Capacitor tolerance importance for high end crossover filters?
 

I am struggling to find any sort of information at all as to how capacitor tolerance applies to accuracy of crossover filters and summed frequency response?

I am building some high end scan revelator/scan ring radiator air circ 2 way towers and am comparing between 5% tolerance vs 3% tolerance caps for this design.

The 3% ones (I deal with one supplier only and am happy with them wont buy elsewhere) are hellishly expensive compared to the 5% ones however if there may be a noticable difference in crossover point shift (example say 1-2db) of summed frequency response output, for this type of high end design, I will just get the 3% ones however I have no experience with summed response measurement output and capacitor tolerance accuracy and if its less than 1db, its just not worth more than double the price.

Cheers in advance

There are standard formulae for calculating the value cap you need - as well as the other components - for a crossover. A change in capacitance will result in a change of frequency. The tighter the tolerance, the less variance from that design frequency.

For example, if you calculate a crossover needs a 10uf cap for some target frequency. Redo the calculation for 10.5uf and 9.5uf. That will tell you in your case how far the crossover frequency will range within a 5% variance of that cap. Now try 10.3uf and 9.7uf. That is the spread for 3%. Then you can see the diference and decide if it matters.

I have tried designing crossovers from mathematical calculations which takes well...lets just say a VERY long time and sound...terrible so now I get someone to design them for me and they sound great!

That is a very good point though once I get the filter/driver responses I could factor in a percentage of tolerance and rework the graphs summed response to +/- reasonable accuracy.

question is how does the capacitors tolerance effect the low pass vs high pass (paralell/series) how much weighting on frequency rolloff does the coil vs cap have on the filters response? eg. if the lowpass series coil has more of a bearing on lowpass the paralell caps tolerance could be negligable...And then what about 18db low pass with 2 coils...this is where it gets really difficult for me to accurately estimate