butterworth and nonstandard values

With both my high an low drivers being 8 ohm and the crossover point at 3500Hz, the calculators for 2nd order butterworth tell me:
L1 = L2 = 0.51mH
C1 = C2 = 4.02uF

I can wind a coil no problem, but that cap. :sad: I could use a 3.3 and a 680nf to get close but is that close enough? That requires fitting 2 caps on the board and metalised PP is expencive.
I could get the correct value http://uk.farnell.com/jsp/endecaSearch/partDetail.jsp?SKU=1190553 Which is metalised PP but meant for motors! Does it matter?

Should I change the x/o frequency to 3000hz to give me an easier value?
This must be a comon problem but I can't find the answer. What is the workaround?

..try re-calculating with closest obtainable parts - might be close to what you want anyway..

Jakob E.

Even with a 3.9uF cap you're still only off by about 3%, making your crossover point 3600Hz instead of 3500. That's probably not going to be audible, but if you're worried, add 3.9uF and 120nF.

Of course, the 3.9uF probably has a 5% tolerance anyway...meaning that if you want to be really accurate you need to buy a batch of them and select.

Peace,
Paul

[quote author="pstamler"]...
Of course, the 3.9uF probably has a 5% tolerance anyway...meaning that if you want to be really accurate you need to buy a batch of them and select.

Peace,
Paul[/quote]

IIRC there's an example of the folly in paralleling caps for a precise value while neglecting the likely tolerance effects, in the datasheet app notes for the ballyhooed new National Semi opamp. You are asked to use 5% caps and place 22nF in parallel with 4.7nF and 500pF (the latter a nonstandard value at that).

They also do something similar with the requested resistor values, suggesting that 100 ohms be placed in series with 3.83k and 909 ohms with 26.1k. But at least with resistors the decent ones are often much tighter than 1% to begin with. One wonders if the writer knew that 3.92k is a standard value though, or that it's pretty easy to get 0.1% tolerance parts.

I need to go away and learn more theory instead of just understanding the principals and using the online calculators!
Meanwhile, I've decided to cross at 3KHz which is still well within the perameters of the drivers.

[quote author="Pbassred"]I need to go away and learn more theory instead of just understanding the principals and using the online calculators!
Meanwhile, I've decided to cross at 3KHz which is still well within the perameters of the drivers.[/quote]

3kHz should be fine. The difference between that and 3.5kHz is only about 2 1/2 semitones. I can't imagine it being so sensitive that you'll notice.