessel, Elliptical (my two favorites ), or butterworth filter, where it would matter.

andy257 said:

Hi all,

Is it good practise to parallel up capapacitors if the values i require are unavailable???? The application is for active filters such as butterworth etc.

some of the capacitor values i require are not common values and it becomes even harder when i have to limit myself to certain types of capacitor to give good results for filter circuits.

Click to expand...

It's perfectly fine, and also normal practice!.

A Sallen and Key Butterworth 2nd order filter can use exactly the same capacitor and resistor values if its gain is about 1.6.
If its gain is 1 then if the resistors have the same values, the capacitors must be half or double. Then just use two of the same capacitor value in parallel.

Be careful with the tolerance - there's no point in calculating and parallelling numerous capacitors with a tolerance of 20% to get say 110nF as it could range between 188nF and 132nF when you might as well use 100nF.

Choose capacitors with a close enough tolerance for your application and select the values within the tolerance range.